As someone who hit 10 mmr in dota (the lowest rating possible, i was learning midlane and was TERRIBLE at it) it was honestly a great experience seeing myself improve and rank up again,, last week i hit 3k mmr, still playing midlane and having fun! So dont let the numbers discourage you, 💪
@@Stanky_Footthe difference between 1 mmr and up to 3500 mmr is the amount of pleasure you'll get during the game. The closer you get to average mmr (the one every goddamn player gets at the start), the worse the games will be. 1 and 5500 are equally close to that rating, meaning the quality of the games goes up. And I say it as a player who's been at every one of those ratings at different points in time
@@Stanky_Foot The crazy part is if you've seen the dialogue that high ranked players provide to you when you play with them is like a incoherent priest speaking in tongues.
I think the points by Activision about the lowest-MMR players eventually quitting are good but perhaps not emphasized enough. As a game gets older this happens more and more so that the average skill of the entire playerbase goes up. This causes the initial "skill wall" that new players need to overcome to get higher and higher, causing new players to "bounce off" and quit immediately unless they have an above-average level of talent and determination to learn despite all the losses. Without an influx of new players, those older players at the bottom of the rankings will have a harder and harder time finding games against equally skilled opponents, and they too are at risk of giving up and quitting. As the lowest skilled players quit, the "losing portion" at the bottom of the ladder will continue to move up, claiming more and more players over time. In the end, the playerbase can shrink to a tiny shadow of its former glory and be composed of only the oldest and most skilled players
This is a legitimate "problem". I find starting AoM multiplayer hard, because I'm going into a 20 year old game and anyone playing already knows all the basics. So I have to lose at minimum 5-10x in a well-working system to reach fellow newbies. Not a welcoming experience (Actually reminds of my first basketball practice sesh, I ragequit after 5 minutes)
i've read that this is why Sakurai has effectively disowned the competitive Smash Bros community. He wants to keep the game friendly to casual players.
@@thefance4708 It makes sense. Competitive scenes are great for elite competitive players and their fans on twitch/youtube but I don't think they do much for sales
The last point you made is a very important one! Most MMR-Systems are still heavily based on the Elo-system, wich was never developed with team based games in mind. There can also never be a universal system for team-rating, as the impact of one very good or very bad player can have varies vastly from game to game. So it's inevitable that good matchmaking gets harder and harder the bigger the team sizes get.
Its also extremely difficult to quantify what someone contributes to a team instead of just awarding points based on win/loss or score at end time. For example in counter strike, in matchmaking you gain or lose elo based on round win difference in the game and the avg elo difference between the 2 teams. So if you beat enemies around your own skill level 13 to 0 you get more points than a close win. The issue with this is, what if you play a game where you massively carry your team, get 30+ kills, (which is more than 1 per round, the average is like 0.67 kills per round) but the game goes long and you win 13-11, you barely gain any elo because your team was not very good. You will only gain a few points for massively overpoerforming relative to what you should be doing at your skill level, your teammates who played badly and died more than they got kills also get points, while they should be losing elo. It also doesnt take into account how many assists you get, how many times you save teammates from dying, how many enemies you flash, etc. Your kill-death ratio doesnt even factor in. Some online services like ESEA tried to improve this by coming up with a system that measures player contributions to a round for ELO gain. The result of this is that everyone wants to play the best spots every round, never pushes first on the attacking side, baits their teammates to get easy kills, and throws useless flashes at enemies without pushing just to make a number go up. Its literally impossible to devise a system that isnt unfair or open to abuse in a 5v5 game.
@@TheSuperappelflap Very true and it is something I have mention before on this discussion. Its actually impossible to quantify player skill, just too many variables. 1vs1 You can just look at who wins, but in a team game, if you were to determine each players contribution to the team it would be impossible, especially in a MOBA for example where you have different roles like support or carry.
@@TheSuperappelflap the issue here is seeing MMR as a reward. You won a hard match and feel like you need to be "rewarded" with more points than an easier stomp. MMR's goal is simply to get you to 50% winrate so the matches are balanced. More balance = more fun for everyone. While stomping players feels fun, it's not fun for the players getting stomped. We can't have a vampiric system where the only fun you can get from a game comes at the expense of others.
What most people are missing is that in team games raw mechnical skills are only part of the story. Social skills can matter even more. I started climbing like crazy in Dota just by coordinating basic plays, redirecting salty players attention onto enemies and I would join with him to throw smack texts on then and do some pep talk when someone wanted to surrender a game. Mechanically I was in same position as before. I climbed 3 full ranks in Dota 2 with just that. Overall it's crazy how huge difference makes when one player just rallies others and as I often played as initiator saying in advance "stunning X" lead to massively different outcomes in team fights.
@andrzejnadgirl2029 Social skills can even matter in FFA based games like risk too. It's not enough to be able to recognize when kills are profitable or how to attack fast amd beat the timer. Negotiating with other players, not making enemies with everyone because having a good reputation at the table with randoms, being able to recognize and take advantage of lower skilled opponents by intentionally not taking territory so they attack someone else etc
Hearthstone and other random based players would disagree with you on that. Some games might manipulate your chances to make "perfect" play so your win rate stays about 50%.
@@andrzejnadgirl2029 true, and its extremely hard to keep attention on your and other people's game at the same time, if you can do that, you're naturally gonna get higher ranks
The biggest complaint about CODs SBMM is that theres no visible rank or MMR. So your reward for playing better is just harder matches and no way to discern if you're actually getting better.
Oh, so the game would suddenly be fun if you could see an arbitrary emblem next to your name, totally makes sense. Yeah man, good games are only good and fun if you can see a meaningless medal
@@tehDmez your progression comes from you. Your own actual progression. If you value some psychological trick, superficial progression system over that, then you are a sheep human
I think the argument for dedicated servers vs. matchmaking is always something to consider. Consider how many old FPS games had individual servers that had their own communities which meant you had fun servers, competitive, and casual ones always go in hand in hand, meaning you had to look for them, but you got the satisfaction of playing the game how you wanted. This can be considered in WC3 by how custom games were their own thing, then there was competitive, but there were also noobs or people that played the base game for fun in the custom maps section to play their own way (such as making self proclaimed rules of "no rush"). Many newer games instead opt for just a matchmaking ranking system for online, meaning that all these people are mashed together and are more likely to lash out because they are trading the convenience of having to search for a game with having that freedom be taken away and feeling like a robot is "forcing them to lose." I also think for any game such as DotA 2, you always feel so much worse for losing for team based affairs, maybe someone running into the enemy team too much, etc. while a fighting game or any 1v1 has less weight to losses because it's based entirely around how you play and who you play as. This isn't even considering sometimes lower MMR is filled with Smurf accounts or people wanting to use more cheese specific builds in an attempt to climb ladder like how early SC2 had many claims Bronze League players would stick to Zergling rushes.
From my experience, the more your team is winning/stomping the other team, the less communication and talking happens. The worse your team is losing/getting stomped by the other team, the more venting and pointing fingers takes place.
3 місяці тому+5
I found if there is more communication before then you tend to win more... be relaxed and destroy others. It's an unfocused flow.
The golden combination for rage venting is (losing + lack of information). This causes people to vehemently attack others specifically so they can assert that they themselves are not the problem. When it's already obvious who the problem is, and that some players on the losing team are still playing well, then there's less of a need to rage. This is why Overwatch was such a salty and ridiculously designed pit of nonsense. It deliberately hid match statistics in an attempt to coddle poor performers and instead created even more rage as people defended themselves in a game that wouldn't tell their own teammates they were currently playing decently.
I think there's a "communication curve". Bad players "communicate" a lot via flaming. Middle mmr players basically dont communicate. And the top players are ultra communicative.
@@ProzacStylings Problem with Overwatch was not hiding match statistics, because match statistics don't really show how well someone is performing. Damage/Kills/Assists/Healing is not directly proportional to how well you are doing. Damage is important, and some classes end up doing more than others. Someone who is standing ready to take out a specific target and succeeding every time, with little damage, high kill rate might be doing just as well as the tank that is in the right position at all times, with zero kills and nearly no damage. Lack of information has in general shown to be better for less toxicity, because it is harder for someone to just "blame a guy". Look at League of Legends, people blame the support that is 0/5/18, while... that can be pretty good stats for a support, especially if the deaths lead to winning fights. Same for the engager that might have almost no kills, tons of deaths, but have ensured most fights are a win. It's harder to just "blame someone" if there aren't any statistics to use. The counter-argument is that someone that might be toxic can see their own terrible performance, but often, you can have a toxic person blaming others for their terrible performance, because it's not their fault regardless. Statistics does not protect from self-delusion. If you take a notice of the games that tends to be the most toxic, it is the games that are team games, where you are highly reliable on your teammates, as well as games with a high population of kids, as well as games with very popular streamers that are toxic. Having or not having statistics tend to be a relatively minor variable in this multi-variable question.
@@SioxerNikita Damage/kills/healing absolutely does tell how good someone is doing. Your attempted counterpoint of a Tank not being represented is solved by adding a Damage Blocked stat. The entire rest of your argument is nonsense based on that incorrect assertion. People who cannot prove they are good at the game by pointing to the stats screen will prove they are good at the game by explaining how their teammates suck. Period. That is why OW is a flame fest.
People hating forced 50% MM when the real problem is anonymous team games. Imagine ANY real life sports that randomly matched you with teammates to compete at a high level. It would be a nightmare just like it is in online games.
i dont think so , cause if you poke a basketball in the ghetto (because you couldn't hit the basket), you'll probably never be able to play any sports again, and you'll only eat from a straw, I would love to play Dota if they gave me the addresses of my teammates after the games.
Except it wasn’t, that’s literally how it used to work and it was great, at first you get destroyed every game then after a month you win 50% of your games then after like 3 months you have the skill and map knowledge to win 90% without even really trying and you feel like a God who over came a real obstacle. Then the cycle repeats when a new game comes out.
@@TannerLindberg Surely with a good enough skill based match making algorithm given enough time your win rate would even out to 50% if that was the developer intention. Why would it be impossible to force 50% in fighting games? Not saying that the match making system in those games is forced 50 just interested in why you think it’s impossible.
His off-meta commentary didnt totally land for me. I played DotA 2 for a long time and got to a 'good' mmr (5k), thats the 'very high' bracket which is top 10 or 15% I cant remember. I started to go off meta because its more fun and my winrate did drop, I went down to 4k. Here is where I dont agree with Grubby, it made the game worse. Even while being 1k mmr higher than my team they would flame non-stop or even throw games out of protest for my non-meta picks, ruining very winnable games. It also lowers the quality of the game because people have a higher than normal likelyhood to throw or tilt. It is NOT as simple as just lowering your rating to the point where your skills equalize the lost efficacy of off meta picks/strats.
@@elvis4868 that isnt good. I dont want to play a game with a perfectly balanced win % if it means I have to be significantly better than the rest of my team to compensate for them actively trying to ruin the game. How is that any kind of good?
@@josephjoebrown11 I was in the same boat and I had reached 6k around 2015 -16. It just became impossible to climb after a while as me being at a 55% winrate, 70% wr with a specific hero, got me matched with people that we obviously the bottom of the barrel as far as mannerisms mattered. I am talking about people that ran mid exclusivelly on the first time you pinged / talked to them. I was being matched with various skillsets, wildly varying in my experiences but mostly just bad overall enjoyment. I just play pubs now and never looked back.
The real issue is that in a lot of games you can queue solo with and against duos and trios. And games usually try to balance it by getting both teams having a duo or a trio if the other does. So in effect solo players are at the mercy of the level of communication in their duo-trio vs the enemy duo-trio.
What imo also needs to be pinpointed is that sometimes the biggest problem is that the multiplier on the mmr gain and the difference of mmr between two ranks/the size of the mmr scale is so disproprotional it requires an absurd amount of games to actually reach your true elo - and many people don't even play that many in one season. This can completly trash yout the entire system. People all over the place for long periods and somewhere in the middle there's a mix of most of the skill levels which creates "elo hell" What Grubby said is true only if we assume most or all people play enough to get to their true elo...many games reset twice or once a year and require 200+ games to get you there.
Yeah i feel like this doesn't get talked about enough - I played LoL for many seasons but never enough to plateau, some seasons 100+ games which should be more than enough to place me. Hovering at 60-70 wr for that long just means I was game ruining for the people placed at the ranks I was passing through - basically forced smurfing. This was basically impossible to escape because I just didn't have the time to play as much as the system asked.
Marvel Snap actually had no multiplier. Because of the "bet rank points on the result of the game" mechanic it had to make wins and losses give a fixed number of points based on bets taken, which also meant that your MMR didn't affect rank gain/loss rate at all. Because of SBMM this also meant that if you played a lot of games your rank point gain/loss should tend to 0 over time. How did they reconcile these things? They didn't! Easily the worst ladder system I've ever seen.
It took me literally 7000 hours to get to the highest rank in CS:GO. There are several ranks in that game where you can get stuck in elo hell where you will be constantly paired with griefers and trolls and lots of cheaters. I skipped most of them relatively easy but the last one was very hard to get through. I actually only got out of it because Valve tweaked the system that rates your performance and I retroactively got boosted up several ranks to the highest one because I was massively overperforming for literal years at that lower rank. Then CS2 came out, which is literally just an engine and graphics update, and for some reason Valve didnt port over everyone's ranks and we all had to start over again in silver. I have now played for a few hundred hours and only made it back up to Nova 3 which is about the middle rank. Because of being forced to play in silver league, I got massively reported by noobs and now again get cheaters, griefers and trolls every game.
Yeah, this video was not done very well or enough research was done. The fact that its possible to smurf at all removes all points in favour of "skill based matchmaking" when games get too hard, you just start again and own people over again. Defeating the point of skill based matchmaking entirely. If it also resets for everyone and it takes too long to get back to the point you were meant to be at, that means its even HARDER to get to playing against people of the same skill as you as then you have people who would never smurf, fighting people worse then them. SBMM is 100% horrible and the wrong way to matchmake.
Yeah, most players will never play enough games in a season to actually reach their true rank. This is going to be particularly true when small swings in MMR can have a large impact on your overall rating.
Yeah, in my defense I was partway through watching it when I decided to write my comment about my MMR matchmaking horror story that I messed up conveying due to being autistic. It was a literal cycle of lose-win-lose-win for nearly 3 days of gametime spread over 2 weeks. It did ruin Heroes of the Storm for me.
In my case, I knew Grubby wouldn't address the issue properly, and I predicted correctly, so I stand by my comment before watching the video to the end. I am an aspiring game designer, so I analyzed this issue in depth, for years. The issue is, and has been reported by Challenger players that LoL became harder to solo carry unless you're in the top minority. In earlier seasons, it was easier for one player to destroy the game. Nowadays, the game is more team reliant, and you have little control over the output. What people are complaining about isn't 50% win rate, it's that it's bad design to use ELO, which was developed for a single player game, in teams of random people. Real world organizations, like companies and sports clubs do not operate like this. Instead the MMR system should track teams of regular people playing together. The advantage ? It's the skill of a real team that is tracked. People don't tilt and go about harassing strangers because they have a bad game, or run it down (feeding). In a real team, people communicate, teach, learn, and analyze games together. They also scrim against other teams. The modern MOBA style of game design is the antithesis of normal social behavior, but AAA game publishers prefer to let toxicity fester, and I am not sure why it makes more money to have the player base so aggravated and unhappy (many Challenger players dislike LoL, but it's their job, so they stick with it). Besides, at their level, arranging teams for skillful play, is probably easier.
@@Edin116 I think it is worse when you get stuck in your own head. I have reached legend rank in hearthstone 3 times as a ftp player, and EVERY TIME I reach diamond 5 and 4 I get nervous and start making mistakes, enjoying the game less, and getting extra toxic. Then I stop for a few days, get a new deck, and all is roses again. I can only imagine in games like Dota and League this is exacerbated because you are sharing your own horror with 9 other people.
@@lilhaxxor If it's harder to carry and the game is more team reliant, then ELO based SBMM is doing it's job perfectly as intended within a team game. You can't cater the entire system for players who want to play within a team game as if it's not a team game (i.e. "carry"). There are plenty of 1v1 games available to play for players who don't want to rely on teammates.
@@nankam It's not what I am talking about. You ignore the main point as if you didn't read, or maybe you only care about your own opinion, or I didn't explain myself well enough. I said the system should not rank players part of a random team using a system created to rank individual performance. What people are pissed about, but fail to articulate about the forced 50% win rate isn't that it doesn't work, it is that it is dysfunctional and frustrating, takes too long to learn, and doesn't reward improvement like it should. If I play solo, or I play with the same players as a team, using elo works, and is fair, and doesn't result in as much toxicity, because you have more control over the result. People are mad because with strangers, you win or lose based on random factors more than your own performance. Obviously, there are limits to what I am saying. A top player will carry most of the time, and a bad player will lose most of the time. However, LoL design, and I am speaking of solo queue here, could be improved easily, to make it more healthy, friendly, and fair/rewarding by a simple change in mindset. The fact that everybody accepts the status quo (including Grubby in this video), as if it was law, and there is no other way to design an enjoyable competitive system... baffles me, because it's just shortsighted (or dare I say lazy). Get to the root of the frustration, and fix that, instead of arguing for punishment systems that only deal with the symptoms.
There are matches where I feel that I'm the worst player in my team. Losing these is very different of matches when my team is way below enemy team on skill. I have no objections when matchmaking gives me better opponents when I'm on a win streak. I have objections when it gives me two noobs and opponents are as usual
Just imagine the amount of times you yourself have been the noob on the team. Or the amount of times you've stomped a noob but chalked it up to your own skill.
@@Dancyspartan 1st of all I always say I'm noob, so ppl not expect much from me. So ppl can either relax and play for fun or try to win with consideration that I won't help much. Many noobs in my matches wont tell anything and prefer to make me find out they a 0.5 of a player in the middle of fight, when you expect some help
@@DancyspartanI don't think that means what you think it means. In the games where he's the noob then the problem is that he's thrashing another one's game. It just reinforces his point really. You tried to make it about him so that he doesn't know how to answer but it is actually not a counter argument
Little HotS story from myself: I was playing ranked with my brother and a friend. I came from DotA and was a pretty good player, so when I started playing HotS, I was performing better than my brother/friend. We played around 100-150 normal games. When we started then playing ranked, something weird happened. My friend/brother, who played more than me, and played ranked before got seeded into diamond/gold, meanwhile I got placed into bronze, even though we played all our placement games together and I had better stats than them in every game.
Matchmakers will account for your performance in previous seasons when assigning a rank to you. So my guess is that your friend has already been marked as diamond adjacent player, before the matchmaking even begun. In fact it probably boosted him higher then it should have because he had an uncertain player with no prior history in all his games (you).
I cannot speak about whether anything is "ruining" multiplayer games as i don't have that kind of experience to make a judgement like that. The only thing i can speak about is what gets in the way of my own enthusiasm for the multiplayer experience, what drags it down and what bolsters it, and right now i can say what's "ruining" multiplayer gaming for me personally is complex team based games. If i lose a game, i want it to be my fault. If i win a game, i want it to be a credit to my abiities. Back when i was younger and heavily invested in multiplayer games, there were only two kind of games. Competitive PvP games where individual player ability was instantly and easily measurable, either because the game was 1vs1 or because the modes were simple and transparent (I.E., capture the flag in a shooter), and team based PvE games (like MMOs), where a bad player could cause the party to lose but was easier for people to get good because the skill ceiling for PvE content was basically fixed at a certain level. Whenever i play a multiplayer game today, it's like i entered this mysterious, esoteric realm where nothing anybody does can be easily measured. I can only imagine this is what leads to a lot of the toxicity you see today. Everybody is suspicious of everybody else. If a game is lost, they blame their team mates, even though it's not immediatly obvious that they played badly. Even the topic of this video can only apply to team based games. No amount of shenanigans from the match making system is going to drag down somebody like Happy. In 1vs1, if you are good you are good, and that's where it ends. And if you are bad, your loss is your own. Nobody is going to hate your for it, except yourself if you are the kind of person who enjoys auto commiseration and dragging themselves down.
Honestly my main take from this video is that team games should indeed go towards performance based matchmaking. Even if the only difference is that improving/failing players rise/drop faster, then it's already a big step towards perception of the game.
At the same time that's why team games with random matchmaking got popular. You lose? Not your fault? You win? You did great! Majority of people tend to lie to themselves and team games where people don't play with their own proper team are great at feeding delusions to those who want it. Also split responsibility means less stress for many - back in the days lobbies were filled almost exclusively with tryhard folks. Quite natural, competitive environment for competitive people. More causal gamers sticked to single player content only. With how modern monetization works, multiplayer gives higher pressure for people to huy skins - so even casual player are pushed into online lobbies deposite it's not really environment for them at all. Modern multiplayer game principles are just... Weird when anybody will think about it for a while.
@@Anterwaare306 Performance based matchmaking is something that almost always sounds better on paper than it is in practice. It can lead to perverse incentives where players start making decisions to maximize their MMR instead of trying to maximize their chances of winning (like intentionally prolonging games, or to try to "kill steal" from teammates even if it's worse for the team as a whole etc.), or it can lead to situations where certain playstyles can result in a 50% win rate because they're playing in a way that gets more/less "bonus MMR points" (if you get a lot of extra MMR points by your playstyle but actually are at the MMR you should be at, then your MMR will increase and then you start losing more than 50% of your games because you're at a MMR that you just shouldn't be at, but because you're getting bonus points you'll stay at that MMR even though you aren't actually good enough to be there). I could see having performance based MMR to handle outliers where someone is obviously performing way way better than they should be, but it should only be designed to handle outliers and shouldn't be something that average players should ever be seeing (and it also should definitely never exist at all at the competitive ranks, it should only be a way to get much higher skilled players through low ranks quickly and nothing else).
@@asdfqwerty14587 Performance based matchmaking works really well. There are a couple of old school games where people have programmed their own MMR according to multiple variables, one of them their individual performance. It gets constantly updated obviously, those systems are not perfect and never will be, but c'mon, a multi million dollar company can't figure out matchmaking and one dude alone can do it in his spare time? Pretty sure matchmaking on these popular videogames is bad on purpose.
Great video. You missed one thing though. It a player climbs to diamond playing Camille top and decides to pick his 25% winrate Teemo in your game. We’re no longer playing at the same skill level with different weaknesses. Instead there’s a gold teemo in your game.
That can also happen for counter picks, where the diamond top gets counter picked by the gold top, and now the diamond top has the game impact of a gold player, and the gold player has the impact of a diamond player, this can only happen because match making happens before champion select, and not after, as well as champion select being dodgeable without losing MMR
@@soldier22881oh good grief get a clue. If a company wants to f up your PC or steal your info they don't need a kernel level anti cheat to do it. They could just put it in the code of the game itself.
I love these type of videos, I could listen you talk about almost anything. Which includes the Warcraft 3 tier lists even though I haven't played or watched W3 in at least 10 or 15 years.
Resetting the MMR wouldn't actually remove smurfs and boosters. It just effectively makes _everyone_ both a smurf and a booster at the same time. Your matches would just stop making sense, and if anything you'd start to feel what it would look like if the matchmaker _really did_ go out of its way to pair you with and against unfairly imbalanced teams.
3 місяці тому+3
Disagree. It resorts in no time and let's people out of Elo hell created by playing too many games after coming back or too late in their day.
A lot of rankings, like LoL, has too many players that are "boosted" and remains at ranks they don't belong at from several seasons back from being boosted once
If it resorts in no time, then you're going to be complaining about it again in no time. If it doesn't actually resort in no time, however, then resetting everyone to zero just puts you in a far, far more extreme Elo hell. Without MMR, the system does not discriminate: it will not hesitate to make half of your games trivially boring, and the other half stupidly unfair. Did you hate playing with a couple of diamond players in a masters level game? If so, then boy are you going to love playing against two diamond players, two masters players, and a champion when you got stuck with three bronze players and an iron. Do you enjoy when your decisions matter in a game? Well, too bad, next game the rest of your team is several ranks above you. You might as well just AFK this match - you won't even come close to changing the outcome anyways, and being AFK lowers your chances of throwing the game by feeding too much.
@@Gofex1337 That's a fair point, boosting does cause a huge amount of problems for the matchmaker when any of the other 4 people in your game can be one. But turning to the matchmaker to solve a problem caused by people circumventing said matchmaker won't really accomplish anything. In fact, if anything, it would make the problem worse. By resetting everyone's rank, the demand for boosted accounts will skyrocket, and boosters will make a fortune off of it. A fortune they can spend by advertising and creating _even more_ boosted accounts to sell. The solution there is to search for clever ways to catch boosters without generating any false positives, so you can ban the accounts in waves. Then people get angry at the boosters, stop using their services, and the boosted accounts all get removed with time.
@@Gofex1337 No one in League remains at a rank that they didn't earn if they keep playing. That is pure copium and complete ignorance of how matchmaking works.
the only thing worse than buffing high skill characters because they are allways bellow 50% win rate , is buffing characters for years in a row because they are bad in high skill or pro play , for example tridamere or nasus in league , or sniper or SF in dota 2
Reading the chat is the same experience. The conspiracies I was seeing were hilarious. All these obscure matchmaking rules people were inventing that are clearly untrue. I wonder why no-one who's Immortal or high rank has these conspiracy theories about matchmaking?
While 50% matchmaking is more of a phenomenon than an intended effect, the "churning" system that league of legends Riot Games have developed and patented is very much real. The churning system is a system designed to keep you at a positive winrate for the first 5-10 games you play, but to slowly reduce your winrate as you play in order to get you to keep playing with the sentiment that "you cant end on a loss". Its stared in the patent that it is a system designed to slowly increase frustration in the player at a rate that doesnt immediately turn them off playing the game. It has a flagging system for "toxic" players who often get reported by their teamates, or also teamates who are in roles they dknt normally play, aswell as putting you on with teamates who are already deep into churning and frustrated before the gsme even starts leading to reduced chances of winning. The thing is, the greifer is never just one person, and they are always on the same team 2/3 of them. The entire game, they will be typing and arguing in the chat over nonesense, and at the end are all calling for reports on their teamates in all chat. What i see from the opponents all chat is "gg". Then in games im "supposed to win" according to the churning algorithm, the greifers are all on the enemy team. 8 times outta 10 someone from the enemy team will say "please report xyz, they are refusing to play with the team" or the argument will spill over into all chat with opponents trying to grt my team to side with one of them to report the other. Evey. Single. Game. One teammis coordinated, clear headed, and on the top of their game. And the other team is a toxic cesspool of a crab bucket with the majority of the enemy team pulling eachother down. As aomeone who was heavily addicted to league, i actually noticed when the matchmaking changed to implement the churning system, because we stopped shit talking the enemy team in all our games, and somehow, started shit talking our allies. So while 50% matchmaking doesnt exist, there are systems currenty active that are designed to make players just wanting thst one hit of dopamine for winning their game, but the game wont give it to them... so they queue again...
Yes, it is 1000% real. when i used to play, i would relish the one game out of 20 that was a fair match and watch the replay over and over again after 19 crappy games
I have 3000 hours in Dota (quit a while ago) and people always said that the 50% forced winrate doesnt exist. I can pretty much confirm that it is there, because im not dumb but I cant test it scientifically obviously, but its actually industry standard to have matchmaking that manipulates the winrates for people. If forced 50% didnt exist, you would expect YOUR TEAM to, on average, be EXACTLY the same skill as the OPPONENTS TEAM. In theory, as you climb the mmr, it is YOU that can't keep up with YOUR TEAM and the OPPONENETS TEAM. You just dont have the skill and drag your team down causing you to lose and drop mmr into a 50% equilibrium. This doesnt actually happen though. What happens is, generally, is that after a winstreak it is YOUR TEAM that becomes WORSE than the OPPONENTS TEAM. It is very evident when you play, all the sudden you start to see people making stupid misstakes or just trolling on your team, despite you having higher mmr than before. This shouldnt happen if the forced 50% mmr did in fact not exist, but we all know it does.
@@Skumtomten1 >YOUR TEAM that becomes WORSE than the OPPONENTS TEAM. Yeah, because there are 5 higher ranked players on the opponents team, and 4 higher ranked players on your team, and you.
@@zym6687 That's not what he said though, what you are describing is expected behaviour that will say this is not your place you are not good enough, you are the reason your team lost. What happens is that you are still pulling your weight, maybe even winning your lane, but your mid and top for example are 0/12/0 combined in 10 minutes. I am talking from personal experience, it happened so many times that it can't be accidental, especially in your first series of promo games. That of course does not mean I was not on opposite side where my opponents are trolling and feeding. But the bottom line is that I don't want that either, even if it means free win for me. I would be so fine with prolonged queue times if it means better match experience. I would rather wait additional 3-5 minutes than get stuck in a guaranteed loss for 25 and have miserable time. I am talking about LOL, that's the game I played the most. Similar problem you can see in world of tanks, where when they changed matchmaking, a 15:3 was most common way the games were resolved (meaning that one team killed all 15 other tanks, and lost only 3). Around 80-85% of the games were one side steamrolling the other and nobody had good time. Not sure how it is now since I quit that as well
Yeah but if you or someone else is carrying then the shit talking just stops. You need to create strategies to stifle mental bleeds and mitigate morale loss. This is not just a performance based game but a team game. If you're good enough it doesn't matter but if you're stuck down here with the rest of us it helps to know how to keep morale high.
Thank you regarding hard MMR resets. I believe it's mostly to promote seasonal rewards and battlepasses but there should always a soft cap when reseting MMR cause the 1st week is just full of bad one-sided games. Also I loved the introduction of sbmm because I remember playing older games which only had lobbies and it was a complete crapshoot where you had master players paired with bronze players by skill averaging and everybody just felt frustrated, unless ur one of those people who enjoy noobstomping.
Even with a soft MMR reset it often skews too much towards the "reset" side. I main LoL and the game's ladder is literally unplayable for at least 1 month; often 1.5~2 months (and they've made it so the reset happens 3 times a year now because retention metrics or something I guess. Woohoo.)
@@FreakattakerThese 3 month resets are terrible imo. First week you play low quality games feels your not learning its just faceroll winning, then you get to the top of the ladder and queues become long and matchmaking quality goes down even more cause now, you essentially have 2-4 pro players matched with 6-8 terrible players, which in mobas where the game snowballs so hard if your adc is pro and the enemy adc is terrible the game is over. Doesn't matter what you do, its over. On the other hand if its in reverse than you lose regardless. So you are forced to play like a 100 trash games in order to play decent players and then its who gets the pros in the key roles. Or which terrible player is capable of not feeding immediately. And also the reason your on top of the ladder is not because you are the best player, its actually because there are many many players who skip this period so they can boost their winrates once there are less pros in low rating. So essentially it takes like 2 months for you to actually be able to play proper games at your skill level. Then you can play for a month and then back again at it. The only way around this is premades. Not only that but you need 5 man team of top 0.1% and even then most of your games are boring easy wins. Mmr resets make mobas terrible, its always this cycle repeating and if you play soloq then you are in bad luck cause at the top end everyone plays premades only so after a while you have no real competition in games and if you do its winning or losing is still not decided by that. Thank god im not invested in being good at mobas anymore.
There's one circumstance when MMR resets make sense, which is during the development of a game when the game is rapidly changing and not finished yet. But besides that, yeah.
@@TheRealXartaX But why reset if someone just plays they will lose their mmr anyway, artifically raising and lowering mmr is annoying for serious players. Its this mentality of give people a sense of accomplishment, but it just doesn't work above a level. People who actually try know its meaningless, your mmr doesn't matter your rank doesn't matter, all that matters is did you play perfect or not. And also games like HoTS have this backwards system of giving people MVP, this is the same issue. How does a game decide who is MVP, is just a joke. It doesn't deeply analyse who plays best it just looks at a few numbers without any context and based on that it tells the player good job. Again this doesn't mean anything, it is completely arbitrary, it is the same as in chess where if you win a piece you get points. Great but I don't win by getting points so it doesn't mean anything, and it just makes most players think they are good and reinforces bad decisions. For most people mmr resets just means one thing, inflated self esteem. Sounds good, but then you realise do players in team games really need more self esteem? It is a recipe for disaster, now not only that skill gaps are all over the place, everyone thinks they are rank1, leads to toxic behaviour. Who came up with this?
@@AnonYmous-spyonmepls Because a reset in that case will literally be closer to people's real mmr than where it was. When a game is rapidly developing with huge changes you will have things like broken strategies that gives people hugely inflated MMR's if they discover it and so on. And I'm not talking about something like a character being a little too powerful, I'm talking about things like being able to clone relics in AOE4 so you generate infinite gold.
I think the issue most people have is that "Casual" game modes have too strict MMR groupings or use ranked MMR as the basis which leads to them feeling too sweaty for people just trying to have some fun or learn or play off meta. As you showed in the video player skill usually sits on a Normal distribution so having tight MMR groupings ultimately doesn't change your overall win rate it just lowers the potential "variance" per game which could be seen as good (close game) or bad (sweaty game) and having lots of "ranks" across that distribution (beyond 3 or 4) is probably meaningless but could be seen as good (yay I am ranking and improving) or bad (we have a 2 rank 5 diamonds on our team and all their team are rank 3 diamond and above). I never played SC2 but I agree with idea that ranking system are at their best when its more a recognition (not fully) of the amount you have played because of what I outlined above and if its designed well it increases positive engagement (losses don't feel as bad) plus you can more easily couple it with re-occurring "resets" which also increases engagement.
I like how the section where you talked about increasing player satisfaction was basically what EOMM was trying to do, but just better. Instead of aiming to essentially rig your games to keep you playing for longer, things like bonus pool give you an additional reward that boosts engagement more effectively than EOMM could without unnecessarily eroding your match quality. The problem with EOMM is that by seeking only to maximize how much time we spend in-game, it completely misses the point of _why_ we choose to spend our time with a game in the first place. It is a system based entirely around doing what it _thinks_ will minimize the chances of us quitting - a metric partially based on statistic like play time and games played, but is also in large part based on your tolerance for losing streaks. Think about it: if you tolerate losing more than the average person, EOMM *should* be harsher on you. If you can handle a 5 loss streak, then you're the perfect fit for 5 higher rank players who would quit after a losing streak of just 1 or 2. Sure, you won't have fun. But you also won't quit either, and that's the only metric that matters - which makes you the perfect candidate to be today's punching bag. Until, of course, the day you decide it's not worth it and quit for good. The algorithm can read data, but it cannot read your mind. And sometimes it cannot collect the data to know when it has pushed you too far until it's already too late. Yet, even if it _does_ learn to let you win barely enough to keep you playing, it begs the question... if the game had to trick you with an imbalanced W/L ratio just to get you to stay, then was it even worth playing to begin with?
Optimizing play time, depending how you collect the data for creating the system, can end up only optimizing for the near future. Because how can they know the system will burn more people out over 10 years when the game has only been out for 2 years. Just because someone is temporarily hooked and plays very consistently doesn't mean they'll be around forever. Maybe satisfying them more but having them play less could net you more hours of play time in total and more secondary traits, like buying microtransactions or new expansions
huh world of tanks uses a similar system instead, the punching bags are free to play players (me) and the winners coincedently happen to be the ones spending money on the game using very souped up tanks. glad i quit wot
@@auberry8613 I don't think gaming companies really care about making evergreen games that would last for 10 years. I'm afraid they will take notes from the successful social media companies which waste our time with ragebait and misinformation to keep us engaged. If these kinds of systems are implemented successfully for matchmaking in games, they might persuade us keep on playing while making the complete experience worse for everybody.
Some thoughts. 1. criterium free matchmaking works fine for casual games like cod or team fortress 2 where most people dont care about winning or losing and just want to have fun, play with certain weapons, farm unlocks and experience, try goofy strats or just practice their skills at certain game aspects. It doesnt only work well for games with low player counts. I had plenty of fun playing cod without rank based matchmaking and so did everyone else back then, and those games were huge in the cod4-mw3 era which is when I played. 2. you learn a new game way faster if you are randomly matched with good and bad players in a wide spectrum than if you only get matched with other noobs at your noob level who dont know how to play either. You will die to the good players, you can spectate them while you are dead, you see what they do and how they kill other noobs. It also gives you clear feedback, because if you manage to kill a good player you will know for sure that the play you made was actually good and not just the other guy playing even worse than you. You will just have to push through losing a lot of games until you git gud, but if there is no ranked matchmaking, you also arent punished for losing in any way, you can just go play the next game against other random players with random skill levels. 3. SOME games do give you worse and worse teammates if you win a lot of games. Often this is defended as a measure to prevent smurfing, people who make new accounts or derank intentionally to noobstomp. This is to punish smurfs for smurfing and prevent them from having fun noobstomping, which discourages smurfing. You wont have fun smurfing if your teammates are such bots that even if you full tryhard you will only get a 50% winrate. This is 100% certain a feature in counter strike's matchmaking. Valve even specifically talked about implementing such a measure to combat people smurfing on new accounts. The thing is, it also happens to me on my main account with 8000 hours. What it actually does is prevent good players from ranking up, forcing them to "smurf" indefinitely, even on their main account, if they get placed in a lower league in a season or after a server reset. 5. As mentioned, I have over 8000 hours in CSGO and achieved Global Elite, the highest rank. When CS2 came out everyones rank was deleted and everyone had to start again from silver. I barely managed to grind up to MG2 in a year since the game released because of these anti-smurfing measures constantly putting me in the absolute worst games with noobs and griefers and cheaters where I have to drop 30 kills every game to have a 50% winrate. If I inform the enemy team that one of my teammates is blatantly cheating, the cheaters report me, making me even more likely to get queued with more griefers and cheaters. I also get kicked all the time by people who think Im cheating because Im not supposed to be at a low rank, Im supposed to be in the highest rank with the other 1% of the best players. This makes a terrible experience for literally everyone on the server. I cant inform enemy team they are playing against cheaters, and I cant overperform relative to how good I "should" be according to my rank, which is way too low, or otherwise I will be punished by my griefing and cheating teammates as well as the matchmaking system. 5. Another issue besides MMR reset is MMR decay. If you dont play counter strike or starcraft 2 for a few weeks, the game will just start steadily lowering your elo. If I come back to SC2 after a few months of not playing, I have to grind up from 3000 to 4000 mmr every time, just to get back to my actual skill level. That means i have to get 40 net game wins (minus losses) to gain 25 mmr per win, 40 times. This means I have to smurf for at least 50 games, assuming I average 80% winrate against players significantly worse than me, just because Blizzard wants to punish me for not playing the game for a while. This is not that much fun for me, although cheesing noobs can be fun, but it is definitely not fun for my opponents. The bonus pool doesnt apply to your visible or hidden mmr, only to your league points which literally nobody cares about and is completely irrelevant. So SC2 just like WC3 does punish you for not playing. 6. EOMM is definitely a thing mostly in games with microtransactions. If you buy them, you are matched against lower skilled players so you win, and if you dont pay, you are matched against harder players, so you lose. This makes people subconsciously associate buying microtransactions with winning more, without being aware of it. It is very effective and probably illegal under EU law. 7. Engagement optimized game design is also a thing. Games often add some weapons or heroes or items that noobs can use to get an easy kill every once in a while. These weapons are often derided by the community and you will get flamed for using them, but devs keep putting them back in. Famous example, in the original COD MW2, the first unlock you would get on every assault rifle after getting like 20 kills with it was a grenade launcher that oneshots players in a radius the size of half the map. Noobs could use this to get some kills on players 500 times better than them and almost everyone hated it, but Activision kept putting the noob tube in their newer games precisely because it levels the skill gap and prevents noobs from losing games with a 0-30 scoreline.
I play Starcraft II, and I actually like MMR decay. I often go a few months without playing, during which I lose some of my ability to play the game well, and the decay normally works out well in terms of matching my actual ability after not playing for a long period. This means I don't have to go through a long series of losses from being matched against players that I _used to be as good as_ while I get back up to speed when I start playing again, and makes my MMR more representative of my actual ability. Different people learn and forget at different rates though, so the system won't work for everyone.
1. Even in casual games, only some of the players actually play casually. Some one did a video on this, back then, people were playing more casually but now there are so many streamers that are playing more sweaty for their content and also push out to others what are the most optimal strategies. 2. If a good player is so much better than a noob, the noob might not even comprehend what’s happening, let alone learn anything. The number of people that are willing to push through an overwhelming number of loses and also do lots of side research and learning of all the advanced mechanics is incredibly small. Baby steps is the better option for most people. 3,4?,5. Only pointing out your contradiction: if you think noobs should play with better players so that they can learn, then why are you upset when you have noobs in your team? You just said that you think it’s a good way for them to learn.
@@nathangamble125 well I lose 1000 MMR when I don't play and my actual skill only decays by maybe 200 points, so maybe you should just lose until you get to your actual skill level instead of the game forcing me to smurf for 50 games.
@@RevCosmin it's good if I get some noobs in my team, not if i get put in a team with 4 noobs against people 5 ranks higher than them and i have to get three kills every round. Ideally every team should have some experienced players, some mid level players and some noobs. Playing meta only has a marginal impact, I can play off meta in a game I'm good at and maybe sacrifice 5% of my optimal results. In fact that's what I'm doing by default because if I spam smokes every round in counter strike and accidentily get some kills my account will get reported even more, I will push my trust factor so far into the ground I will be in the third basement.
@@RevCosmin "the number of people who want to push through losses, do research and learn mechanics is very small". Yes, I agree. The people who don't want to do those things should play casual games instead of hardcore competitive games that require you to do those things to be able to win games. Instead of playing csgo or sc2 they could just go play cod or some autobattler that doesn't require in depth knowledge of mechanics and strategy and have 300 apm and high accuracy and being able to communicate while doing all those things. Or idk play Minecraft or something. Inviting people who don't want to grind for skill into games that require grinding just sets them up to either have a miserable time or start cheating.
Just to correct you on the SC2 system a bit, SC2 has MMR as well as ladder points. The matchmaking and your league is dependent on your MMR, while ladder points is just fluff (although it is slightly correlated to your MMR within your division). Bonus points added points to your ladder points, so it is just fluff on top of fluff.
I think most of the frustration comes from not understanding the matchmaking rules and feeling "cheated". This is understandable since matchmaking code is almost never open-source and many games will hide your true mmr score. It also makes it impossible to determine whether the matchmaking system is truly broken. And the system *could* be broken. MMR depends on an iterative mathematical model where the MMR scores of players are used to predict the likelihood of an outcome and the true result is then used to adjust the MMRs. The quality of this depends heavily on how accurate the mmr-based predictions are and how mmr's get adjusted. If mmr's get adjusted by too little then the system takes ages to stabilize while if it gets adjusted by too much then the system becomes chaotic and may never stabilize. This is made even harder when you consider that a player's true mmr changes over time. I don't think we can discard the possibility that the matchmaking for some games is deeply flawed and it's frustrating that we do not have the data to prove it one way or another.
Let's not pretend the matchmaking in team games like dota actually works for majority of people. When I still bothered to play finding actually balanced game was about as unlikely as in a game I'm currently playing that has NO sbmm at all. Its usually a stompy win/stompy loss
The problem is skill based matchmaking DOES match you with potatoes. Many games don't or poorly calculate individual skill on a team, so you have people who get a good rank because they got carried, or have bad rank because their team sucks. Skill based matchmaking is bad because skill calculation is bad. Another reason it doesn't work, is the human element, because some people have off days or really good days. I think every game should have a ranked and unranked. It's fun to be able to destroy noobs. It's also fun when you have that one god tier player you have to really avoid.
This is true, especialy in like 1v1 games. In team games, this is true aswell, but there a win is a team achievement, not depending on a single person. I agree with you that you can always improve and you definitly shouldnt try to always shift the blame onto others. But you should know that solo carrying in a team game is extremely difficult, especialy if the odds are heavily against you, which is why a forced 50% winrate isnt rly fair for players.
@@crisscore157 If I wanted to win; I would play meta. Hell if most people wanted to win; they'd play meta teams. You're going to have bad teammates; tough luck, just like with a job you're going have to try to do the best you can with the hand you got. There are other gameplay systems that could be implemented to make things less hard carry if it's such an issue.
@@soundrogue4472 i agree, but you cant pick for your team mates, can you ? Im not disagreeing with you and i do believe that you should always strive to improve (and ofc dont shift blame if you could perform better). I just wanted to add that you kinda need a certain amount off luck and if that is rigged, it gets harder to win
@@crisscore157 Sure you can pick teammates; get friends. I am stubborn and not going to budge on this. This is like going to the Casino and being shocked when the government lets them not give the pay out for what you were owed if you find a way to game the system.
Lmao stop yapping there’s a reason even the best players in the world max out at 50% across the games he used as examples in the screen such as dota and league. Obviously you’re not the best player in the world yet the statement stands, there’s a 50% wr system holding players back in which yeah if you’re better than everyone else you can win 1v9 yet having more difficult games due to having higher wr and the game not wanting you climb too fast is clearly unfair.
@@mackparker9064there is kind of viral theory of how matchmaker deliberately throws you to players with a bad time or lack of skill or/and gives you strong opponents so your win rate stays the same. It is particularly popular in DOTA community
@@mackparker9064but the conspiracy is that matchmaker of DOTA matches teams not only by their shown MMR, but "hidden" MMR (which is calculated by performance statistics and maybe something else) also. And these hidden performance statistics could be manipulated to match teams so that one team loses and other wins.
@@gregory7090 I can assure you, it's not just popular in DotA community, but in all MOBA community. It comes with the genre I guess, being 5v5 game and all.
What a great video.. btw a good recent case study is Overwatch 2. People used to complain so much about it but not any more and it increased participation in ranked games. Basically, among many things, they made the system update more frequently, more transparent, and flexible. Rank used to update only every 7 games, now it's 5. They tightened the MMR spread AND allowed wider grouping in ranked games ... resulting in two kinds of matches.. the tight matches and the wider matches.. and told people that wider means more punishing losses for the higher MMR people.
A lot of good "theoretical" points, but until MMR algos in games are open source, that's all it is. DotA2 was also "talked of" as using a fair and well implemented MMR system, and then it became known that it intentionally took extremes of both Elo and Attitude to create "equalized" teams, leading to hilariously bad matchups and experiences. It is always easier to create a system that cheats, and claim that it works (make a system that forces 50% win rates, and then say that the 50% is how you know the system is working), rather than make a system that actually works.
Dota2 matched feels more like playing a slot machine than a matching system. At max behavior and communication scores I often get placed with crazy toxic griefers and leavers after a win streak. I then go onto lose like 5+ games in a row where I will still have a very good game but I can't out carry the tozic behavior, role abuse picks, etc. I was still climbing but it was a long exhausting grind where most games are stomps, many matches there are people with under 500 matches pitted against players with thousands of matches. Other matchmaking in other games did not feel like this. Something must be wrong. I feel like it might be because they're cooking the books for a paid subscriber base that is mostly toxic and terrible.
@@parawizard I have 3000 hours in Dota (quit a while ago) and people always said that the 50% forced winrate doesnt exist. I can pretty much confirm that it is there, because im not dumb but I cant test it scientifically obviously, but its actually industry standard to have matchmaking that manipulates the winrates for people. If forced 50% didnt exist, you would expect YOUR TEAM to, on average, be EXACTLY the same skill as the OPPONENTS TEAM. In theory, as you climb the mmr, it is YOU that can't keep up with YOUR TEAM and the OPPONENETS TEAM. You just dont have the skill and drag your team down causing you to lose and drop mmr into a 50% equilibrium. This doesnt actually happen though. What happens is, generally, is that after a winstreak it is YOUR TEAM that becomes WORSE than the OPPONENTS TEAM. It is very evident when you play, all the sudden you start to see people making stupid misstakes or just trolling on your team, despite you having higher mmr than before. This shouldnt happen if the forced 50% mmr did in fact not exist, but we all know it does.
Amazing video as always Grubby. I find this a very interesting topic (as an engineer working on similar non-gaming algorithms) I'd love to see a follow up that explores the insights that has come from different 'less subtle' variations to the matchmaking algorithms to target certain outcomes that has been published publicly or leaked (and confirmed). Some of these systems are developed and sold to studios (or in the case of Activison, they have spent a lot of time and money developing and testing multiple iterations and published (or had leak) their findings).
One thing that wasn't mentioned and i think is the real problem with our games mmr system. Is the hidden mmr. The hidden mmr is the real number the matchmaking system is using to decide what players you should be playing with, instead of the mmr that you currently see on your account. For example Wow arenas has this system visible to the player. Hidden mmr is bad for the matchmaking system because when you win a game it increases the hidden mmr based on your whether you win or lose and your performance relative to the other players, sometimes you can have a really good game just because you had a good matchup. So due to you being extremely lucky versus those opponents you will get a high performance, causing the game to increase your hidden mmr by a huge margin. Then when you decide to play the next game you are being matched with players that are literally 200mmr higher than yourself, causing you to feel like you are just a totally complete noob and making the game feel more unfair than usual. This hidden mmr has a compounding effect as well, so if you get into a high win streak your account mmr will be even further behind your hidden mmr and raise the bar of enemy players even higher, eventually resulting in a lose streak bringing you back down to your true mmr. The blizzard post about the 50% forced win rate compares players/people to a random coin flip, this isn't true people are more consistent than a random coin flip even including factors like tilting and so on. if a player is 2k mmr that means that player is 2k mmr, if he goes against someone that is 2.1k the 2k player should lose unless they ARE better than the 2.1k player. It's not a random coin flip that makes the 2.1k player better than the 2k player, the 2.1k player IS just better than the 2k player. If games literally just used the mmr you see on your account regardless of your win streak or loss streak the game would feel more fair and you would consistently see more players with 1W 1L back to back in their match histories.
This definition of hidden MMR is a little misleading. Usually, hidden MMR operates with a higher k-factor(gain), such that if you win a match, your MMR might increase by 10, and your hidden MMR might increase by 15. If you consistently outperform your MMR, you will enter a new rating bracket with a new K-factor for hidden MMR that is closer to the one that the regular MMR has. This is done for variance control, and under the assumption that skill increase and skill decrease are not equivalent. Your skills grow faster than they decay in this model. Players at the absolute top experience very little or no difference between their MMR and the hidden MMR, while players in the middle and bottom are given the opportunity to more rapidly climb towards a 50% winrate when they experience skill growth. The number of data points is critical. In your example with the conflip it does indeed hold true that human performance is more consistent than a coinflip given maybe 10 or 100 data points. But as the number of datapoints increase, the coinflip will prove more consistent. I have no specific experience with Wow arenas match making system, but given your description, either the k-factor is too high for your rating bracket, OR you are not providing enough data points for the match making system to create balanced matches. In general these are the only 2 mathematical answers.
@rasmusvinther9190 from my experience the hidden mmr increase seems to be extremely volatile, one win my hidden mmr will increase by let's say 20, second win in a row it will increase by 40 then on the third win in a row it will increase by about 80, meanwhile my actual account mmr is only increasing by around 10, 15, 25 each time. Creating a massive gap between my real mmr and my hidden mmr. This rings true for pretty much every single pvp game I've played. Wow, arenas just make their hidden mmr visible. The problem I'm pointing out is that the game then starts to give you players that are way better than you are and thus makes the player feel like they are horrible at the game and had no chance. Rather than slowly graduating through the ranks and giving them a proper chance to help win for their team. As for the coin flip, yes coin flips can be streaky and over the course of 1000s of flips it will result in 50% heads and tails. What I was talking about is that humans are consistent if you are in the correct mmr you should not be getting ridiculous win streaks, you should be losing to players that are better than you and winning against players that are worse than you. Thus giving you 1W 1L most of the time.
The way arena in wow does this drives me nuts. Play new characters, start at 1500 mmr go up to 2k mmr almost instantly since you're old veteran players, but you're playing completely new specs and classes, trying to get those 1600 CR rewards, but you have to fight 2k+ players for them. It's annoying as hell. Just getting by on fundamentals and having to uber sweat every victory when playing to get your first epic pvp item for the season xD
@@kristoferkravis1474 Sometimes the hidden criteria include things outside the match. Be it behaviour or how long have you logged in. Stockholm Syndrome is a thing, i notice in moba when i talks shit, i get pair up with people that talkin shit. Sometimes when i too good, i got paired up with lower skillset, sometimes also the unlucky matchmaking. The game literally set you up to lose. At that point i just play for killin time and even got apathy on one point which actually help getting better team
Yall keep forgetting smurfs who ruin the who situation overall anyway because they dont want to play their main "forced 50% account" because that would require trying, so they want to clown on lower skilled people. For fun. Guess whos not having fun when this happens. Anyone else in a forced 50% mmr being forcefully carried or forcefully denied a win by someone who spends their life in the game. And fucking their actual MMR because funzies.
Didn't watch the entire video but it seems you were just talking about players personal perspective bias but didn't talk about seasonal ranking and MMR decay and hidden MMR which is what i expected the forced 50% WR dilemma was supposed to be about
It's no coincidence that community servers went away when games with micro transactions started coming out. The matchmaking is designed to reward players that pay and punish those that don't.
More importantly not watching the video. I personally wouldn't laugh at anyone for believing what people put in their titles etc, but taking time to talk about it without watching is a mistake.
@@jpVari is it wrong for people to want to engage n conversation without spending a lot of time watching a video? also, how do you know if they watched it or not?
@@snuffeldjuret These comments judging whether or not someone watched the video always come down to one question. "Does person agree with my stance?" "Yes? They watched the video." "Does person agree with my stance?" "No? They clearly didn't watch and just flew straight into the comments section."
I'm pretty sure most of the complaints about "forced 50/50" (at least that I've seen) aren't so much about the existence of sbmm resulting in that outcome, but rather people saying that devs take the easy route and chase the metric of "if people get 50/50 winrates matches are balanced", which can also help keep queue times consistent if it cares more about average mmr between teams than everyone having close mmr. That's ignoring Engagement Based MatchMaking, which I feel was glossed over when it's apparently becoming more and more common. I feel server browsers are easily superior. People typically will sort themselves out into enjoyable games and will take into account things the game simply can't like if they're tilted or on fire, rather than trying to jam everybody into one or two different queues.
Thought this was going to be about faction/character balance, which I agree with. Electric Underground did a video on street fighter 3 not that long ago, and one of his main takeaways was that an imbalanced roster where some characters have noticeably worse win rates than others is good for the game in the long run because it allows for a more varied competitive environment where good players can play bad characters and win because their opponents have less matchup experience.
Interesting idea to compare the meta to the FGC. It's actually similar in many ways. Hate to say it, but fighting games in general could benefit from the pokemon style tier bracket. i.e teir 1 everything is playable, tier 2 the top 10 are banned from competing tier 3 top 20 etc etc... It would be a cool way for lower tier characters or builds in wc3s case to have their own meta. Problem is, it worked for pokemon because it had 152 creatures to play with. No fighting game or RTS had THAT many variations.
@grizzgreen6821 I think it would be interesting for more games to have a tiered bracket system like you described, but I don't think it's really necessary. Just think of the example that Grubby gives in the video with the teemo player. It's a good thing that league has characters like teemo because his playstyle provides players with a change of pace, but you would never want teemo to be meta viable because of how much he differs from the norm, this naturally leads to good teemo players potentially having a game knowledge advantage that can allow them to sometimes cheese out wins, which is the main appeal of playing low-tiers.
Theoretically, anyway. In reality just about any 3S tournament worth a damn is absolutely dominated by the top tier characters. There is a nice grey area on the tier list where there are several powerful characters who can, in theory, beat the top tiers with sufficient skill but in reality the character specialists only really shine when the stakes are low (so nobody is really trying hard) or when the field is soft enough that they can win regardless of who is played.
@MrOrdinaryundone well, yeah. that's kind of a tautology though. Characters with low win rates don't often win. I do think it's disingenuous to say that upsets only happen when people aren't taking things seriously though. Every competitive game that has ever existed has had a meta that evolved over time with underdogs winning when they really shouldn't have. Not all of those victories were because their opponent was toying with them. Edit: Like, just think about moment 38. For a Hugo player, getting #4 at Evo is no small feat
@@danily11 It's not a tautology because the reason they have low win rates are because they are statisically inferior to the top tiers. Their own weakness informs their low win rate, its not like the are underplayed because they are difficult to grasp or play 3S is a very old and well studied game, if there was some tech that made Q or Alex able to hang with Chun or Yun it would have been found by now even if it was difficult or obscure. And by "serious" I don't mean the players BMing or not taking their opponents seriously I mean low tiers only really get spotlighted when the tournament's rules enable them, like team formats or bans. You have exceptions, like Hayao's Hugo play at Evo this year, but even as well as he played 4th place just reinforces the status quo when he's beaten out by Chun and Yun.
Great video with good explanations. There are just so many people that don't understand these systems but I can definitely understand some of the frustration.
I think the biggest hurdle is psychological and can't really be ignored. It is simply poisonous to enjoyment to feel like you're playing your heart out, and doing well, yet your efforts are rewarded with a loss by other people on the team. Sometimes you're mistaken on your skill level, sometimes they're having a bad day - but there is a concept of "Some games are clear wins 40$ of the time, clear losses 40% of the time, and it's the 20% of the time where your play mattered that causes your MMR to rise" which simply ignores how terrible it feels to have wasted your time. This is one of the reasons why, in DotA at least, some of the most toxic players are pos 5 supports - because they feel like their entire effort was wasted and they feel trapped by the system. This is what Performance-based MMR helps against, and why I'd ultimately advocate for it if I could. Because it, at very least, says to players: "You are struggling and I see you". And sometimes, that's all that's needed.
Yes, that's the problem entirely. It's absolutely psychological and not just for the person who's playing their heart out, because they will in turn rip on the not so skilled players who are forcing you to play your heart out to begin with. Toxicity is a natural consequence of your actions going unrewarded, and even punished because it's your turn to lose. They don't want even games, far from it. It's an absolute grind, and people have different temperaments either suited for that or not. But temperament should never trump skill. If you aren't rewarded for being the best performing player, you're doing it wrong and people will act accordingly
Imo any performance metric will be gamed by elo boosters and de-boosters. But it's nice to have some sort of reward for dying little, getting kills, etc to motivate ppl.
they need to make games where you can 1v5 instead of relying on teams, which is why tf2 is so much more fun to play despite it being a "team" game i checked the dev commentary and they balanced it around anyone being able to carry thats why its so good, you can carry against 15 players and it feels awsome.
the biggest flaw imo with sbmm is that it is a concept invented for 1v1 games and then applied to team based games. It is also often applied to f2p team based games which opens up the door for a lot of bad actors (smurfs, trolls, cheaters) to have multiple accounts. Combine this with our tendency to remember and emphasize bad things (ie loss streak, smurf, leaver) more so than good things (win streak, got carried etc) and you end up with a brew that feels bad for most people to grind through every day. In a team based sbmm game you could very well be a plat player stuck in gold, which can never happen in a 1v1 sbmm game. Are you better than gold players yes, but not by enough of a margin to just carry 4 lemmings when the odds are too stacked against you. the closer you are to where you are supposed to be, the harder it gets to make up for mistakes made by your teammates. A solid argument is that, with time, and many, many games, you will eventually get to your actual rank. But most people aren't interested in grinding a game for 8 hours a day for months just to get to their actual rank.
But now you are forgetting what Grubby said as well: Those 4 lemmings are as often on your opponent side and you get easy wins against them even when you don't deserve it. So myth busted
@@YesThisIsCrass It's not ideal, but it's obvious that matching 10 people with similar skill level is harder than 2. Also the problem is not that sbmm is made for 1v1's, it would be exactly like it is if it would have been developed for team games
3 weaks ago I implemented the ELO match making in my sports team. We play a kind of medieval duel and team (battle) fights in a kind of HEMA inspired sport. ELO is awesome to our context. Very happy to see the timing of this video ^^
This may be true if the assessment of your skill is accurate. However, if the game thinks your skill is far above your bracket yet still keeps you in that bracket, it will absolutely put potatoes on your team for hte purpose of averaging the team skill MMR. Even if it is a team slayer and not objective based game, you are still getting a 3v4 game dynamic if ONE person on your team is a feeder vs. their whole team being mediocre yet outnumbering you, ala halo ranked matchmaking. It is a LOT worse when it is an objective based gameplay
then you LOSE a lot more points at the end of the match than everyone else becaues the game thinks your skill level is a lot higher (yet they keep you in that rank and give you potato teammates???) therefore keeping you in that bracket of potatoes. Absolutely NOT myth busted ROFL
In halo infinite, for example, if you have an ancient account like mine (since around inception of the game), for whatever reason, you gain a LOT less "exp" than someone who makes a new account and just dominates in ranked. They can go all the way to diamond or even onyx when their actual skill level is close to a low diamond as opposed to people I play in ranked who are just platinum players who are a LOT better than diamond players who only use their teammates as meat shields and have no concept of teamwork whatsoever and only high FPS performance CPUS. I do not play heroes of the storm or anything but I do remember playing dota 2 but it is now all about how polite you are and if you have an account with a history of BM (like mine) bad teammates, you will never match decent teammates, however before they implimented the hush order, the match making was excellent. Completely unpredictable
I had this with SC2 and CS:GO. I watched a few videos on SC2 starting tactics and the game's MM thought I was some expert. Got crushed ~66% of the time and still kept my rank and kept getting crushed because I knew how to start the match and for some reason that gave ranking points. In CS:GO, I used to be a really good 1.6 player but after a decade of not playing couldn't hit a wall. In CS:GO you had skill based MM. I knew tactics, I knew when to sacrifice but I couldn't hit an enemy to save anyone. Way too high I go while having 1 frag to 2 deaths and I can't get down to my shooting level without intentionally losing all the time by just rushing towards the enemy. So much fun.
You might want to look into why people don't like SBMM in Call of Duty and not just assume that every game works like Warcraft. That system ruins the new games and the whole thing would be fixable without even removing the system.
Strange definition of new; CoD has had SBMM since 2007. You can read about it in Activision's whitepaper on multiplayer game methodology published around that time
The main problem is some games (might) separate MMR and ranks. And depends on how much it's separate. I think LoL had a system like that. When you reached threshold between rating. Say you go form silver to gold, you need to win 2/3 games next. Failing to win would keep you in silver, and still match you vs silver and low gold players. But it does keep track of your skill regardless, and will try and match games around both your skill and rank. And winning the 2/3 games will put you in gold depending on your skill, so can catapult you to mid gold. It did kinda work out with enough games played, but was weird. No idea if it's like that now. Overwatch also had something going on with non-rank skill tracking.
Actually, in League of Legends, you might end up consistently matching vs players in mid-high gold until you actually win the promotion series to rank up or have enough disparity in MMR and rank to skip it (IIRC, you can only skip rank tiers, but not the 5 game promo series to advance in rank). This is something that's happened to me and some friends of mine on occasion (when I actually played ranked in LoL; at least one of those friends still runs into this problem regularly). LoL will also occasionally match teams with significant disparities in MMR/rank, resulting in occasionally going up against a team of much higher ranked players (EG: a team of mid-silver players vs a team of mid-gold to low platinum players, or against a team of mostly bronze players - about full rank of disparity) - this is less common in ranked, but it does still happen occasionally, and matching up is particularly common during those promotion series.
Its even worse in Starcraft, where you have 3 different ratings. You have your rank, your visible mmr, and your invisible mmr. These can be completely disconnected from each other. If you win a bunch of games in a row you will probably not even rank up, you will gain maybe a few hundred visible mmr but your invisible mmr will be boosted thousands of points, matching you against players several ranks higher than you.
Thanks for immediately explaining that this is about team games within the first 30 seconds. I was confused because I always think WC3 and SC2 when I see your face :)
There's a few things I'd bring up that you either ignored or rattled off at the end without fully addressing. - snowball factor of characters/items/etc.; depending on game, these are not all equal. Ex: a genji grandmaster with those 4 diamond will have more carry potential than a lili - sometimes items, roles, etc. are intentionally made unbalanced to incentivize/disincentivize play and influence meta - bots, smurfs, and solo vs duo "boosting" of a duo's wr - strategy, skill floor for various roles/characters/items all vary based on ranked; this means you may be more or less inclined to climb based on whether your rank, role, character caters to your strengths; there are of course outliers at any rank too ex: league - adc baseline is higher skill floor, but is currently stronger than other roles at equal skill once the floor is met - all these factors can vary by patch - somewhat like a mini mmr reset I agree that - given enough playtime - a well designed sbmm will place you at or probably within a tier of your proper rank. But all the above factors can influence to keep someone delayed from achieving that. So players who put limited time into a game and run into various temporary roadblocks can easily be left with inflated/deflated rank - especially given the last point I brought up. This is not meant as some attack on sbmm or to say it's wrong. Just to point out some of the issues that make it far more chaotic and troublesome for matchmakers to do their job well on a consistent basis. While one individual instance or factor may not be much on it's own, they add up via ripple effect. Sort of like zerglings. One isn't a problem. A whole bunch all crashing the same pylon makes for a bad day for protoss.
A few more I forgot to add: ping - some fluctuate more time of day - who is playing drunk/sober at work vs at home (yes, had guy stop in middle of HOTS 5v5 for 10 min cuz boss walked over to check on him)
Great video, I think 1 thing is missing tho : in team games, some games are more "carryable" than others. In LOL you can get fed and 1v3 in a game carrying it to the end. But in the case of balanced/fullteam games like HOTS, you can almost never win a 2v1. Thats also why you can carry easier in LOL or more in FPS than games like HOTS.
There are 1 and 1/2 things that you didn't touch on. For the half: While you did talk about wide matches, and you did bring up how a game with 5 closely skilled players could be favored over a team with a carry and some noobs, what needs to also be heavily considered is that in MOBAs, and other role based hero games, like OW, each role has a different amount of direct impact on a game. Pairing a GM Support with Diamond DPS and a Tank against a full squad of Masters, probably isn't winnable for the GM, as his character's kit doesn't usually have the ability to carry a team. And this is especially true in games where certain roles depend on other roles for help. In Overwatch 2, the Tank role is the strongest role, without a doubt, but is entirely dependent on Supports to enable them to do their job. If they don't get the minimum support that they need for their role to function, then it will also be like the Tank is a noob. As for the 1: This possibly could have been included under the umbrella of engagement based matchmaking, but not really. Blizzard has a patent for Overwatch, and you can look this up, it is 100% legit, that states that newer players may be matched together with veteran players that have generally paid for skins, with the intention of getting the newer players to also buy those skins. This can go as far as pairing an inferior player with a superior player that has paid for skins, so that the superior player can look cool and awesome to the inferior player. And furthermore, there is also a patant that says that once a player buys a skin, they will be rewarded and made to feel good, get a dopamine rush, by the matchmaker attempting to guarantee their games will be wins, when using their new skin, to encourage them to buy more.
Depends a lot on how far in GM, diamond, and master you were talking about, but mostly a GM support can absolutely hard carry a game like that. As long as it's not a Mercy, Ana, or Brig otp, it's not a stretch. You can currently hard carry OW in any role. It's much more a matter of hero pool within a role, like how most low mobility heroes are worthless in the face of proper tank gap. You also quoted the patent slightly wrong. Both things you mentioned are from the same patent. It does not say anything about giving easy games after a purchase. It talks about putting the player into a game where the newly purchased item can be used effectively, heavily implying game play affecting items. From US20160001181A1: "For example, if the in-game item is a weapon (e.g., an accurate and powerful sniper rifle), the second gameplay session may be selected because the weapon is highly effective in the second gameplay session. Doing so may encourage the first player to make subsequent purchases due to the satisfaction of effectively using the in-game item during the second gameplay session." Much of what's in the patent follows schemes used in mobile games mobile games. It's also not Blizzard's patent or ABK's, but specifically Activision's. Considering it's timing, it's largely thought to have been specifically planned for or created during CoD Mobile's development.
Good video! Btw Grubby, these sorts of "meta talks about gaming" videos are my favourite. Ones that discuss gaming and the gaming industry in general. Similar to your video on ranking games by complexity/depth. I find these far more insightful than a deep dive into any particular game.
Here's a league story: League came up with the swarm mode a while ago, and when you first enter the match making, you find a game really fast But after you play for a few games, and you win all of them because it's too easy and no one loses in that mode, your que will go up to 15 20 minutes Why? Because the system is trying to make you lost but it can't find anyone with low win rate because everyone is winning And you know it's not that no one is playing, cuz if you simply log in to another account you can play normally again
Or it's because you're in a higher tier of rating than others, rather than the system failing at trying to find bad players which are plentiful. That's a well known matchmaking issue when it's for something new, and it works on the other accounts because they have lower rating and thus more matches. System isn't trying to make you lose, you just went ahead faster than the rest of the playerbase because you won them all and kept playing.
@@averageds3player699 Yes, some people are just not going to be at higher rankings, especially when it's new and people may not all be grinding the mode.
@@averageds3player699 A pve game mode with Vampire Survivor style gameplay. Yes, some people are bad at it, and not everyone is going to grind it out, resulting in lower amounts of high performing players, especially when it just released. The only ignorance is yours.
Just put same elo players with similar wr in the same game, matching people with 60% and 40% in the same, team just makes the experience worse, make it more competitive, not stupid lol
I'll be honest, I learned to hate the matchmaking systems. And I long for the old times, when you simply had player-maintained lobby with servers. Unfortunately these days, if these exist, they are infested with cheats or gimmick servers and you are forced into matchmaker-made PUGs. In team-based games even normal non-skill based matchmaker drags you towards 48-52% winrate by the virtue of carrying or being carried. My real issue is that you, as the player, lack any actual control. In a way, MM in most games that I know is actually very punishing, especially when you consider the fact that plenty of companies monetise these minor conveniences. You found the team you liked and enjoyed playing with, or had this once in a million match where the team worked like a clockwork? The matchmaker says "F-you, random monkey team it is". A map you love and want to play on it again? MM says "F-you, the boring-ass map you hate it is". Another thing is confirmation bias - even 50% winrate looks like losing more. We have a binary system - win or lose. Our minds are wired to consider a negative result as an anomaly. This means we disregard the wins, we treat them as a norm. Losses become vastly more visible. This becomes incredibly annoying - in team based games, you often feel punished with a loss for no fault of your own. And it tends to create landslide victories or losses. In solo, skill based MM games, you are also often in a state of flux. You do well, your ELO goes up, MM decides to match you higher and you get curb-stomped. You drop and now you curb-stomp the newbies. Actual CLOSE, even matches are extremely rare from what I noticed. Again, it is an equivalent of landslide victories/losses plague in team games. Statistically, you are doing even - but actual individual matches are rarely close or fair. Skill based in team games is another can of worms as well, because what metric do you use? Average is probably the most obvious but also the worst - because it will be punishing good players with monkey teams. Players rated 10, 1, 1, 1, 2 have the same average as 3, 3, 3, 3, 3. So which team will win? Their average is identical but one team has consistent skill level and the other puts everything on shoulders of one player. Team 1 is going to have a frustrating experience - 1-rated guys will have a feeling of being overwhelmed by the enemy and feel inferior to obviously superior ally, while the 10 ranked guy will not be happy with a bunch of dead-weights.
your last point is not that difficult to tackle, they literally have a massive amount of data that they can analyze how intra skill imbalance effect the outcome.
@@snuffeldjuret It is more more difficult than you might expect - I am considering only quality of matches here - it is just rambling is the vacuum - but speed at which the matchmaking works is also a factor. Nobody will care if MM gives you a perfectly fine matches, if you have to wait 10 minutes for it to "cook". And if the match itself is short, people will quickly get annoyed. It gets even more complicated if you need to factor in people's connection quality for that matter.
@@hideshisface1886 I 100% agree with you on your initial last point. A matchmaking system which puts a 10 with 4 1's to "even out" the game is not fair. The 10 should have an equal chance of getting 3s and 1s. The 10 should have a higher win rate, while everyone else has a roughly 40-45 win rate because they are just as likely to be on red team or blue team. If you think about it, a game where the only strong player in the matchmaking pool DOES NOT have a higher than 50% win rate literally means the GAMES are unfair, artificially made more difficult for him to "counter balance" his skill. Imagine if in the Olympics, they added more weight to the fastest skiers to make the results always more "50-50", would that be considered fair?
A match maker should only care about finding players of closest skill level, and NOT the result of the game being 50/50, which is what the last chart in the video showed. A GM should not be artificially given weaker teammates to artificially give the game a 50% win rate. Rather, he should be randomly assigned to one side. He will be more likely to win, naturally, as he is better than the rest of the lobby. However, it is still fair for the other players, as they are just as likely to be on the GM's team as anyone else. Overall it would make the games more fair. "50-50" isn't the same thing as "fair"
Thank you grubby , amazing content, you clarified everything so well. I will be less tilted now if my team goes 0/15 in 5min. I need to carry somehow without getting tilted.😂 I learned alot. Thanks again. Actually those games were winnable, But the thing is my teammates tried new champions for first time. Matchmaking expected them to perform well.
I'm sorry, but a fair match system does not lead to addiction and larger purchases per game. So modern match systems are not made to be fair, but made to make players spends as much time as possible on climbing without leaving the game. Nothing personal, just business and your own ego that does not allow you to leave a toxic relationship with the game.
Oh my god finally someone big said what I have been saying since forever. I remember when I played Warcraft 2 online and not having such a strict ladder was so much fun because you could really feel how much you improving instead of just seeing a meaningless number online.
The dev is another gaslighter who defends the company's practice of infinite hamsterwheel. Nothing of what he said is true and can be tested by anyone who had experience playing in early 2010's.
As a Data Scientist and Programmer here is my take: There are many reasons the whole MMR system is stupid. Note: Not the matchmaker which finds games. The Rating System is what's fkd. 1. it doesn't bias winning, meaning someone who has played another similar game (like DOTA or LOL) with lots of transferable skill still needs to play hundreds of games vs potatoes before the system eventually finds real players for him. What's going to happen? Good players will just leave the game because why would someone good want to play 100 matches to finally play some good games??? 2. It encourages smurfing because it takes soo long to climb you can abuse dozens of players before the system climbs you out. 3. It doesn't give you the opportunity to learn. Let's face it, the idea of playing with equally skilled players sounds good, but that's also the reason why most people learn bad habits they never break. Because they never get punished for their mistakes since they tend to play vs "equally-skilled" players. 4. Climbing in MMR is directly counter to what the system rewards. If you want to get better at the game you need to try new things, push limits and play against better players. If the game punishes all of those by me losing MMR then when am I going to learn. This is why most people in Platinum and Diamond never grow. SOLUTION: 1. Most *INTELLIGENT* mmr systems (like chess or go) use a self-determined ? rating (meaning you say "I think I'm Diamond 2") and it lets you play a few games, if your winrate is
Grubby, I first heard of you when you were doing your dota climb… now I had urges to replay Warcraft 3 and UA-cam starts showing me all your videos hahaha. Thanks for confirming I should replay it. Also, a great video!
The issue I have is that a single MMR rating that is permanently on is extremely detrimental to the gameplay experience. Forcing everyone to play optimally to preserve their rating kills off the posibility for finding the fun in suboptimal play. Most games purport to have a casual mode but the reality is that it's not casual. You still have a rating, it's just not shown to you. It's ranked lite, and it sucks the fun out of games. It also condenses your entire play pattern to a mean value, which is throwing out alot of data. We're not static points on a line. We're lines on a line. And sometimes we're multiple different lines on a line. Another thing is the problematic part of systems based on Elo: a new player is assumed to be of "Median" skill. Which is blatantly false in online games. In tournament context, most players will not go to tournament until they've played in casual (truely casual) settings a fair bit. With online games this is not the case. Being thrown in the deep end with nothing more than a extremely barebones tutorial is just bad for everyone involved. Also, having a matchmaking rating and a unrelated ranking system just doesn't make sense, and the two heavily fight each other. But lastly, the biggest problem is that losing isn't fun, and on top of that you're punished for it. You get less rewards, your number goes down, and it's just all around a negativity spike. In a sense, losing a game is a service you provide. You lose so someone else can win. While being rewarded for losing can quickly lead to people intentionally tanking games, it is a mechanism that has to be investigated nonetheless. Losing has to feel as good as winning. So in conclusion, skill based matchmaking in isolation isn't bad in theory. In practise, it has a lot of pitfalls that seemingly have yet to be solved. We have yet to find a new-player appropriate matchmaking system, it makes every game a high-stakes game when games are often at their best when they're low-stakes, it contradicts having a more formal ranked system, and IMO it's more important to find ways to make losing fun rather than get everyone to a win half the time.
Yes, exactly. Once I ranked up I felt like I could never go back to my worse race/faction/build/etc as it'd undo all my hardwon slightly over 50% grind up. And the whole reason I stopped playing SC2 was every single match felt stressful. Dunno if it was me aging out or the MMR, but I spent years playing tons of CS and SC1, win/lose whatever, games were fun. SC2 multiplayer was stress. Which is fine if you're trying to do like an occasional tourney, but I no longer wanted to log in daily and bust out a few hours of games.
The only reason why it would "force everyone to play optimally" is if you enforce the mentality of "forcing everyone to play optimally"... in a casual mode, where the rank is not visible, then... trying to "preserve your mmr" is irrelevant. "But lastly, the biggest problem is that losing isn't fun, and on top of that you're punished for it. You get less rewards, your number goes down, and it's just all around a negativity spike. In a sense, losing a game is a service you provide. You lose so someone else can win. While being rewarded for losing can quickly lead to people intentionally tanking games, it is a mechanism that has to be investigated nonetheless. Losing has to feel as good as winning. " THIS!!! is a YOU problem... And most people, even if they complain, would rather have skillbased matchmaking than not. Imagine playing... let's say, Age of Empires... Let's imagine they didn't have a skillbased matchmaking system, and you just start a game... you are a mid level player... most games for you will either be a stomp, or you will get stomped... The low level player will almost always be stomped... and the high level player will almost always stomp... that's not fun either... even when playing casually, that stops being fun really quickly... The pitfalls you mention are rather player attitudes that have evolved, rather than intrinsic properties of the matchmaking systems, especially because games having various matchmaking systems are not complained about like that. Chess (which most systems have been based around in one form or another) has had these systems for a long time, and doesn't have that toxic culture itself... and the ranking system is used heavily so you can find interesting chess games. It's not particularly fun playing against someone that is far worse than you, because you can counter their moves without even thinking, in a game about thinking. It's not going to be easy to find ways to make losing fun... since winning is kind of the reason why competitive games are fun. There are party games that have been made that can be fun to lose in, because you end up engaging with the game while you have lost, until the final winner has been declared, but that doesn't make it more "fun" losing... regardless of that, those party games that are wacky and random that makes it less frustrating to lose, those... they have a tendency to be... well... not really played en masse. The Battle Royale genre kind of did something like that. If you lose, you can start again very quickly, and the whole looting part of it is also a fun part, so even if you die, it doesn't feel like wasted time, and due to the random nature, you will inevitably get high in some games, regardless of your skill level. It gives engagement from the feeling of "I almost won!", and since there are so many teams/players, and the likelihood of winning is lower, then losing at the higher levels can be quite engaging. But that doesn't really work for a MOBA, does it? Or a general team shooter? Or a 1v1 fighting game? To make "losing fun" you have to make a significantly different game...
Totally agree with the video! One small thing to clarify though, when you mention flaming teammates for playing off meta picks (specifically in MOBAs), it's usually because players are at their peak elo on a specific role/champ. I do not mind having a long time off meta player on my team, but it does tilt me when my Janna OTP first times their AP Shaco top lane because they watched one too many Pink Ward montages.
"but it does tilt me when..." that's a you problem then. You cannot control what your team is doing, you can only control your own reactions. Game is already hard because X person is playing "bad", you don't want to make it even harder by playing on tilt yourself.
@@Lightn0xWhy put in effort to win when your teammate plays 4fun and soft ints and griefs his way down, ruining games as he tumbles down to his deserved rank? It's not enjoyable to have one guy sacrifice his teammates ability to have fun for his own.
In Crusade Dota 2 most of the off-meta flaming I've seen wasn't warranted. For example I played support Weaver for a while not long ago, despite it being very strong I had issues with my team when I picked it because they didn't understand it. But I think the real issue you're saying is that they're picking a new hero they don't play well. I'd say that's a tangential issue from the meta or which heroes are actually good. It's more about specialization and learning heroes inside the ranked mode when you shouldn't be.
@@famulanrevengeance3044 You missing the point that the game always set you up to lose. Simply for profits and statistic for higher active player, but also introduce stockholm syndrome. The more rich/succesful people actually have higher stockholm syndrome rate, than someone who always chill and feel 'enough'. To be fair, everyone goin to have their sinful pleasure onto something else.
I'd love to have a real ranked system like you're describing. Problem is, I just don't have the trust in game devs these days that they aren't "cooking the books" behind the scenes and manipulating the matchmaking to create a certain experience that they've found retains players the best. Idk how exactly to fix a trust issue, but I'd start by demanding total transparency of how the system works and see where things go.
Cooking the books could definitely happen. I think it's really common in Gacha p2w mobile games, and AAA game devs may take note, when infested with money hungry stockholder lap dogs. I think many devs still rail against that ethic, though
@GrubbyTalks even AAA commonly use scummy tactics like shadow pools or weighted pools to improve matchmaking in certain pools/regions which hurts people individually.
Matchmaking is garbage. Back in my day we just played games for fun. Now the only reason people play games is sweat, meta, ego stroking. We can't just goof off and have fun because everyone is in matchmaking.... ffs It's impossible to just dick around and have fun when you're constantly being nagged at about your lifetime performance...
Seriously, right? This is why I still play on community TF2 servers to this day… I have literally no interest in being “ranked” in an FPS game (or any other type of game, for that matter) I just want to hop online and shoot at some people for fun. I can spend hours just “scrimming” and enjoying the experience without stressing about anything 🤷
I'd love to see more experimentation with PvP multiplayer that doesn't care about skill. Most games are made to work around this model, but I'd like to see some that are more experimental in nature. Imagine if Starcraft didn't care who won, you could add a lot crazier systems in because strict balance is less important. Then you could tie meta progression to games played instead of ladder advancement through winrate, and suddenly you have a system where you can introduce new unlocks that everyone gets to partake in.
EOMM is definitely being used in AAA games, one example being TrueSkill 2.0 which Microsoft uses for halo infinite. TS2 can predict which team will win a given match with 68% accuracy which if you think about for 3 seconds you’ll recognize why that’s a bad thing. If two teams are of equal skill then you should not be able to predict who will win because they should both have a 50% win rate. I recommend watching Favyn’s video on sbmm because it coincides with a lot of the ideas in this video but from a fps perspective.
But the hability to predict match outcome doesn't mean it tries to screw with players, it just means it is good at evaluating their respective skill level. There's nothing that garantees there will always be enough players available to create perfectly balanced matches.
@@theslay66 That's the thing about Engagement Optimized Matchmaking - it doesn't care if the match is perfectly balanced, or even close to fair. It just cares what the chances of you quitting after the match is over are. The longer it keeps you playing, the better it believes to have done its job, so if you're a player who tends to keep playing on tilt when presented with a 10 game loss streak, the algorithm is actually going to _favor_ putting you in that position. Anything to get you to play for longer.
@@scrapbotcommander And like explained in the video, this argument doesn't hold. If you have fun, you play. If you don't, you quit. Maximizing engagement is just about maximizing fun for the players, and having balanced matches is generally what is overall better for that. Nobody plays more when they are presented with a long loss streak. And even if there are, they are outliers, not your average player. There's no sense in building your system around the behavior of a minority. As there is no sense building a system with purposefully unbalanced matches to screw with the players. There's no need for that. Randomness of available player, and a good ranking system, is all you need. This kind of streaks will naturally emerge from the system. Random is random, streaks will happen. Seriously, most people complaining about how the matchmaker makes them lose on purpose are just biased. We tend to attribute success on ourselves, and failures to external causes, that's natural. But why the system would target any player specifically ? It must work with all players simultanously, while working with a limited pool of available players at any time. Making a system that try to decide in advance who should win and who should lose is just senseless, and far more difficult than simply trying to make balanced matches. The rest is just speculation, complotism and persecution syndrom.
Eomm straight up rig the games using aim assist aim resist, the strong players all have aim resist, except when they are meant to win the game for the noobs on their team, the noobs all have aim assist, the lobbies are rigged to generate dopamine in the noobs, get them addicted to being better than they are, making them believe they are good, and thus spending money, rewarding the company for giving them a fantastic gaming experience where they were made to be the hero. It's somewhat of a twisted business model but it is the best business model you can have in a shooter environment, it's what makes the most money, it preys on the weak. So there is two ladders juxtaposed, one real, that is hidden, one manipulated in the open where they make people who spends climb and the people who don't stall, with the caveat that depending on who you are they may model their abuse of you based on you. Other way to look at it, imagine you have a garden, the weeds are the good players, they perform well in the environment, and the noobs are your fruits and veggies that needs nurturing but produce food. So you use the weeds as biomass to fuel your production, perfectly logical. It's perfectly fine in nature, not so much in a society. Tl:dr: eomm is a con.
@@Deucely I mean, the matchmaking system can't control the presence or absence of aim assist, it just pairs groups of players together. That isn't the problem with EOMM, the problem is that it will sometimes match you with an opponent of higher or lower skill than you, like matching a platinum player with a diamond player. In the case of team games, that would appear as one team's average rank being somewhat higher than the other team's average rank. Not enough to be blatant, but enough to skew the results in favor of one team, by maybe a 60%/40% chance. You should still absolutely take responsibility for your own actions as a player. It also won't do it constantly, @theslay66 is correct that this would make most players quit, and my example was admittedly absurd. But the problem is that EOMM may sometimes decide that skewing a match is the best thing to do based on how much it values the outcome of a game for a specific player. And yes, unlike SBMM, EOMM does actually value using the predicted outcome as a metric for optimizing player engagement. If you can predict the chances of a match's outcome, you can control how often a person will win or lose. You can break up lose streaks by handing out pity matches with slightly lower MMR opponents at key timings designed to keep you from quitting in frustration, then throw in a close game after feeding an easy win so you still feel challenged as a player, then throw in a clear losing scenario so someone else can have their pity match as well once the system thinks you can tolerate it without calling it for the day. That does not mean it is necessarily a fun or fair method to use. While fun always creates engagement, it is not the _only_ source of engagement. (I would not consider getting flamed for 30 minutes straight in a losing match of LoL to be fun, for instance.) And EOMM, sadly, does not discriminate: To it, engagement is engagement, no matter what that looks like in practice. That, however, also _doesn't_ mean the algorithm has some personal vendetta against you specifically, or that it will always disrupt a streak. That's not true either, it's just trying to maximize the average engagement among all players. And sometimes that just ends up involving unfair matches... the only time it doesn't, it matches SBMM anyways, and in my opinion SBMM seems to be to just be a better system overall, at least for 1v1.
it's really hard to balance any game, so that you both have a clear progression of strength that you feel stronger all the time, and also aren't bored by the lack of challenge once you've become strong
12:00 - People that defend 50% completely missing the point every time. It's not about "we can't get to where we belong". it's about: "Why do I need to play 300 games instead of 100 to get where I belong?". LoL players would I understand what I am talking about. For everyone else, imagine if Happy/Sok/Lyn/Hawk would've lost 3 games out 10 when climbing through adept/diamond leagues towards the grandmaster.
I was stuck on a low elo on leage recently. well am but not played lately. Either you completely stomp your lane win and go to the next game and get a free lose and get right back to where you were. Or you completely stomp your lane and still lose because too many of your team mates are completely uncarryable. even if the other players get fed they still manage to throw. there comes a point where it is so rigged it stops being fun. why do you need to grind through 100s of games like that just to climb one division closer to where you were.
@@elvis4868Smurf is basically a skilled player that plays on newb matchmaking *Obviously the skilled player will win evertime against someone that's entirely new*
@@lightlayagajoie5739 There is a solution, which is to be less secretive about mmr and perceived skill. Tell the people playing who's getting matched up and who's getting matched down. The expectation of being matched with equals breeds resentment and toxicity when you get people way below your skill. It also allows the game to be more transparent and say that you get a big mmr boost for pulling through and winning this game despite the odds, while you only get a small hit if you lose. Vice versa as well. Sure, those games will still suck, but being clear about that helps players rationalize wins/losses rather than bemoaning "elo hell"
@@antarath517 the problem is not only lack of transparency but also that some games are clearly intended to be predetermined to be a win or a lose. because somehow that makes it more addictive I think. If I win one game I don't need to be tested if I can still win against 2 smurfs and 3 useless allies. yes sure I'm good but I'm not that good, or whatever. This is just another game you have to sit out before you can continue to climb.
The problem with MMR that you have ignored is that you dont feel progress. The only thing you have to go on is the rank up because the matches are overtuned. The sensation of being good cannot be felt because your opponent is always too close. You never get matches against 'your old self' to showcase how far you have come, you are constantly locked in cuthroat matches and can never relax and try new things.
In 1v1 it's nonsense, but in team based games there is a nugget of truth to it. Not in that it matches you against harder opponents - that makes sense. The problem is when the matchmaking does things like put you in a group where people main the same role (e.g. 3-5x jungle mains) so that people on your team are forced to off-role, just because the people in your team are on win streaks. While the other team is populated by an even spread of main-rolers (e.g. top/mid/bottom/jungle), thus inflating their chance of victory. I'll watch and see if you're talking about the same thing now : D
That is a real problem, but it's not really related to the win streak idea. Matchmaking rating in team games is just noisier, because there's so many things that are outside of your control during a game with 10 people compared to 2 people. Over a theoretical infinite game sample, you should still arrive at your true MMR eventually. But over a realistic sample of 50 games or 100 games, you could easily win or lose an extra 10 games due to this sort of what-random-teammates-did-I-get-today thing, which makes it much harder for the matchmaker algorithm to understand what your MMR ought to be. Role selection adds another layer of complexity on top of that: For example in Starcraft, it's very reasonable to imagine your "true" MMR with Terran and with Protoss may not be the same, but at least there the matchmaker knows which race you chose to queue with and can adjust appropriately. In DotA, maybe your "true" MMR as main-carry is significantly different than with a support hero, or on a finer detail maybe there's only 2 specific heros out of 100-however-many that you have learned to play efficiently, and the matchmaker can't really know in advance which hero you will select this game, or whether it will mesh will with the limited hero set of the rest of your team. Arguably having a wide selection of potential heros is itself a skill that makes you better, but all of this adds noise to the matchmaker.
@@Kerostasis I don't think you quite understood. Imagine if in SC2 you had to play 3v3, and each player on each team must play a different race. Then one team gets a terran main, a protoss main, and a zerg main. While the other team gets three zerg mains and two of them have to pick protoss/terran. Effectively basically deciding the victors of the game before it even started. In a sane world, each team would get two zerg players so one player on each team had to offrace.
I mean, if you're going into a moba solo queue (without role-based queuing) and you can only play one role well...that's a skill issue. Your skillset is limited, so you're going have a rough time when a change of plans is called for.
@@thekingofcheese9005 You don't get it. You select your role when you queue. But it puts main rolers of diverse roles on one team, while the other team gets stacked by redundant main rolers. We're not talking about having to offrole, we're talking about one team having to offrole massively more than the other team (and the matchmaker knows the main role of every player).
Finished watching the video now. Grubby didn't really touch on what I was talking about. But he got into the periphery regarding team games with the 1 champ + 4 mid rankers vs 5 high rankers. To add to that, there's more complexity than you'd think. For example, in this matchup the champion player is obviously intended to hard carry, that's the only way it can be fair. But if the champion is forced to pick his hero before his direct opponent, it means he can be hardcounter-picked which makes that extremely difficult when you can't rely on your teammates to cover for you. Stuff like that.
As usual, a reasonable take from the most level-headed guy around. Looking at the history of matchmaking, I love the "psychological arms race against ego damage" the devs have been at for decades. That being said, the difficulties with team matchmaking are the second biggest reason (after game length) why I left MOBAs and got into RTS and later into fighting games. It gets easier to see rank for what it is, and a combination of ranked and casual games (with looser MMR matching) has been the best way for me to feel the improvement, and enjoy the journey.
So essentially the result of a system that has matchmaking (mmr,elo) is that everyone will have 50% win rate, but they will be assigned a different badge of honor (silver, gold, etc) and the result of a system that has no matchmaking is that your badge of honor will be your win rate - lower skilled players will have lower win rate, and the best - the highest. Given that overview which do you like more? All your games are a coin flip but you get a different colored thing in game, or winning/losing proportional to your skill? I would take the second. Being one of the best players, but still getting 50% win rate is not something I would consider fun. Why did I become better after so much hard work if I am not going to win more? And losing when you are the worst should be something people understand as normal and natural. Something to overcome, not to avoid or mask with a 50% win rate. If I perform badly I want to lose, If I perform good I want to win. And I am talking about actually winning games, not winning more mmr points. I have played a game with no rating and it was damn fun progressing from the bottom to someone that can kill almost everyone. As opposed to a different game with mmr that always feels medium levels of rewarding no matter how much you improve. There are other considerations: if you play at your skill level you MUST always use the best performing strategies/other best choices or you risk performing badly. It ruins fun even more by making things stale. Not to mention that every game is the hardest. There are no breaks. I don't want to do 100% effort 100% of the time. It's a video game, not life or death. Some times I want to chill with something goofy. The most fun I had is when I was a good player, but used shitty, weak, fun strategies against worse players without sacrificing winning. Do that when you are in an mmr game at your skill level and you will have a bad time. Don't tell me that is OK. It ruins diversity and fun. But there is something even worse than this topic, it is when a game is a team game (like league or dota). That is truly something that would ruin your fun. You know if you played. I avoid games like that now.
Man I feel this so hard. Moving through the sludge of modern mmr systems is slow and unrewarding. You always feel like you're getting nowhere fast even if you know logically you are in fact getting better. You're right too - that winning more is literally what we're trying to do when we improve at something competitive. When you rob people of that catharsis is when they start getting frustrated and quitting, blaming "elo hell" and other things, with most people not even realizing what's happening or that they are actually improving. In reality it's a system that feels stagnant and is actively causing the frustration. If you are scared of losing and need to have your hand held and maintain a 50% win rate no matter what to stay engaged with the game then i'm sorry but you are not a competitive person at heart. Unfortunately capitalism could never allow for such an unexploited demographic and leave potential customers on the table however.
@@coreyrachar9694 i mean... you want to have a really high win rate that comes from smashing noobs and people you KNOW youre better than, and you claim to be competitive? thats just jerkin yourself off
@coreyrachar9694 tf? Poor players in a sbmm would eventually lose at below 50% winrate until they are matched against other equally poor players, then if they improve they will win more and climb mmr until they reach a higher skill lvl whats the problem with that? If your motivation to improve is to smurf on worse players then thats just incredibly selfish, i play badminton and when much better players play agaisnt weaker oppos they know to chill cuz otherwise its just bullying their opponent
If performing at your best to have a 50% chance of winning isnt enjoyable to you then yes sbmm is not for you, but some people to enjoy the challenge and strive to get better still as they can find enjoyment in improving game to game to climb to a higher rank, you want to be allowed to destroy people worse than you which would actually kill the game as the worse players wont even be able to have a game experience to learn and improve
THANK YOU GRUBBY! I had a friend who complained so much about 50% winrate being BS and as someone who has studied statistics I found his arguments infuriating. Especially for 1v1 games, which we was playing, when you drill down into their arguments it is all about 'oh I think I am good and I want to win more.' It is never from the perspective that finding challenging opponents is actually a virtue of a good system and the quickest way to improve. No, it's just 'I want to feel good and curb stomp noobs' deep down. There's a quote on the wall of my dojo that says, "Your opponent is your greatest teacher. Your ego is your greatest enemy." People would hate it but that should be on the loading screen of all competitive games.
That last example of grandmaster with diamond teammates is exactly why I stopped playing HOTS and Overwatch. I put in a bunch of time and learning only to get punished for slowly building up my skills. It's so exhausting and frustrating knowing that if you don't HARD CARRY, you WILL lose badly and it just gets worse over time because no one wants to be a perfect robot tryharding every single game just to have a chance at winning.
I mean, if you were only put in games with other people as good as you, you would still have to tryhard every game to have a chance at winning, because you have to tryhard to get 50% winrate at the highest level. Thats how competitive multiplayer games work. If you dont want that, just intentionally derank to diamond and tryhard when you feel like it and throw games to derank back if you win too many times. That is called smurfing.
>ranked >waaa waaa why do I have to tryhard I don't know man, geez, no clue whatsoever why people care about rank and actually try to win games and won't let you rank up for free.
Once I played like 200 games of Hots over a week or 2. During that time I became a very good player, as with that many games you see the patterns develop quickly. You are most likely to lose around minute 17 when one of your teammates tries to solo a camp, doesnt watch the map and get ganked by the enemy team, and you cant recover the momentum. I would argue the fact that I understand this and my teammates dont means that the matchmaking is flawed as I am paired with teammates who are less skillful. Anyway my rank barely moved over those games and played my heart out tanking for teammates....still doesnt matter.
@@CommanderReplay its a team game. Unless you are extremely more skilled than the opponents at your rank, you won't be able to significantly carry randoms. You would need to find a team with 4 friends who are at the same level of cognitive abilities.
People that hate SBMM are all people that are slightly above average and were used to dominate in lan parties, but the moment they go online with a big player pool, they realize they're not that good.
Before Apex legends had ranked mode I managed to win 15 games straight as a solo player (20 teams with 3 players in each game). The fun stopped when I played ranked, and I quit shortly there after.
You were probably a top 90% ELO player. The problem is, SBMM extends ELO hell to about the top 95% ELO players. You got placed in top 85% to 95% ELO lobbies after SBMM was implemented. This video fails to mention the extension of ELO hell that is caused by SBMM, which is why SBMM can work in theory but feel bad in practice.
37:45 I remember splitgate doing this. I was mid diamond and ended up playing against 3 people of the highest ranks in the game who had developer tags next to their names and I reconized them from the dev team. They also had a mid diamond on their team and my team had top 100 players on it too. I'd also get paired as a mid diamond with a team of silvers to fight a team of golds with me being the outlier. So this ends up happening relatively frequently. I had to carry lots of gems because I was the highest rank in my team and it was a glorified 1v1 between 2 diamond ranks and a bunch of lower ranked players.
Well, it's easier in FPS games, especially in a game like Quake where one person can literally carry 1v10 if he's so much better. Can't really do that in Counter Strike for example, just because of how the match, movement, weapons and maps are designed... Games like dota and lol are especially annoying because of how you can "feed" the enemy. You choose an "isolated" lane. Consider a scenario where, let's say you're the worst in your team and you go into a lane vs the enemy team best player. Your best player will get matched with someone "mediocore" meanwhile in their lane. You're probably going to make the game so much harder for your team because of how easy your enemy will snowball, enabling his rotation and plays through "experience" and "gold" gained in your lane. On average this scenario doesn't matter much, but it will happen pretty often, the opposite scenario will happen too, where you dominate your "opponent" and then lose the game anyway, and we're all ready to blame the rest.
@@grimonce So basically you are arguing that you can't solve bad game design with match making? Overwatch was supposed to be an FPS but the hero design totally ruined it, even though that was supposed to be the main selling point. CS seems a bit more feasible, I have barely played that game but brought back a round from 1v4 once. LoL in particular just really sounds like a snowball fest, idk why you'd want to play that game. But I can really only compare OW and Quake from the list... Also my experience in CS was very cheater heavy.
This was the most "You didn't actually watch the video" video of all time based on peoples comments saying they agree that SBMM is bad. Never seen so many people get fooled
My boy grubby did a great video! To summarize : everyone is silver and mmr is a rat race regardless of game. the points and rewards system work! I play Marvel Snap(card game) with challenges on parallel with the ladder and it is very enjoyable. For Sweaty boys , open tournaments could be a very challenging and rewarding experience than the ladder.
Thanks for making this video. All this video did was confirmed that the changes made in Wild Rift was the changes that the community. was asking for when it came to rank. (My TL:DR for the day. The bottom doesn't truly matter.) Note: when I say people or community I mean all of North America. Not Japan, not Europe. Strictly NA. I remember that points were a thing Diamond and up. The first season I've experienced this was in fact the last season for me. I was stuck in low emerald for the second season in a row. I've played with a friend that was low platinum. 3 straight games in a platinum room that I got wins at, and the system decided to put me in diamond for the first time. Cool. At that time, many people were either climbing or dropping with an average of +10/-15 rating. Riot decides to fix this mid season before the dominoes fell into place later on. Points were replaced with marks, and Riot made it easier to play with friends in rank regardless of rank. Autofill was already a thing during pc helping to guarantee faster matchmaking. It's still being strongly adapted to this day. So unless you're in legendary queue you now have a system that rewards strictly based off of wins and losses with the performances being measured towards getting your next shield mark protection. Because autofill is there the community now expects you to know how to play all roles really well, and if you can't people are just going to surrender until they're able to find a team that can play perfectly. If the other side is being totally dominated they're going to surrender since they're on the other side of the table. The community enforces this, because most is fighting to be a 1v9 carry for the team. Whether you want to admit this or not these current systems only benefit the casual playerbase. They also benefit the lazy playerbase: those that don't to take responsibility and would rather buffed and nerf everything to their liking just so that can finally have a shot at winning. This game is for them. After all, Riot themselves wanted the entire playerbase to be at a diamond level. So now we're at a point that since the amount of players at the lower ranks are way too small everyone between iron and gold are now considered one group to make matchmaking easier. That's what happens when it comes to boosting. While I was on the fence about whether the forced 50% was real this has basically cleared my suspicions. Thank you. The develop.ers are at fault for making the system that some people hate. At the same time, some have been asking for this. Do NOT put all the blame on the developers
@@zerg0s correct, you can climb up pretty quickly in a stack, and then play stably in this rank even in solo. if you're in the broader part of a bell curve, you can be placed whereever and play pretty comfortable.
@@zerg0s If you got a skills and you got a team, you can easily cheat a system and get 55-70% up to 80%. This system sucks but devs are loving it because it forces your engagement. More engagement = more $$$.
@@MrSam1804 you climb faster up until the natural elo of the worst member of your duo/group. Then you climb slower, because that player just became an anchor for the others. And at that point, you are still a silver player. You're just playing in gold because you got carried and you'll slowly sink back down over time. Elo kinda sucks because it needs 200+ games to place you (as the video points out), so if you play less games than that per season you're just kinda adrift.
An excellent analysis and explanation of how mmr systems actually work over extended periods of time. While listening to you I caught my past self getting emotional when losing a match after a pride-filled winstreak because of griefing / trolling / whatever. It feels bad, but like you said on the other hand, it isn't nearly as annoying when such a case happens on the enemy team. I feel like I've gained something by watching this video. Thank you!
I'm at a 22% winrate, so I guess I haven't reached my true MMR. God have mercy on me.
As someone who hit 10 mmr in dota (the lowest rating possible, i was learning midlane and was TERRIBLE at it) it was honestly a great experience seeing myself improve and rank up again,, last week i hit 3k mmr, still playing midlane and having fun! So dont let the numbers discourage you, 💪
@@dj_vanx there is no way you are 10 mmr and can form a coherent sentence
@@Stanky_Footthe difference between 1 mmr and up to 3500 mmr is the amount of pleasure you'll get during the game. The closer you get to average mmr (the one every goddamn player gets at the start), the worse the games will be. 1 and 5500 are equally close to that rating, meaning the quality of the games goes up.
And I say it as a player who's been at every one of those ratings at different points in time
@@dj_vanx as the saying goes, when you're at the bottom the only way left is UP
@@Stanky_Foot The crazy part is if you've seen the dialogue that high ranked players provide to you when you play with them is like a incoherent priest speaking in tongues.
I think the points by Activision about the lowest-MMR players eventually quitting are good but perhaps not emphasized enough. As a game gets older this happens more and more so that the average skill of the entire playerbase goes up. This causes the initial "skill wall" that new players need to overcome to get higher and higher, causing new players to "bounce off" and quit immediately unless they have an above-average level of talent and determination to learn despite all the losses. Without an influx of new players, those older players at the bottom of the rankings will have a harder and harder time finding games against equally skilled opponents, and they too are at risk of giving up and quitting. As the lowest skilled players quit, the "losing portion" at the bottom of the ladder will continue to move up, claiming more and more players over time. In the end, the playerbase can shrink to a tiny shadow of its former glory and be composed of only the oldest and most skilled players
This is a legitimate "problem". I find starting AoM multiplayer hard, because I'm going into a 20 year old game and anyone playing already knows all the basics. So I have to lose at minimum 5-10x in a well-working system to reach fellow newbies. Not a welcoming experience
(Actually reminds of my first basketball practice sesh, I ragequit after 5 minutes)
Hey, thats a cool way to describe fighting games
i've read that this is why Sakurai has effectively disowned the competitive Smash Bros community. He wants to keep the game friendly to casual players.
@@thefance4708 It makes sense. Competitive scenes are great for elite competitive players and their fans on twitch/youtube but I don't think they do much for sales
can be fixed by regular "events" for newbies to gather existing ones, or get new players.
The last point you made is a very important one! Most MMR-Systems are still heavily based on the Elo-system, wich was never developed with team based games in mind.
There can also never be a universal system for team-rating, as the impact of one very good or very bad player can have varies vastly from game to game. So it's inevitable that good matchmaking gets harder and harder the bigger the team sizes get.
Its also extremely difficult to quantify what someone contributes to a team instead of just awarding points based on win/loss or score at end time.
For example in counter strike, in matchmaking you gain or lose elo based on round win difference in the game and the avg elo difference between the 2 teams. So if you beat enemies around your own skill level 13 to 0 you get more points than a close win.
The issue with this is, what if you play a game where you massively carry your team, get 30+ kills, (which is more than 1 per round, the average is like 0.67 kills per round) but the game goes long and you win 13-11, you barely gain any elo because your team was not very good. You will only gain a few points for massively overpoerforming relative to what you should be doing at your skill level, your teammates who played badly and died more than they got kills also get points, while they should be losing elo.
It also doesnt take into account how many assists you get, how many times you save teammates from dying, how many enemies you flash, etc. Your kill-death ratio doesnt even factor in.
Some online services like ESEA tried to improve this by coming up with a system that measures player contributions to a round for ELO gain. The result of this is that everyone wants to play the best spots every round, never pushes first on the attacking side, baits their teammates to get easy kills, and throws useless flashes at enemies without pushing just to make a number go up.
Its literally impossible to devise a system that isnt unfair or open to abuse in a 5v5 game.
@@TheSuperappelflap Very true and it is something I have mention before on this discussion. Its actually impossible to quantify player skill, just too many variables. 1vs1 You can just look at who wins, but in a team game, if you were to determine each players contribution to the team it would be impossible, especially in a MOBA for example where you have different roles like support or carry.
@@TheSuperappelflap the issue here is seeing MMR as a reward. You won a hard match and feel like you need to be "rewarded" with more points than an easier stomp.
MMR's goal is simply to get you to 50% winrate so the matches are balanced. More balance = more fun for everyone.
While stomping players feels fun, it's not fun for the players getting stomped. We can't have a vampiric system where the only fun you can get from a game comes at the expense of others.
That's the one nice thing about 1v1 games. They can't really sabotage you if you're good enough.
If you're not good enough to carry 4 chimps in league/dota then you don't deserve to rank up
What most people are missing is that in team games raw mechnical skills are only part of the story. Social skills can matter even more.
I started climbing like crazy in Dota just by coordinating basic plays, redirecting salty players attention onto enemies and I would join with him to throw smack texts on then and do some pep talk when someone wanted to surrender a game. Mechanically I was in same position as before. I climbed 3 full ranks in Dota 2 with just that.
Overall it's crazy how huge difference makes when one player just rallies others and as I often played as initiator saying in advance "stunning X" lead to massively different outcomes in team fights.
@andrzejnadgirl2029 Social skills can even matter in FFA based games like risk too.
It's not enough to be able to recognize when kills are profitable or how to attack fast amd beat the timer.
Negotiating with other players, not making enemies with everyone because having a good reputation at the table with randoms, being able to recognize and take advantage of lower skilled opponents by intentionally not taking territory so they attack someone else etc
Hearthstone and other random based players would disagree with you on that. Some games might manipulate your chances to make "perfect" play so your win rate stays about 50%.
@@andrzejnadgirl2029 true, and its extremely hard to keep attention on your and other people's game at the same time, if you can do that, you're naturally gonna get higher ranks
The biggest complaint about CODs SBMM is that theres no visible rank or MMR. So your reward for playing better is just harder matches and no way to discern if you're actually getting better.
That's my only problem with it. Without showing the MMR it's just player manipulation.
Oh, so the game would suddenly be fun if you could see an arbitrary emblem next to your name, totally makes sense. Yeah man, good games are only good and fun if you can see a meaningless medal
@@RequiemOfSolo Well it would be an improvement for sure. The sense of progress it gives can be significant.
@@RequiemOfSolo Sense of progression is important for every game, this is a smooth brain take.
@@tehDmez your progression comes from you. Your own actual progression. If you value some psychological trick, superficial progression system over that, then you are a sheep human
I think the argument for dedicated servers vs. matchmaking is always something to consider. Consider how many old FPS games had individual servers that had their own communities which meant you had fun servers, competitive, and casual ones always go in hand in hand, meaning you had to look for them, but you got the satisfaction of playing the game how you wanted. This can be considered in WC3 by how custom games were their own thing, then there was competitive, but there were also noobs or people that played the base game for fun in the custom maps section to play their own way (such as making self proclaimed rules of "no rush"). Many newer games instead opt for just a matchmaking ranking system for online, meaning that all these people are mashed together and are more likely to lash out because they are trading the convenience of having to search for a game with having that freedom be taken away and feeling like a robot is "forcing them to lose." I also think for any game such as DotA 2, you always feel so much worse for losing for team based affairs, maybe someone running into the enemy team too much, etc. while a fighting game or any 1v1 has less weight to losses because it's based entirely around how you play and who you play as. This isn't even considering sometimes lower MMR is filled with Smurf accounts or people wanting to use more cheese specific builds in an attempt to climb ladder like how early SC2 had many claims Bronze League players would stick to Zergling rushes.
Survivor bias. The people who managed to find communities at their skill levels stuck around. All the others bounced off.
From my experience, the more your team is winning/stomping the other team, the less communication and talking happens. The worse your team is losing/getting stomped by the other team, the more venting and pointing fingers takes place.
I found if there is more communication before then you tend to win more... be relaxed and destroy others. It's an unfocused flow.
The golden combination for rage venting is (losing + lack of information). This causes people to vehemently attack others specifically so they can assert that they themselves are not the problem. When it's already obvious who the problem is, and that some players on the losing team are still playing well, then there's less of a need to rage. This is why Overwatch was such a salty and ridiculously designed pit of nonsense. It deliberately hid match statistics in an attempt to coddle poor performers and instead created even more rage as people defended themselves in a game that wouldn't tell their own teammates they were currently playing decently.
I think there's a "communication curve". Bad players "communicate" a lot via flaming. Middle mmr players basically dont communicate. And the top players are ultra communicative.
@@ProzacStylings Problem with Overwatch was not hiding match statistics, because match statistics don't really show how well someone is performing. Damage/Kills/Assists/Healing is not directly proportional to how well you are doing. Damage is important, and some classes end up doing more than others. Someone who is standing ready to take out a specific target and succeeding every time, with little damage, high kill rate might be doing just as well as the tank that is in the right position at all times, with zero kills and nearly no damage.
Lack of information has in general shown to be better for less toxicity, because it is harder for someone to just "blame a guy". Look at League of Legends, people blame the support that is 0/5/18, while... that can be pretty good stats for a support, especially if the deaths lead to winning fights.
Same for the engager that might have almost no kills, tons of deaths, but have ensured most fights are a win.
It's harder to just "blame someone" if there aren't any statistics to use.
The counter-argument is that someone that might be toxic can see their own terrible performance, but often, you can have a toxic person blaming others for their terrible performance, because it's not their fault regardless. Statistics does not protect from self-delusion.
If you take a notice of the games that tends to be the most toxic, it is the games that are team games, where you are highly reliable on your teammates, as well as games with a high population of kids, as well as games with very popular streamers that are toxic.
Having or not having statistics tend to be a relatively minor variable in this multi-variable question.
@@SioxerNikita Damage/kills/healing absolutely does tell how good someone is doing. Your attempted counterpoint of a Tank not being represented is solved by adding a Damage Blocked stat.
The entire rest of your argument is nonsense based on that incorrect assertion.
People who cannot prove they are good at the game by pointing to the stats screen will prove they are good at the game by explaining how their teammates suck. Period. That is why OW is a flame fest.
People hating forced 50% MM when the real problem is anonymous team games. Imagine ANY real life sports that randomly matched you with teammates to compete at a high level. It would be a nightmare just like it is in online games.
i dont think so , cause if you poke a basketball in the ghetto (because you couldn't hit the basket), you'll probably never be able to play any sports again, and you'll only eat from a straw,
I would love to play Dota if they gave me the addresses of my teammates after the games.
Except it wasn’t, that’s literally how it used to work and it was great, at first you get destroyed every game then after a month you win 50% of your games then after like 3 months you have the skill and map knowledge to win 90% without even really trying and you feel like a God who over came a real obstacle. Then the cycle repeats when a new game comes out.
@@AntiTheBirdThis was Tekken but now Tekken 8 is essentially doing the 50% bs
@KyeCreates you literally can't force a win rate in a fighting game lmao you're just a scrub i have a 60% win rate in t8 and sf6
@@TannerLindberg Surely with a good enough skill based match making algorithm given enough time your win rate would even out to 50% if that was the developer intention. Why would it be impossible to force 50% in fighting games? Not saying that the match making system in those games is forced 50 just interested in why you think it’s impossible.
His off-meta commentary didnt totally land for me. I played DotA 2 for a long time and got to a 'good' mmr (5k), thats the 'very high' bracket which is top 10 or 15% I cant remember. I started to go off meta because its more fun and my winrate did drop, I went down to 4k.
Here is where I dont agree with Grubby, it made the game worse. Even while being 1k mmr higher than my team they would flame non-stop or even throw games out of protest for my non-meta picks, ruining very winnable games. It also lowers the quality of the game because people have a higher than normal likelyhood to throw or tilt.
It is NOT as simple as just lowering your rating to the point where your skills equalize the lost efficacy of off meta picks/strats.
Well then you will drop to a point where your skill can compensate for people on your team griefing because of said off meta pick
@@elvis4868 that isnt good. I dont want to play a game with a perfectly balanced win % if it means I have to be significantly better than the rest of my team to compensate for them actively trying to ruin the game. How is that any kind of good?
@@josephjoebrown11 I was in the same boat and I had reached 6k around 2015 -16. It just became impossible to climb after a while as me being at a 55% winrate, 70% wr with a specific hero, got me matched with people that we obviously the bottom of the barrel as far as mannerisms mattered. I am talking about people that ran mid exclusivelly on the first time you pinged / talked to them. I was being matched with various skillsets, wildly varying in my experiences but mostly just bad overall enjoyment.
I just play pubs now and never looked back.
@@josephjoebrown11 Well, have you considered you're the one actively trying to ruin their game
@@zym6687 Im not. Didnt take much consideration to figure that one out.
The real issue is that in a lot of games you can queue solo with and against duos and trios. And games usually try to balance it by getting both teams having a duo or a trio if the other does. So in effect solo players are at the mercy of the level of communication in their duo-trio vs the enemy duo-trio.
What imo also needs to be pinpointed is that sometimes the biggest problem is that the multiplier on the mmr gain and the difference of mmr between two ranks/the size of the mmr scale is so disproprotional it requires an absurd amount of games to actually reach your true elo - and many people don't even play that many in one season. This can completly trash yout the entire system. People all over the place for long periods and somewhere in the middle there's a mix of most of the skill levels which creates "elo hell"
What Grubby said is true only if we assume most or all people play enough to get to their true elo...many games reset twice or once a year and require 200+ games to get you there.
Yeah i feel like this doesn't get talked about enough - I played LoL for many seasons but never enough to plateau, some seasons 100+ games which should be more than enough to place me. Hovering at 60-70 wr for that long just means I was game ruining for the people placed at the ranks I was passing through - basically forced smurfing. This was basically impossible to escape because I just didn't have the time to play as much as the system asked.
Marvel Snap actually had no multiplier. Because of the "bet rank points on the result of the game" mechanic it had to make wins and losses give a fixed number of points based on bets taken, which also meant that your MMR didn't affect rank gain/loss rate at all. Because of SBMM this also meant that if you played a lot of games your rank point gain/loss should tend to 0 over time. How did they reconcile these things? They didn't! Easily the worst ladder system I've ever seen.
It took me literally 7000 hours to get to the highest rank in CS:GO. There are several ranks in that game where you can get stuck in elo hell where you will be constantly paired with griefers and trolls and lots of cheaters. I skipped most of them relatively easy but the last one was very hard to get through. I actually only got out of it because Valve tweaked the system that rates your performance and I retroactively got boosted up several ranks to the highest one because I was massively overperforming for literal years at that lower rank.
Then CS2 came out, which is literally just an engine and graphics update, and for some reason Valve didnt port over everyone's ranks and we all had to start over again in silver. I have now played for a few hundred hours and only made it back up to Nova 3 which is about the middle rank.
Because of being forced to play in silver league, I got massively reported by noobs and now again get cheaters, griefers and trolls every game.
Yeah, this video was not done very well or enough research was done.
The fact that its possible to smurf at all removes all points in favour of "skill based matchmaking" when games get too hard, you just start again and own people over again. Defeating the point of skill based matchmaking entirely.
If it also resets for everyone and it takes too long to get back to the point you were meant to be at, that means its even HARDER to get to playing against people of the same skill as you as then you have people who would never smurf, fighting people worse then them. SBMM is 100% horrible and the wrong way to matchmake.
Yeah, most players will never play enough games in a season to actually reach their true rank. This is going to be particularly true when small swings in MMR can have a large impact on your overall rating.
Blizzard has literally stated that they rig matchmaker to prevent streaks in OW2. Forcing wins/losses.
Some of these comments self-reporting that they didn't watch the video.
Yeah, in my defense I was partway through watching it when I decided to write my comment about my MMR matchmaking horror story that I messed up conveying due to being autistic. It was a literal cycle of lose-win-lose-win for nearly 3 days of gametime spread over 2 weeks. It did ruin Heroes of the Storm for me.
In my case, I knew Grubby wouldn't address the issue properly, and I predicted correctly, so I stand by my comment before watching the video to the end.
I am an aspiring game designer, so I analyzed this issue in depth, for years. The issue is, and has been reported by Challenger players that LoL became harder to solo carry unless you're in the top minority. In earlier seasons, it was easier for one player to destroy the game. Nowadays, the game is more team reliant, and you have little control over the output.
What people are complaining about isn't 50% win rate, it's that it's bad design to use ELO, which was developed for a single player game, in teams of random people.
Real world organizations, like companies and sports clubs do not operate like this. Instead the MMR system should track teams of regular people playing together.
The advantage ? It's the skill of a real team that is tracked. People don't tilt and go about harassing strangers because they have a bad game, or run it down (feeding). In a real team, people communicate, teach, learn, and analyze games together. They also scrim against other teams.
The modern MOBA style of game design is the antithesis of normal social behavior, but AAA game publishers prefer to let toxicity fester, and I am not sure why it makes more money to have the player base so aggravated and unhappy (many Challenger players dislike LoL, but it's their job, so they stick with it). Besides, at their level, arranging teams for skillful play, is probably easier.
@@Edin116 I think it is worse when you get stuck in your own head. I have reached legend rank in hearthstone 3 times as a ftp player, and EVERY TIME I reach diamond 5 and 4 I get nervous and start making mistakes, enjoying the game less, and getting extra toxic. Then I stop for a few days, get a new deck, and all is roses again. I can only imagine in games like Dota and League this is exacerbated because you are sharing your own horror with 9 other people.
@@lilhaxxor If it's harder to carry and the game is more team reliant, then ELO based SBMM is doing it's job perfectly as intended within a team game. You can't cater the entire system for players who want to play within a team game as if it's not a team game (i.e. "carry"). There are plenty of 1v1 games available to play for players who don't want to rely on teammates.
@@nankam It's not what I am talking about. You ignore the main point as if you didn't read, or maybe you only care about your own opinion, or I didn't explain myself well enough.
I said the system should not rank players part of a random team using a system created to rank individual performance. What people are pissed about, but fail to articulate about the forced 50% win rate isn't that it doesn't work, it is that it is dysfunctional and frustrating, takes too long to learn, and doesn't reward improvement like it should.
If I play solo, or I play with the same players as a team, using elo works, and is fair, and doesn't result in as much toxicity, because you have more control over the result. People are mad because with strangers, you win or lose based on random factors more than your own performance.
Obviously, there are limits to what I am saying. A top player will carry most of the time, and a bad player will lose most of the time. However, LoL design, and I am speaking of solo queue here, could be improved easily, to make it more healthy, friendly, and fair/rewarding by a simple change in mindset.
The fact that everybody accepts the status quo (including Grubby in this video), as if it was law, and there is no other way to design an enjoyable competitive system... baffles me, because it's just shortsighted (or dare I say lazy).
Get to the root of the frustration, and fix that, instead of arguing for punishment systems that only deal with the symptoms.
There are matches where I feel that I'm the worst player in my team. Losing these is very different of matches when my team is way below enemy team on skill. I have no objections when matchmaking gives me better opponents when I'm on a win streak. I have objections when it gives me two noobs and opponents are as usual
Just imagine the amount of times you yourself have been the noob on the team. Or the amount of times you've stomped a noob but chalked it up to your own skill.
@@Dancyspartan 1st of all I always say I'm noob, so ppl not expect much from me. So ppl can either relax and play for fun or try to win with consideration that I won't help much. Many noobs in my matches wont tell anything and prefer to make me find out they a 0.5 of a player in the middle of fight, when you expect some help
@@DancyspartanI don't think that means what you think it means. In the games where he's the noob then the problem is that he's thrashing another one's game. It just reinforces his point really. You tried to make it about him so that he doesn't know how to answer but it is actually not a counter argument
Sorry the matchmaker isnt god that can predict how every player is gonna play, maybe your teammate just woke up on the wrong side of bed that day
@@elvis4868 but it actually good in predictions, cause after reaching 70% winrate it gives me teams that persistently lower my winrate back to 50%
Little HotS story from myself:
I was playing ranked with my brother and a friend. I came from DotA and was a pretty good player, so when I started playing HotS, I was performing better than my brother/friend. We played around 100-150 normal games. When we started then playing ranked, something weird happened. My friend/brother, who played more than me, and played ranked before got seeded into diamond/gold, meanwhile I got placed into bronze, even though we played all our placement games together and I had better stats than them in every game.
Matchmakers will account for your performance in previous seasons when assigning a rank to you. So my guess is that your friend has already been marked as diamond adjacent player, before the matchmaking even begun. In fact it probably boosted him higher then it should have because he had an uncertain player with no prior history in all his games (you).
Normal games don't count for ranked? What did you expect?
I cannot speak about whether anything is "ruining" multiplayer games as i don't have that kind of experience to make a judgement like that.
The only thing i can speak about is what gets in the way of my own enthusiasm for the multiplayer experience, what drags it down and what bolsters it, and right now i can say what's "ruining" multiplayer gaming for me personally is complex team based games.
If i lose a game, i want it to be my fault. If i win a game, i want it to be a credit to my abiities.
Back when i was younger and heavily invested in multiplayer games, there were only two kind of games. Competitive PvP games where individual player ability was instantly and easily measurable, either because the game was 1vs1 or because the modes were simple and transparent (I.E., capture the flag in a shooter), and team based PvE games (like MMOs), where a bad player could cause the party to lose but was easier for people to get good because the skill ceiling for PvE content was basically fixed at a certain level.
Whenever i play a multiplayer game today, it's like i entered this mysterious, esoteric realm where nothing anybody does can be easily measured. I can only imagine this is what leads to a lot of the toxicity you see today. Everybody is suspicious of everybody else. If a game is lost, they blame their team mates, even though it's not immediatly obvious that they played badly.
Even the topic of this video can only apply to team based games. No amount of shenanigans from the match making system is going to drag down somebody like Happy. In 1vs1, if you are good you are good, and that's where it ends. And if you are bad, your loss is your own. Nobody is going to hate your for it, except yourself if you are the kind of person who enjoys auto commiseration and dragging themselves down.
Honestly my main take from this video is that team games should indeed go towards performance based matchmaking. Even if the only difference is that improving/failing players rise/drop faster, then it's already a big step towards perception of the game.
At the same time that's why team games with random matchmaking got popular.
You lose? Not your fault? You win? You did great!
Majority of people tend to lie to themselves and team games where people don't play with their own proper team are great at feeding delusions to those who want it.
Also split responsibility means less stress for many - back in the days lobbies were filled almost exclusively with tryhard folks. Quite natural, competitive environment for competitive people. More causal gamers sticked to single player content only. With how modern monetization works, multiplayer gives higher pressure for people to huy skins - so even casual player are pushed into online lobbies deposite it's not really environment for them at all.
Modern multiplayer game principles are just... Weird when anybody will think about it for a while.
@@Anterwaare306 Performance based matchmaking is something that almost always sounds better on paper than it is in practice. It can lead to perverse incentives where players start making decisions to maximize their MMR instead of trying to maximize their chances of winning (like intentionally prolonging games, or to try to "kill steal" from teammates even if it's worse for the team as a whole etc.), or it can lead to situations where certain playstyles can result in a 50% win rate because they're playing in a way that gets more/less "bonus MMR points" (if you get a lot of extra MMR points by your playstyle but actually are at the MMR you should be at, then your MMR will increase and then you start losing more than 50% of your games because you're at a MMR that you just shouldn't be at, but because you're getting bonus points you'll stay at that MMR even though you aren't actually good enough to be there).
I could see having performance based MMR to handle outliers where someone is obviously performing way way better than they should be, but it should only be designed to handle outliers and shouldn't be something that average players should ever be seeing (and it also should definitely never exist at all at the competitive ranks, it should only be a way to get much higher skilled players through low ranks quickly and nothing else).
@@asdfqwerty14587 Performance based matchmaking works really well. There are a couple of old school games where people have programmed their own MMR according to multiple variables, one of them their individual performance. It gets constantly updated obviously, those systems are not perfect and never will be, but c'mon, a multi million dollar company can't figure out matchmaking and one dude alone can do it in his spare time?
Pretty sure matchmaking on these popular videogames is bad on purpose.
Aaah soyboy mentality
Great video. You missed one thing though. It a player climbs to diamond playing Camille top and decides to pick his 25% winrate Teemo in your game. We’re no longer playing at the same skill level with different weaknesses. Instead there’s a gold teemo in your game.
riot mandated loser q
Pentaless off nunu, get me out.
That can also happen for counter picks, where the diamond top gets counter picked by the gold top, and now the diamond top has the game impact of a gold player, and the gold player has the impact of a diamond player, this can only happen because match making happens before champion select, and not after, as well as champion select being dodgeable without losing MMR
dont play league anymore because of vanguard, i dont trust rito and they went woke
@@soldier22881oh good grief get a clue. If a company wants to f up your PC or steal your info they don't need a kernel level anti cheat to do it. They could just put it in the code of the game itself.
I love these type of videos, I could listen you talk about almost anything. Which includes the Warcraft 3 tier lists even though I haven't played or watched W3 in at least 10 or 15 years.
Resetting the MMR wouldn't actually remove smurfs and boosters.
It just effectively makes _everyone_ both a smurf and a booster at the same time. Your matches would just stop making sense, and if anything you'd start to feel what it would look like if the matchmaker _really did_ go out of its way to pair you with and against unfairly imbalanced teams.
Disagree. It resorts in no time and let's people out of Elo hell created by playing too many games after coming back or too late in their day.
A lot of rankings, like LoL, has too many players that are "boosted" and remains at ranks they don't belong at from several seasons back from being boosted once
If it resorts in no time, then you're going to be complaining about it again in no time.
If it doesn't actually resort in no time, however, then resetting everyone to zero just puts you in a far, far more extreme Elo hell.
Without MMR, the system does not discriminate: it will not hesitate to make half of your games trivially boring, and the other half stupidly unfair.
Did you hate playing with a couple of diamond players in a masters level game? If so, then boy are you going to love playing against two diamond players, two masters players, and a champion when you got stuck with three bronze players and an iron.
Do you enjoy when your decisions matter in a game? Well, too bad, next game the rest of your team is several ranks above you. You might as well just AFK this match - you won't even come close to changing the outcome anyways, and being AFK lowers your chances of throwing the game by feeding too much.
@@Gofex1337 That's a fair point, boosting does cause a huge amount of problems for the matchmaker when any of the other 4 people in your game can be one. But turning to the matchmaker to solve a problem caused by people circumventing said matchmaker won't really accomplish anything.
In fact, if anything, it would make the problem worse. By resetting everyone's rank, the demand for boosted accounts will skyrocket, and boosters will make a fortune off of it. A fortune they can spend by advertising and creating _even more_ boosted accounts to sell.
The solution there is to search for clever ways to catch boosters without generating any false positives, so you can ban the accounts in waves. Then people get angry at the boosters, stop using their services, and the boosted accounts all get removed with time.
@@Gofex1337 No one in League remains at a rank that they didn't earn if they keep playing. That is pure copium and complete ignorance of how matchmaking works.
Honestly matchmaking ran by developers has ruined multi-player. The best times for online gaming was when it was all community ran servers.
Grebi baiting all the noobs with this title
the only thing worse than buffing high skill characters because they are allways bellow 50% win rate , is buffing characters for years in a row because they are bad in high skill or pro play , for example tridamere or nasus in league , or sniper or SF in dota 2
Reading the chat is the same experience. The conspiracies I was seeing were hilarious. All these obscure matchmaking rules people were inventing that are clearly untrue. I wonder why no-one who's Immortal or high rank has these conspiracy theories about matchmaking?
"Grebi" made me confused for a second lol
it made me confused for a while
@@SirTatManTat but i've seen some... cant remember now cause I dont follow gaming people too much but I've seen someone in dota
While 50% matchmaking is more of a phenomenon than an intended effect, the "churning" system that league of legends Riot Games have developed and patented is very much real.
The churning system is a system designed to keep you at a positive winrate for the first 5-10 games you play, but to slowly reduce your winrate as you play in order to get you to keep playing with the sentiment that "you cant end on a loss".
Its stared in the patent that it is a system designed to slowly increase frustration in the player at a rate that doesnt immediately turn them off playing the game.
It has a flagging system for "toxic" players who often get reported by their teamates, or also teamates who are in roles they dknt normally play, aswell as putting you on with teamates who are already deep into churning and frustrated before the gsme even starts leading to reduced chances of winning.
The thing is, the greifer is never just one person, and they are always on the same team 2/3 of them. The entire game, they will be typing and arguing in the chat over nonesense, and at the end are all calling for reports on their teamates in all chat.
What i see from the opponents all chat is "gg".
Then in games im "supposed to win" according to the churning algorithm, the greifers are all on the enemy team. 8 times outta 10 someone from the enemy team will say "please report xyz, they are refusing to play with the team" or the argument will spill over into all chat with opponents trying to grt my team to side with one of them to report the other.
Evey. Single. Game.
One teammis coordinated, clear headed, and on the top of their game. And the other team is a toxic cesspool of a crab bucket with the majority of the enemy team pulling eachother down.
As aomeone who was heavily addicted to league, i actually noticed when the matchmaking changed to implement the churning system, because we stopped shit talking the enemy team in all our games, and somehow, started shit talking our allies.
So while 50% matchmaking doesnt exist, there are systems currenty active that are designed to make players just wanting thst one hit of dopamine for winning their game, but the game wont give it to them... so they queue again...
Yes, it is 1000% real. when i used to play, i would relish the one game out of 20 that was a fair match and watch the replay over and over again after 19 crappy games
I have 3000 hours in Dota (quit a while ago) and people always said that the 50% forced winrate doesnt exist. I can pretty much confirm that it is there, because im not dumb but I cant test it scientifically obviously, but its actually industry standard to have matchmaking that manipulates the winrates for people.
If forced 50% didnt exist, you would expect YOUR TEAM to, on average, be EXACTLY the same skill as the OPPONENTS TEAM. In theory, as you climb the mmr, it is YOU that can't keep up with YOUR TEAM and the OPPONENETS TEAM. You just dont have the skill and drag your team down causing you to lose and drop mmr into a 50% equilibrium.
This doesnt actually happen though. What happens is, generally, is that after a winstreak it is YOUR TEAM that becomes WORSE than the OPPONENTS TEAM. It is very evident when you play, all the sudden you start to see people making stupid misstakes or just trolling on your team, despite you having higher mmr than before. This shouldnt happen if the forced 50% mmr did in fact not exist, but we all know it does.
@@Skumtomten1 >YOUR TEAM that becomes WORSE than the OPPONENTS TEAM.
Yeah, because there are 5 higher ranked players on the opponents team, and 4 higher ranked players on your team, and you.
@@zym6687 That's not what he said though, what you are describing is expected behaviour that will say this is not your place you are not good enough, you are the reason your team lost. What happens is that you are still pulling your weight, maybe even winning your lane, but your mid and top for example are 0/12/0 combined in 10 minutes. I am talking from personal experience, it happened so many times that it can't be accidental, especially in your first series of promo games. That of course does not mean I was not on opposite side where my opponents are trolling and feeding. But the bottom line is that I don't want that either, even if it means free win for me. I would be so fine with prolonged queue times if it means better match experience. I would rather wait additional 3-5 minutes than get stuck in a guaranteed loss for 25 and have miserable time. I am talking about LOL, that's the game I played the most. Similar problem you can see in world of tanks, where when they changed matchmaking, a 15:3 was most common way the games were resolved (meaning that one team killed all 15 other tanks, and lost only 3). Around 80-85% of the games were one side steamrolling the other and nobody had good time. Not sure how it is now since I quit that as well
Yeah but if you or someone else is carrying then the shit talking just stops. You need to create strategies to stifle mental bleeds and mitigate morale loss. This is not just a performance based game but a team game. If you're good enough it doesn't matter but if you're stuck down here with the rest of us it helps to know how to keep morale high.
Thank you regarding hard MMR resets. I believe it's mostly to promote seasonal rewards and battlepasses but there should always a soft cap when reseting MMR cause the 1st week is just full of bad one-sided games.
Also I loved the introduction of sbmm because I remember playing older games which only had lobbies and it was a complete crapshoot where you had master players paired with bronze players by skill averaging and everybody just felt frustrated, unless ur one of those people who enjoy noobstomping.
Even with a soft MMR reset it often skews too much towards the "reset" side. I main LoL and the game's ladder is literally unplayable for at least 1 month; often 1.5~2 months (and they've made it so the reset happens 3 times a year now because retention metrics or something I guess. Woohoo.)
@@FreakattakerThese 3 month resets are terrible imo. First week you play low quality games feels your not learning its just faceroll winning, then you get to the top of the ladder and queues become long and matchmaking quality goes down even more cause now, you essentially have 2-4 pro players matched with 6-8 terrible players, which in mobas where the game snowballs so hard if your adc is pro and the enemy adc is terrible the game is over. Doesn't matter what you do, its over. On the other hand if its in reverse than you lose regardless. So you are forced to play like a 100 trash games in order to play decent players and then its who gets the pros in the key roles. Or which terrible player is capable of not feeding immediately. And also the reason your on top of the ladder is not because you are the best player, its actually because there are many many players who skip this period so they can boost their winrates once there are less pros in low rating. So essentially it takes like 2 months for you to actually be able to play proper games at your skill level. Then you can play for a month and then back again at it. The only way around this is premades. Not only that but you need 5 man team of top 0.1% and even then most of your games are boring easy wins. Mmr resets make mobas terrible, its always this cycle repeating and if you play soloq then you are in bad luck cause at the top end everyone plays premades only so after a while you have no real competition in games and if you do its winning or losing is still not decided by that. Thank god im not invested in being good at mobas anymore.
There's one circumstance when MMR resets make sense, which is during the development of a game when the game is rapidly changing and not finished yet. But besides that, yeah.
@@TheRealXartaX But why reset if someone just plays they will lose their mmr anyway, artifically raising and lowering mmr is annoying for serious players. Its this mentality of give people a sense of accomplishment, but it just doesn't work above a level. People who actually try know its meaningless, your mmr doesn't matter your rank doesn't matter, all that matters is did you play perfect or not. And also games like HoTS have this backwards system of giving people MVP, this is the same issue. How does a game decide who is MVP, is just a joke. It doesn't deeply analyse who plays best it just looks at a few numbers without any context and based on that it tells the player good job. Again this doesn't mean anything, it is completely arbitrary, it is the same as in chess where if you win a piece you get points. Great but I don't win by getting points so it doesn't mean anything, and it just makes most players think they are good and reinforces bad decisions.
For most people mmr resets just means one thing, inflated self esteem. Sounds good, but then you realise do players in team games really need more self esteem? It is a recipe for disaster, now not only that skill gaps are all over the place, everyone thinks they are rank1, leads to toxic behaviour. Who came up with this?
@@AnonYmous-spyonmepls Because a reset in that case will literally be closer to people's real mmr than where it was. When a game is rapidly developing with huge changes you will have things like broken strategies that gives people hugely inflated MMR's if they discover it and so on. And I'm not talking about something like a character being a little too powerful, I'm talking about things like being able to clone relics in AOE4 so you generate infinite gold.
I think the issue most people have is that "Casual" game modes have too strict MMR groupings or use ranked MMR as the basis which leads to them feeling too sweaty for people just trying to have some fun or learn or play off meta. As you showed in the video player skill usually sits on a Normal distribution so having tight MMR groupings ultimately doesn't change your overall win rate it just lowers the potential "variance" per game which could be seen as good (close game) or bad (sweaty game) and having lots of "ranks" across that distribution (beyond 3 or 4) is probably meaningless but could be seen as good (yay I am ranking and improving) or bad (we have a 2 rank 5 diamonds on our team and all their team are rank 3 diamond and above).
I never played SC2 but I agree with idea that ranking system are at their best when its more a recognition (not fully) of the amount you have played because of what I outlined above and if its designed well it increases positive engagement (losses don't feel as bad) plus you can more easily couple it with re-occurring "resets" which also increases engagement.
I like how the section where you talked about increasing player satisfaction was basically what EOMM was trying to do, but just better. Instead of aiming to essentially rig your games to keep you playing for longer, things like bonus pool give you an additional reward that boosts engagement more effectively than EOMM could without unnecessarily eroding your match quality.
The problem with EOMM is that by seeking only to maximize how much time we spend in-game, it completely misses the point of _why_ we choose to spend our time with a game in the first place.
It is a system based entirely around doing what it _thinks_ will minimize the chances of us quitting - a metric partially based on statistic like play time and games played, but is also in large part based on your tolerance for losing streaks. Think about it: if you tolerate losing more than the average person, EOMM *should* be harsher on you. If you can handle a 5 loss streak, then you're the perfect fit for 5 higher rank players who would quit after a losing streak of just 1 or 2. Sure, you won't have fun. But you also won't quit either, and that's the only metric that matters - which makes you the perfect candidate to be today's punching bag.
Until, of course, the day you decide it's not worth it and quit for good. The algorithm can read data, but it cannot read your mind. And sometimes it cannot collect the data to know when it has pushed you too far until it's already too late. Yet, even if it _does_ learn to let you win barely enough to keep you playing, it begs the question... if the game had to trick you with an imbalanced W/L ratio just to get you to stay, then was it even worth playing to begin with?
Optimizing play time, depending how you collect the data for creating the system, can end up only optimizing for the near future. Because how can they know the system will burn more people out over 10 years when the game has only been out for 2 years. Just because someone is temporarily hooked and plays very consistently doesn't mean they'll be around forever. Maybe satisfying them more but having them play less could net you more hours of play time in total and more secondary traits, like buying microtransactions or new expansions
huh world of tanks uses a similar system instead, the punching bags are free to play players (me) and the winners coincedently happen to be the ones spending money on the game using very souped up tanks. glad i quit wot
@@auberry8613 I don't think gaming companies really care about making evergreen games that would last for 10 years.
I'm afraid they will take notes from the successful social media companies which waste our time with ragebait and misinformation to keep us engaged. If these kinds of systems are implemented successfully for matchmaking in games, they might persuade us keep on playing while making the complete experience worse for everybody.
10 years later and you're still going. Good luck brother! Was a big fan of your more creative plays back in the WC3 days.
Great video, Grubby!
I was really confused by the title and 2 min intro^^ It sounded so wrong, yet was delivered superbly professionally, causing a lot of confusion.
Some thoughts.
1. criterium free matchmaking works fine for casual games like cod or team fortress 2 where most people dont care about winning or losing and just want to have fun, play with certain weapons, farm unlocks and experience, try goofy strats or just practice their skills at certain game aspects. It doesnt only work well for games with low player counts.
I had plenty of fun playing cod without rank based matchmaking and so did everyone else back then, and those games were huge in the cod4-mw3 era which is when I played.
2. you learn a new game way faster if you are randomly matched with good and bad players in a wide spectrum than if you only get matched with other noobs at your noob level who dont know how to play either. You will die to the good players, you can spectate them while you are dead, you see what they do and how they kill other noobs.
It also gives you clear feedback, because if you manage to kill a good player you will know for sure that the play you made was actually good and not just the other guy playing even worse than you.
You will just have to push through losing a lot of games until you git gud, but if there is no ranked matchmaking, you also arent punished for losing in any way, you can just go play the next game against other random players with random skill levels.
3. SOME games do give you worse and worse teammates if you win a lot of games. Often this is defended as a measure to prevent smurfing, people who make new accounts or derank intentionally to noobstomp. This is to punish smurfs for smurfing and prevent them from having fun noobstomping, which discourages smurfing. You wont have fun smurfing if your teammates are such bots that even if you full tryhard you will only get a 50% winrate. This is 100% certain a feature in counter strike's matchmaking. Valve even specifically talked about implementing such a measure to combat people smurfing on new accounts. The thing is, it also happens to me on my main account with 8000 hours.
What it actually does is prevent good players from ranking up, forcing them to "smurf" indefinitely, even on their main account, if they get placed in a lower league in a season or after a server reset.
5. As mentioned, I have over 8000 hours in CSGO and achieved Global Elite, the highest rank. When CS2 came out everyones rank was deleted and everyone had to start again from silver.
I barely managed to grind up to MG2 in a year since the game released because of these anti-smurfing measures constantly putting me in the absolute worst games with noobs and griefers and cheaters where I have to drop 30 kills every game to have a 50% winrate.
If I inform the enemy team that one of my teammates is blatantly cheating, the cheaters report me, making me even more likely to get queued with more griefers and cheaters. I also get kicked all the time by people who think Im cheating because Im not supposed to be at a low rank, Im supposed to be in the highest rank with the other 1% of the best players.
This makes a terrible experience for literally everyone on the server. I cant inform enemy team they are playing against cheaters, and I cant overperform relative to how good I "should" be according to my rank, which is way too low, or otherwise I will be punished by my griefing and cheating teammates as well as the matchmaking system.
5. Another issue besides MMR reset is MMR decay. If you dont play counter strike or starcraft 2 for a few weeks, the game will just start steadily lowering your elo. If I come back to SC2 after a few months of not playing, I have to grind up from 3000 to 4000 mmr every time, just to get back to my actual skill level. That means i have to get 40 net game wins (minus losses) to gain 25 mmr per win, 40 times. This means I have to smurf for at least 50 games, assuming I average 80% winrate against players significantly worse than me, just because Blizzard wants to punish me for not playing the game for a while. This is not that much fun for me, although cheesing noobs can be fun, but it is definitely not fun for my opponents.
The bonus pool doesnt apply to your visible or hidden mmr, only to your league points which literally nobody cares about and is completely irrelevant. So SC2 just like WC3 does punish you for not playing.
6. EOMM is definitely a thing mostly in games with microtransactions. If you buy them, you are matched against lower skilled players so you win, and if you dont pay, you are matched against harder players, so you lose. This makes people subconsciously associate buying microtransactions with winning more, without being aware of it. It is very effective and probably illegal under EU law.
7. Engagement optimized game design is also a thing. Games often add some weapons or heroes or items that noobs can use to get an easy kill every once in a while. These weapons are often derided by the community and you will get flamed for using them, but devs keep putting them back in.
Famous example, in the original COD MW2, the first unlock you would get on every assault rifle after getting like 20 kills with it was a grenade launcher that oneshots players in a radius the size of half the map. Noobs could use this to get some kills on players 500 times better than them and almost everyone hated it, but Activision kept putting the noob tube in their newer games precisely because it levels the skill gap and prevents noobs from losing games with a 0-30 scoreline.
I play Starcraft II, and I actually like MMR decay.
I often go a few months without playing, during which I lose some of my ability to play the game well, and the decay normally works out well in terms of matching my actual ability after not playing for a long period. This means I don't have to go through a long series of losses from being matched against players that I _used to be as good as_ while I get back up to speed when I start playing again, and makes my MMR more representative of my actual ability. Different people learn and forget at different rates though, so the system won't work for everyone.
1. Even in casual games, only some of the players actually play casually.
Some one did a video on this, back then, people were playing more casually but now there are so many streamers that are playing more sweaty for their content and also push out to others what are the most optimal strategies.
2. If a good player is so much better than a noob, the noob might not even comprehend what’s happening, let alone learn anything.
The number of people that are willing to push through an overwhelming number of loses and also do lots of side research and learning of all the advanced mechanics is incredibly small.
Baby steps is the better option for most people.
3,4?,5. Only pointing out your contradiction: if you think noobs should play with better players so that they can learn, then why are you upset when you have noobs in your team? You just said that you think it’s a good way for them to learn.
@@nathangamble125 well I lose 1000 MMR when I don't play and my actual skill only decays by maybe 200 points, so maybe you should just lose until you get to your actual skill level instead of the game forcing me to smurf for 50 games.
@@RevCosmin it's good if I get some noobs in my team, not if i get put in a team with 4 noobs against people 5 ranks higher than them and i have to get three kills every round. Ideally every team should have some experienced players, some mid level players and some noobs.
Playing meta only has a marginal impact, I can play off meta in a game I'm good at and maybe sacrifice 5% of my optimal results. In fact that's what I'm doing by default because if I spam smokes every round in counter strike and accidentily get some kills my account will get reported even more, I will push my trust factor so far into the ground I will be in the third basement.
@@RevCosmin "the number of people who want to push through losses, do research and learn mechanics is very small". Yes, I agree. The people who don't want to do those things should play casual games instead of hardcore competitive games that require you to do those things to be able to win games.
Instead of playing csgo or sc2 they could just go play cod or some autobattler that doesn't require in depth knowledge of mechanics and strategy and have 300 apm and high accuracy and being able to communicate while doing all those things.
Or idk play Minecraft or something.
Inviting people who don't want to grind for skill into games that require grinding just sets them up to either have a miserable time or start cheating.
Just to correct you on the SC2 system a bit, SC2 has MMR as well as ladder points. The matchmaking and your league is dependent on your MMR, while ladder points is just fluff (although it is slightly correlated to your MMR within your division). Bonus points added points to your ladder points, so it is just fluff on top of fluff.
I think most of the frustration comes from not understanding the matchmaking rules and feeling "cheated". This is understandable since matchmaking code is almost never open-source and many games will hide your true mmr score. It also makes it impossible to determine whether the matchmaking system is truly broken.
And the system *could* be broken. MMR depends on an iterative mathematical model where the MMR scores of players are used to predict the likelihood of an outcome and the true result is then used to adjust the MMRs. The quality of this depends heavily on how accurate the mmr-based predictions are and how mmr's get adjusted. If mmr's get adjusted by too little then the system takes ages to stabilize while if it gets adjusted by too much then the system becomes chaotic and may never stabilize. This is made even harder when you consider that a player's true mmr changes over time.
I don't think we can discard the possibility that the matchmaking for some games is deeply flawed and it's frustrating that we do not have the data to prove it one way or another.
Let's not pretend the matchmaking in team games like dota actually works for majority of people. When I still bothered to play finding actually balanced game was about as unlikely as in a game I'm currently playing that has NO sbmm at all. Its usually a stompy win/stompy loss
The explanations are amazing, thank you.
TALK GRUBBY TALK. TALK THIS ONE TO THE MOON.
The problem is skill based matchmaking DOES match you with potatoes. Many games don't or poorly calculate individual skill on a team, so you have people who get a good rank because they got carried, or have bad rank because their team sucks.
Skill based matchmaking is bad because skill calculation is bad. Another reason it doesn't work, is the human element, because some people have off days or really good days. I think every game should have a ranked and unranked. It's fun to be able to destroy noobs. It's also fun when you have that one god tier player you have to really avoid.
0:29 no I'm losing because I'm bad at the game. You can still improve and be quite skillful.
This is true, especialy in like 1v1 games. In team games, this is true aswell, but there a win is a team achievement, not depending on a single person.
I agree with you that you can always improve and you definitly shouldnt try to always shift the blame onto others. But you should know that solo carrying in a team game is extremely difficult, especialy if the odds are heavily against you, which is why a forced 50% winrate isnt rly fair for players.
@@crisscore157 If I wanted to win; I would play meta. Hell if most people wanted to win; they'd play meta teams.
You're going to have bad teammates; tough luck, just like with a job you're going have to try to do the best you can with the hand you got.
There are other gameplay systems that could be implemented to make things less hard carry if it's such an issue.
@@soundrogue4472 i agree, but you cant pick for your team mates, can you ?
Im not disagreeing with you and i do believe that you should always strive to improve (and ofc dont shift blame if you could perform better). I just wanted to add that you kinda need a certain amount off luck and if that is rigged, it gets harder to win
@@crisscore157 Sure you can pick teammates; get friends. I am stubborn and not going to budge on this.
This is like going to the Casino and being shocked when the government lets them not give the pay out for what you were owed if you find a way to game the system.
Lmao stop yapping there’s a reason even the best players in the world max out at 50% across the games he used as examples in the screen such as dota and league. Obviously you’re not the best player in the world yet the statement stands, there’s a 50% wr system holding players back in which yeah if you’re better than everyone else you can win 1v9 yet having more difficult games due to having higher wr and the game not wanting you climb too fast is clearly unfair.
I will share this video EVERYWHERE. Everyone playing these games need to hear these words.
Grubby converted to the eastern european dota 2 player school of thought.
Can you explain me how? Don't know exactly what does that mean.
@@mackparker9064there is kind of viral theory of how matchmaker deliberately throws you to players with a bad time or lack of skill or/and gives you strong opponents so your win rate stays the same. It is particularly popular in DOTA community
@@gregory7090 is this to cope or a real theory?
@@mackparker9064but the conspiracy is that matchmaker of DOTA matches teams not only by their shown MMR, but "hidden" MMR (which is calculated by performance statistics and maybe something else) also. And these hidden performance statistics could be manipulated to match teams so that one team loses and other wins.
@@gregory7090
I can assure you, it's not just popular in DotA community, but in all MOBA community. It comes with the genre I guess, being 5v5 game and all.
What a great video.. btw a good recent case study is Overwatch 2.
People used to complain so much about it but not any more and it increased participation in ranked games.
Basically, among many things, they made the system update more frequently, more transparent, and flexible.
Rank used to update only every 7 games, now it's 5.
They tightened the MMR spread AND allowed wider grouping in ranked games ... resulting in two kinds of matches.. the tight matches and the wider matches.. and told people that wider means more punishing losses for the higher MMR people.
A lot of good "theoretical" points, but until MMR algos in games are open source, that's all it is.
DotA2 was also "talked of" as using a fair and well implemented MMR system, and then it became known that it intentionally took extremes of both Elo and Attitude to create "equalized" teams, leading to hilariously bad matchups and experiences.
It is always easier to create a system that cheats, and claim that it works (make a system that forces 50% win rates, and then say that the 50% is how you know the system is working), rather than make a system that actually works.
this tbh
Dota2 matched feels more like playing a slot machine than a matching system. At max behavior and communication scores I often get placed with crazy toxic griefers and leavers after a win streak. I then go onto lose like 5+ games in a row where I will still have a very good game but I can't out carry the tozic behavior, role abuse picks, etc. I was still climbing but it was a long exhausting grind where most games are stomps, many matches there are people with under 500 matches pitted against players with thousands of matches. Other matchmaking in other games did not feel like this. Something must be wrong. I feel like it might be because they're cooking the books for a paid subscriber base that is mostly toxic and terrible.
@@parawizard I have 3000 hours in Dota (quit a while ago) and people always said that the 50% forced winrate doesnt exist. I can pretty much confirm that it is there, because im not dumb but I cant test it scientifically obviously, but its actually industry standard to have matchmaking that manipulates the winrates for people.
If forced 50% didnt exist, you would expect YOUR TEAM to, on average, be EXACTLY the same skill as the OPPONENTS TEAM. In theory, as you climb the mmr, it is YOU that can't keep up with YOUR TEAM and the OPPONENETS TEAM. You just dont have the skill and drag your team down causing you to lose and drop mmr into a 50% equilibrium.
This doesnt actually happen though. What happens is, generally, is that after a winstreak it is YOUR TEAM that becomes WORSE than the OPPONENTS TEAM. It is very evident when you play, all the sudden you start to see people making stupid misstakes or just trolling on your team, despite you having higher mmr than before. This shouldnt happen if the forced 50% mmr did in fact not exist, but we all know it does.
Insightful, informative, and most importantly the not pointing out weaknesses is hilarious
overwatch admit to a forced 50% winrate system
Amazing video as always Grubby. I find this a very interesting topic (as an engineer working on similar non-gaming algorithms) I'd love to see a follow up that explores the insights that has come from different 'less subtle' variations to the matchmaking algorithms to target certain outcomes that has been published publicly or leaked (and confirmed). Some of these systems are developed and sold to studios (or in the case of Activison, they have spent a lot of time and money developing and testing multiple iterations and published (or had leak) their findings).
One thing that wasn't mentioned and i think is the real problem with our games mmr system. Is the hidden mmr.
The hidden mmr is the real number the matchmaking system is using to decide what players you should be playing with, instead of the mmr that you currently see on your account. For example Wow arenas has this system visible to the player.
Hidden mmr is bad for the matchmaking system because when you win a game it increases the hidden mmr based on your whether you win or lose and your performance relative to the other players, sometimes you can have a really good game just because you had a good matchup. So due to you being extremely lucky versus those opponents you will get a high performance, causing the game to increase your hidden mmr by a huge margin.
Then when you decide to play the next game you are being matched with players that are literally 200mmr higher than yourself, causing you to feel like you are just a totally complete noob and making the game feel more unfair than usual. This hidden mmr has a compounding effect as well, so if you get into a high win streak your account mmr will be even further behind your hidden mmr and raise the bar of enemy players even higher, eventually resulting in a lose streak bringing you back down to your true mmr.
The blizzard post about the 50% forced win rate compares players/people to a random coin flip, this isn't true people are more consistent than a random coin flip even including factors like tilting and so on. if a player is 2k mmr that means that player is 2k mmr, if he goes against someone that is 2.1k the 2k player should lose unless they ARE better than the 2.1k player. It's not a random coin flip that makes the 2.1k player better than the 2k player, the 2.1k player IS just better than the 2k player.
If games literally just used the mmr you see on your account regardless of your win streak or loss streak the game would feel more fair and you would consistently see more players with 1W 1L back to back in their match histories.
This definition of hidden MMR is a little misleading.
Usually, hidden MMR operates with a higher k-factor(gain), such that if you win a match, your MMR might increase by 10, and your hidden MMR might increase by 15. If you consistently outperform your MMR, you will enter a new rating bracket with a new K-factor for hidden MMR that is closer to the one that the regular MMR has.
This is done for variance control, and under the assumption that skill increase and skill decrease are not equivalent. Your skills grow faster than they decay in this model.
Players at the absolute top experience very little or no difference between their MMR and the hidden MMR, while players in the middle and bottom are given the opportunity to more rapidly climb towards a 50% winrate when they experience skill growth.
The number of data points is critical. In your example with the conflip it does indeed hold true that human performance is more consistent than a coinflip given maybe 10 or 100 data points. But as the number of datapoints increase, the coinflip will prove more consistent.
I have no specific experience with Wow arenas match making system, but given your description, either the k-factor is too high for your rating bracket, OR you are not providing enough data points for the match making system to create balanced matches. In general these are the only 2 mathematical answers.
@rasmusvinther9190 from my experience the hidden mmr increase seems to be extremely volatile, one win my hidden mmr will increase by let's say 20, second win in a row it will increase by 40 then on the third win in a row it will increase by about 80, meanwhile my actual account mmr is only increasing by around 10, 15, 25 each time. Creating a massive gap between my real mmr and my hidden mmr. This rings true for pretty much every single pvp game I've played. Wow, arenas just make their hidden mmr visible.
The problem I'm pointing out is that the game then starts to give you players that are way better than you are and thus makes the player feel like they are horrible at the game and had no chance. Rather than slowly graduating through the ranks and giving them a proper chance to help win for their team.
As for the coin flip, yes coin flips can be streaky and over the course of 1000s of flips it will result in 50% heads and tails. What I was talking about is that humans are consistent if you are in the correct mmr you should not be getting ridiculous win streaks, you should be losing to players that are better than you and winning against players that are worse than you. Thus giving you 1W 1L most of the time.
The way arena in wow does this drives me nuts. Play new characters, start at 1500 mmr go up to 2k mmr almost instantly since you're old veteran players, but you're playing completely new specs and classes, trying to get those 1600 CR rewards, but you have to fight 2k+ players for them. It's annoying as hell. Just getting by on fundamentals and having to uber sweat every victory when playing to get your first epic pvp item for the season xD
@@kristoferkravis1474 Sometimes the hidden criteria include things outside the match.
Be it behaviour or how long have you logged in. Stockholm Syndrome is a thing, i notice in moba when i talks shit, i get pair up with people that talkin shit. Sometimes when i too good, i got paired up with lower skillset, sometimes also the unlucky matchmaking.
The game literally set you up to lose. At that point i just play for killin time and even got apathy on one point which actually help getting better team
Yall keep forgetting smurfs who ruin the who situation overall anyway because they dont want to play their main "forced 50% account" because that would require trying, so they want to clown on lower skilled people. For fun. Guess whos not having fun when this happens. Anyone else in a forced 50% mmr being forcefully carried or forcefully denied a win by someone who spends their life in the game. And fucking their actual MMR because funzies.
Didn't watch the entire video but it seems you were just talking about players personal perspective bias but didn't talk about seasonal ranking and MMR decay and hidden MMR which is what i expected the forced 50% WR dilemma was supposed to be about
You're seriously underplaying the impact of EOMM. Activision patented it for a reason.
It's no coincidence that community servers went away when games with micro transactions started coming out. The matchmaking is designed to reward players that pay and punish those that don't.
Excellent job Grubby. Thx for taking the time to explore this heated topic :).
Holy crap, people in the comment section are actually taking it seriously :D
More importantly not watching the video. I personally wouldn't laugh at anyone for believing what people put in their titles etc, but taking time to talk about it without watching is a mistake.
@@jpVari is it wrong for people to want to engage n conversation without spending a lot of time watching a video?
also, how do you know if they watched it or not?
ITS HAPPENING FOR LIKE ALMOST 10 YEARS NOW AT LEAST, WHY DID IT TAKE SO LONG??? PEOPLE ARE STUPID THUS IT WORKS!
He might not be serious, but that doesn't mean he is wrong.
@@snuffeldjuret These comments judging whether or not someone watched the video always come down to one question.
"Does person agree with my stance?" "Yes? They watched the video."
"Does person agree with my stance?" "No? They clearly didn't watch and just flew straight into the comments section."
I'm pretty sure most of the complaints about "forced 50/50" (at least that I've seen) aren't so much about the existence of sbmm resulting in that outcome, but rather people saying that devs take the easy route and chase the metric of "if people get 50/50 winrates matches are balanced", which can also help keep queue times consistent if it cares more about average mmr between teams than everyone having close mmr.
That's ignoring Engagement Based MatchMaking, which I feel was glossed over when it's apparently becoming more and more common.
I feel server browsers are easily superior. People typically will sort themselves out into enjoyable games and will take into account things the game simply can't like if they're tilted or on fire, rather than trying to jam everybody into one or two different queues.
EOMM is the elephant in the room, and yes it was glossed over here
Thought this was going to be about faction/character balance, which I agree with. Electric Underground did a video on street fighter 3 not that long ago, and one of his main takeaways was that an imbalanced roster where some characters have noticeably worse win rates than others is good for the game in the long run because it allows for a more varied competitive environment where good players can play bad characters and win because their opponents have less matchup experience.
Interesting idea to compare the meta to the FGC. It's actually similar in many ways. Hate to say it, but fighting games in general could benefit from the pokemon style tier bracket. i.e teir 1 everything is playable, tier 2 the top 10 are banned from competing tier 3 top 20 etc etc... It would be a cool way for lower tier characters or builds in wc3s case to have their own meta. Problem is, it worked for pokemon because it had 152 creatures to play with. No fighting game or RTS had THAT many variations.
@grizzgreen6821 I think it would be interesting for more games to have a tiered bracket system like you described, but I don't think it's really necessary. Just think of the example that Grubby gives in the video with the teemo player. It's a good thing that league has characters like teemo because his playstyle provides players with a change of pace, but you would never want teemo to be meta viable because of how much he differs from the norm, this naturally leads to good teemo players potentially having a game knowledge advantage that can allow them to sometimes cheese out wins, which is the main appeal of playing low-tiers.
Theoretically, anyway. In reality just about any 3S tournament worth a damn is absolutely dominated by the top tier characters. There is a nice grey area on the tier list where there are several powerful characters who can, in theory, beat the top tiers with sufficient skill but in reality the character specialists only really shine when the stakes are low (so nobody is really trying hard) or when the field is soft enough that they can win regardless of who is played.
@MrOrdinaryundone well, yeah. that's kind of a tautology though. Characters with low win rates don't often win. I do think it's disingenuous to say that upsets only happen when people aren't taking things seriously though. Every competitive game that has ever existed has had a meta that evolved over time with underdogs winning when they really shouldn't have. Not all of those victories were because their opponent was toying with them.
Edit: Like, just think about moment 38. For a Hugo player, getting #4 at Evo is no small feat
@@danily11 It's not a tautology because the reason they have low win rates are because they are statisically inferior to the top tiers. Their own weakness informs their low win rate, its not like the are underplayed because they are difficult to grasp or play 3S is a very old and well studied game, if there was some tech that made Q or Alex able to hang with Chun or Yun it would have been found by now even if it was difficult or obscure. And by "serious" I don't mean the players BMing or not taking their opponents seriously I mean low tiers only really get spotlighted when the tournament's rules enable them, like team formats or bans. You have exceptions, like Hayao's Hugo play at Evo this year, but even as well as he played 4th place just reinforces the status quo when he's beaten out by Chun and Yun.
Great video with good explanations. There are just so many people that don't understand these systems but I can definitely understand some of the frustration.
I think the biggest hurdle is psychological and can't really be ignored. It is simply poisonous to enjoyment to feel like you're playing your heart out, and doing well, yet your efforts are rewarded with a loss by other people on the team. Sometimes you're mistaken on your skill level, sometimes they're having a bad day - but there is a concept of "Some games are clear wins 40$ of the time, clear losses 40% of the time, and it's the 20% of the time where your play mattered that causes your MMR to rise" which simply ignores how terrible it feels to have wasted your time. This is one of the reasons why, in DotA at least, some of the most toxic players are pos 5 supports - because they feel like their entire effort was wasted and they feel trapped by the system.
This is what Performance-based MMR helps against, and why I'd ultimately advocate for it if I could. Because it, at very least, says to players: "You are struggling and I see you". And sometimes, that's all that's needed.
Yes, that's the problem entirely. It's absolutely psychological and not just for the person who's playing their heart out, because they will in turn rip on the not so skilled players who are forcing you to play your heart out to begin with. Toxicity is a natural consequence of your actions going unrewarded, and even punished because it's your turn to lose.
They don't want even games, far from it. It's an absolute grind, and people have different temperaments either suited for that or not. But temperament should never trump skill. If you aren't rewarded for being the best performing player, you're doing it wrong and people will act accordingly
Imo any performance metric will be gamed by elo boosters and de-boosters. But it's nice to have some sort of reward for dying little, getting kills, etc to motivate ppl.
As a support, you wont be able to perform well if your team is getting rolled, so performance-based ranking wont help you.
Dota pos 5 buying potion and tango mango for carry to -25
they need to make games where you can 1v5 instead of relying on teams, which is why tf2 is so much more fun to play despite it being a "team" game i checked the dev commentary and they balanced it around anyone being able to carry thats why its so good, you can carry against 15 players and it feels awsome.
Great video Grubby, really insightful! I think this view will help me with PMA in my games when it feels like someone is a griefer.
the biggest flaw imo with sbmm is that it is a concept invented for 1v1 games and then applied to team based games. It is also often applied to f2p team based games which opens up the door for a lot of bad actors (smurfs, trolls, cheaters) to have multiple accounts. Combine this with our tendency to remember and emphasize bad things (ie loss streak, smurf, leaver) more so than good things (win streak, got carried etc) and you end up with a brew that feels bad for most people to grind through every day.
In a team based sbmm game you could very well be a plat player stuck in gold, which can never happen in a 1v1 sbmm game. Are you better than gold players yes, but not by enough of a margin to just carry 4 lemmings when the odds are too stacked against you. the closer you are to where you are supposed to be, the harder it gets to make up for mistakes made by your teammates. A solid argument is that, with time, and many, many games, you will eventually get to your actual rank. But most people aren't interested in grinding a game for 8 hours a day for months just to get to their actual rank.
But now you are forgetting what Grubby said as well:
Those 4 lemmings are as often on your opponent side and you get easy wins against them even when you don't deserve it. So myth busted
Also, without sbmm the results would be even worse. I don't know what are you suggesting bro
@glendaal67 they're saying that matchmaking system isn't ideal for team games.
"actual rank" lol
@@YesThisIsCrass It's not ideal, but it's obvious that matching 10 people with similar skill level is harder than 2. Also the problem is not that sbmm is made for 1v1's, it would be exactly like it is if it would have been developed for team games
3 weaks ago I implemented the ELO match making in my sports team. We play a kind of medieval duel and team (battle) fights in a kind of HEMA inspired sport. ELO is awesome to our context.
Very happy to see the timing of this video ^^
This may be true if the assessment of your skill is accurate. However, if the game thinks your skill is far above your bracket yet still keeps you in that bracket, it will absolutely put potatoes on your team for hte purpose of averaging the team skill MMR. Even if it is a team slayer and not objective based game, you are still getting a 3v4 game dynamic if ONE person on your team is a feeder vs. their whole team being mediocre yet outnumbering you, ala halo ranked matchmaking. It is a LOT worse when it is an objective based gameplay
then you LOSE a lot more points at the end of the match than everyone else becaues the game thinks your skill level is a lot higher (yet they keep you in that rank and give you potato teammates???) therefore keeping you in that bracket of potatoes. Absolutely NOT myth busted ROFL
In halo infinite, for example, if you have an ancient account like mine (since around inception of the game), for whatever reason, you gain a LOT less "exp" than someone who makes a new account and just dominates in ranked. They can go all the way to diamond or even onyx when their actual skill level is close to a low diamond as opposed to people I play in ranked who are just platinum players who are a LOT better than diamond players who only use their teammates as meat shields and have no concept of teamwork whatsoever and only high FPS performance CPUS. I do not play heroes of the storm or anything but I do remember playing dota 2 but it is now all about how polite you are and if you have an account with a history of BM (like mine) bad teammates, you will never match decent teammates, however before they implimented the hush order, the match making was excellent. Completely unpredictable
I had this with SC2 and CS:GO. I watched a few videos on SC2 starting tactics and the game's MM thought I was some expert. Got crushed ~66% of the time and still kept my rank and kept getting crushed because I knew how to start the match and for some reason that gave ranking points.
In CS:GO, I used to be a really good 1.6 player but after a decade of not playing couldn't hit a wall. In CS:GO you had skill based MM. I knew tactics, I knew when to sacrifice but I couldn't hit an enemy to save anyone. Way too high I go while having 1 frag to 2 deaths and I can't get down to my shooting level without intentionally losing all the time by just rushing towards the enemy. So much fun.
"If a matchmaker is badly designed it'll have bad results"
I mean... Yeah
@@nahuel3433All skill based matchers have bad design. :)
You might want to look into why people don't like SBMM in Call of Duty and not just assume that every game works like Warcraft. That system ruins the new games and the whole thing would be fixable without even removing the system.
Strange definition of new; CoD has had SBMM since 2007. You can read about it in Activision's whitepaper on multiplayer game methodology published around that time
The main problem is some games (might) separate MMR and ranks. And depends on how much it's separate.
I think LoL had a system like that. When you reached threshold between rating. Say you go form silver to gold, you need to win 2/3 games next. Failing to win would keep you in silver, and still match you vs silver and low gold players.
But it does keep track of your skill regardless, and will try and match games around both your skill and rank.
And winning the 2/3 games will put you in gold depending on your skill, so can catapult you to mid gold.
It did kinda work out with enough games played, but was weird. No idea if it's like that now.
Overwatch also had something going on with non-rank skill tracking.
Actually, in League of Legends, you might end up consistently matching vs players in mid-high gold until you actually win the promotion series to rank up or have enough disparity in MMR and rank to skip it (IIRC, you can only skip rank tiers, but not the 5 game promo series to advance in rank). This is something that's happened to me and some friends of mine on occasion (when I actually played ranked in LoL; at least one of those friends still runs into this problem regularly).
LoL will also occasionally match teams with significant disparities in MMR/rank, resulting in occasionally going up against a team of much higher ranked players (EG: a team of mid-silver players vs a team of mid-gold to low platinum players, or against a team of mostly bronze players - about full rank of disparity) - this is less common in ranked, but it does still happen occasionally, and matching up is particularly common during those promotion series.
Yeah, it's weird. They should just display an ELO or sth. Then again there's probably a marketing reason behind this.
Its even worse in Starcraft, where you have 3 different ratings. You have your rank, your visible mmr, and your invisible mmr. These can be completely disconnected from each other. If you win a bunch of games in a row you will probably not even rank up, you will gain maybe a few hundred visible mmr but your invisible mmr will be boosted thousands of points, matching you against players several ranks higher than you.
Thanks for immediately explaining that this is about team games within the first 30 seconds. I was confused because I always think WC3 and SC2 when I see your face :)
There's a few things I'd bring up that you either ignored or rattled off at the end without fully addressing.
- snowball factor of characters/items/etc.; depending on game, these are not all equal. Ex: a genji grandmaster with those 4 diamond will have more carry potential than a lili
- sometimes items, roles, etc. are intentionally made unbalanced to incentivize/disincentivize play and influence meta
- bots, smurfs, and solo vs duo "boosting" of a duo's wr
- strategy, skill floor for various roles/characters/items all vary based on ranked; this means you may be more or less inclined to climb based on whether your rank, role, character caters to your strengths; there are of course outliers at any rank too ex: league - adc baseline is higher skill floor, but is currently stronger than other roles at equal skill once the floor is met
- all these factors can vary by patch - somewhat like a mini mmr reset
I agree that - given enough playtime - a well designed sbmm will place you at or probably within a tier of your proper rank. But all the above factors can influence to keep someone delayed from achieving that. So players who put limited time into a game and run into various temporary roadblocks can easily be left with inflated/deflated rank - especially given the last point I brought up.
This is not meant as some attack on sbmm or to say it's wrong. Just to point out some of the issues that make it far more chaotic and troublesome for matchmakers to do their job well on a consistent basis. While one individual instance or factor may not be much on it's own, they add up via ripple effect. Sort of like zerglings. One isn't a problem. A whole bunch all crashing the same pylon makes for a bad day for protoss.
A few more I forgot to add:
ping - some fluctuate more
time of day - who is playing
drunk/sober
at work vs at home (yes, had guy stop in middle of HOTS 5v5 for 10 min cuz boss walked over to check on him)
Great video, I think 1 thing is missing tho : in team games, some games are more "carryable" than others. In LOL you can get fed and 1v3 in a game carrying it to the end.
But in the case of balanced/fullteam games like HOTS, you can almost never win a 2v1. Thats also why you can carry easier in LOL or more in FPS than games like HOTS.
There are 1 and 1/2 things that you didn't touch on.
For the half: While you did talk about wide matches, and you did bring up how a game with 5 closely skilled players could be favored over a team with a carry and some noobs, what needs to also be heavily considered is that in MOBAs, and other role based hero games, like OW, each role has a different amount of direct impact on a game.
Pairing a GM Support with Diamond DPS and a Tank against a full squad of Masters, probably isn't winnable for the GM, as his character's kit doesn't usually have the ability to carry a team. And this is especially true in games where certain roles depend on other roles for help. In Overwatch 2, the Tank role is the strongest role, without a doubt, but is entirely dependent on Supports to enable them to do their job. If they don't get the minimum support that they need for their role to function, then it will also be like the Tank is a noob.
As for the 1: This possibly could have been included under the umbrella of engagement based matchmaking, but not really. Blizzard has a patent for Overwatch, and you can look this up, it is 100% legit, that states that newer players may be matched together with veteran players that have generally paid for skins, with the intention of getting the newer players to also buy those skins.
This can go as far as pairing an inferior player with a superior player that has paid for skins, so that the superior player can look cool and awesome to the inferior player.
And furthermore, there is also a patant that says that once a player buys a skin, they will be rewarded and made to feel good, get a dopamine rush, by the matchmaker attempting to guarantee their games will be wins, when using their new skin, to encourage them to buy more.
that part about engagement based matchmaking is so far beyond abominable.... Yay capitalism...
Depends a lot on how far in GM, diamond, and master you were talking about, but mostly a GM support can absolutely hard carry a game like that. As long as it's not a Mercy, Ana, or Brig otp, it's not a stretch. You can currently hard carry OW in any role. It's much more a matter of hero pool within a role, like how most low mobility heroes are worthless in the face of proper tank gap.
You also quoted the patent slightly wrong. Both things you mentioned are from the same patent. It does not say anything about giving easy games after a purchase. It talks about putting the player into a game where the newly purchased item can be used effectively, heavily implying game play affecting items.
From US20160001181A1:
"For example, if the in-game item is a weapon (e.g., an accurate and powerful sniper rifle), the second gameplay session may be selected because the weapon is highly effective in the second gameplay session. Doing so may encourage the first player to make subsequent purchases due to the satisfaction of effectively using the in-game item during the second gameplay session."
Much of what's in the patent follows schemes used in mobile games mobile games. It's also not Blizzard's patent or ABK's, but specifically Activision's. Considering it's timing, it's largely thought to have been specifically planned for or created during CoD Mobile's development.
Good video! Btw Grubby, these sorts of "meta talks about gaming" videos are my favourite. Ones that discuss gaming and the gaming industry in general. Similar to your video on ranking games by complexity/depth. I find these far more insightful than a deep dive into any particular game.
Here's a league story:
League came up with the swarm mode a while ago, and when you first enter the match making, you find a game really fast
But after you play for a few games, and you win all of them because it's too easy and no one loses in that mode, your que will go up to 15 20 minutes
Why? Because the system is trying to make you lost but it can't find anyone with low win rate because everyone is winning
And you know it's not that no one is playing, cuz if you simply log in to another account you can play normally again
Or it's because you're in a higher tier of rating than others, rather than the system failing at trying to find bad players which are plentiful.
That's a well known matchmaking issue when it's for something new, and it works on the other accounts because they have lower rating and thus more matches.
System isn't trying to make you lose, you just went ahead faster than the rest of the playerbase because you won them all and kept playing.
@@matirion lmao you think literally anyone is low rating on the swarm mode?
@@averageds3player699 Yes, some people are just not going to be at higher rankings, especially when it's new and people may not all be grinding the mode.
@@matirion tell me what the swarm is to prove you are not completely fucking ignorant on the topic you are talking about
@@averageds3player699 A pve game mode with Vampire Survivor style gameplay. Yes, some people are bad at it, and not everyone is going to grind it out, resulting in lower amounts of high performing players, especially when it just released.
The only ignorance is yours.
Just put same elo players with similar wr in the same game, matching people with 60% and 40% in the same, team just makes the experience worse, make it more competitive, not stupid lol
I'll be honest, I learned to hate the matchmaking systems.
And I long for the old times, when you simply had player-maintained lobby with servers. Unfortunately these days, if these exist, they are infested with cheats or gimmick servers and you are forced into matchmaker-made PUGs.
In team-based games even normal non-skill based matchmaker drags you towards 48-52% winrate by the virtue of carrying or being carried.
My real issue is that you, as the player, lack any actual control. In a way, MM in most games that I know is actually very punishing, especially when you consider the fact that plenty of companies monetise these minor conveniences.
You found the team you liked and enjoyed playing with, or had this once in a million match where the team worked like a clockwork? The matchmaker says "F-you, random monkey team it is".
A map you love and want to play on it again? MM says "F-you, the boring-ass map you hate it is".
Another thing is confirmation bias - even 50% winrate looks like losing more. We have a binary system - win or lose. Our minds are wired to consider a negative result as an anomaly. This means we disregard the wins, we treat them as a norm. Losses become vastly more visible.
This becomes incredibly annoying - in team based games, you often feel punished with a loss for no fault of your own.
And it tends to create landslide victories or losses.
In solo, skill based MM games, you are also often in a state of flux. You do well, your ELO goes up, MM decides to match you higher and you get curb-stomped. You drop and now you curb-stomp the newbies. Actual CLOSE, even matches are extremely rare from what I noticed. Again, it is an equivalent of landslide victories/losses plague in team games. Statistically, you are doing even - but actual individual matches are rarely close or fair.
Skill based in team games is another can of worms as well, because what metric do you use? Average is probably the most obvious but also the worst - because it will be punishing good players with monkey teams. Players rated 10, 1, 1, 1, 2 have the same average as 3, 3, 3, 3, 3. So which team will win? Their average is identical but one team has consistent skill level and the other puts everything on shoulders of one player. Team 1 is going to have a frustrating experience - 1-rated guys will have a feeling of being overwhelmed by the enemy and feel inferior to obviously superior ally, while the 10 ranked guy will not be happy with a bunch of dead-weights.
your last point is not that difficult to tackle, they literally have a massive amount of data that they can analyze how intra skill imbalance effect the outcome.
@@snuffeldjuret It is more more difficult than you might expect - I am considering only quality of matches here - it is just rambling is the vacuum - but speed at which the matchmaking works is also a factor. Nobody will care if MM gives you a perfectly fine matches, if you have to wait 10 minutes for it to "cook". And if the match itself is short, people will quickly get annoyed.
It gets even more complicated if you need to factor in people's connection quality for that matter.
@@hideshisface1886 I literally dealt with that problem when I coded a pickup bot for my dota2 community.
@@hideshisface1886 I 100% agree with you on your initial last point. A matchmaking system which puts a 10 with 4 1's to "even out" the game is not fair. The 10 should have an equal chance of getting 3s and 1s. The 10 should have a higher win rate, while everyone else has a roughly 40-45 win rate because they are just as likely to be on red team or blue team. If you think about it, a game where the only strong player in the matchmaking pool DOES NOT have a higher than 50% win rate literally means the GAMES are unfair, artificially made more difficult for him to "counter balance" his skill. Imagine if in the Olympics, they added more weight to the fastest skiers to make the results always more "50-50", would that be considered fair?
A match maker should only care about finding players of closest skill level, and NOT the result of the game being 50/50, which is what the last chart in the video showed. A GM should not be artificially given weaker teammates to artificially give the game a 50% win rate. Rather, he should be randomly assigned to one side. He will be more likely to win, naturally, as he is better than the rest of the lobby. However, it is still fair for the other players, as they are just as likely to be on the GM's team as anyone else. Overall it would make the games more fair. "50-50" isn't the same thing as "fair"
Thank you grubby , amazing content, you clarified everything so well.
I will be less tilted now if my team goes 0/15 in 5min.
I need to carry somehow without getting tilted.😂
I learned alot.
Thanks again.
Actually those games were winnable,
But the thing is my teammates tried new champions for first time.
Matchmaking expected them to perform well.
I'm sorry, but a fair match system does not lead to addiction and larger purchases per game. So modern match systems are not made to be fair, but made to make players spends as much time as possible on climbing without leaving the game. Nothing personal, just business and your own ego that does not allow you to leave a toxic relationship with the game.
Oh my god finally someone big said what I have been saying since forever. I remember when I played Warcraft 2 online and not having such a strict ladder was so much fun because you could really feel how much you improving instead of just seeing a meaningless number online.
The dev is another gaslighter who defends the company's practice of infinite hamsterwheel. Nothing of what he said is true and can be tested by anyone who had experience playing in early 2010's.
Heroes of the Storm came out in 2015
Grubby does the best lectures I swear, I got so into the upgrade mega explanation and I don't play WC3 anymore.
As a Data Scientist and Programmer here is my take: There are many reasons the whole MMR system is stupid. Note: Not the matchmaker which finds games. The Rating System is what's fkd.
1. it doesn't bias winning, meaning someone who has played another similar game (like DOTA or LOL) with lots of transferable skill still needs to play hundreds of games vs potatoes before the system eventually finds real players for him. What's going to happen? Good players will just leave the game because why would someone good want to play 100 matches to finally play some good games???
2. It encourages smurfing because it takes soo long to climb you can abuse dozens of players before the system climbs you out.
3. It doesn't give you the opportunity to learn. Let's face it, the idea of playing with equally skilled players sounds good, but that's also the reason why most people learn bad habits they never break. Because they never get punished for their mistakes since they tend to play vs "equally-skilled" players.
4. Climbing in MMR is directly counter to what the system rewards. If you want to get better at the game you need to try new things, push limits and play against better players. If the game punishes all of those by me losing MMR then when am I going to learn. This is why most people in Platinum and Diamond never grow.
SOLUTION:
1. Most *INTELLIGENT* mmr systems (like chess or go) use a self-determined ? rating (meaning you say "I think I'm Diamond 2") and it lets you play a few games, if your winrate is
Grubby, I first heard of you when you were doing your dota climb… now I had urges to replay Warcraft 3 and UA-cam starts showing me all your videos hahaha. Thanks for confirming I should replay it.
Also, a great video!
The issue I have is that a single MMR rating that is permanently on is extremely detrimental to the gameplay experience. Forcing everyone to play optimally to preserve their rating kills off the posibility for finding the fun in suboptimal play. Most games purport to have a casual mode but the reality is that it's not casual. You still have a rating, it's just not shown to you. It's ranked lite, and it sucks the fun out of games.
It also condenses your entire play pattern to a mean value, which is throwing out alot of data. We're not static points on a line. We're lines on a line. And sometimes we're multiple different lines on a line.
Another thing is the problematic part of systems based on Elo: a new player is assumed to be of "Median" skill. Which is blatantly false in online games. In tournament context, most players will not go to tournament until they've played in casual (truely casual) settings a fair bit. With online games this is not the case. Being thrown in the deep end with nothing more than a extremely barebones tutorial is just bad for everyone involved.
Also, having a matchmaking rating and a unrelated ranking system just doesn't make sense, and the two heavily fight each other.
But lastly, the biggest problem is that losing isn't fun, and on top of that you're punished for it. You get less rewards, your number goes down, and it's just all around a negativity spike. In a sense, losing a game is a service you provide. You lose so someone else can win. While being rewarded for losing can quickly lead to people intentionally tanking games, it is a mechanism that has to be investigated nonetheless. Losing has to feel as good as winning.
So in conclusion, skill based matchmaking in isolation isn't bad in theory. In practise, it has a lot of pitfalls that seemingly have yet to be solved. We have yet to find a new-player appropriate matchmaking system, it makes every game a high-stakes game when games are often at their best when they're low-stakes, it contradicts having a more formal ranked system, and IMO it's more important to find ways to make losing fun rather than get everyone to a win half the time.
Yes, exactly. Once I ranked up I felt like I could never go back to my worse race/faction/build/etc as it'd undo all my hardwon slightly over 50% grind up. And the whole reason I stopped playing SC2 was every single match felt stressful. Dunno if it was me aging out or the MMR, but I spent years playing tons of CS and SC1, win/lose whatever, games were fun. SC2 multiplayer was stress. Which is fine if you're trying to do like an occasional tourney, but I no longer wanted to log in daily and bust out a few hours of games.
if your enjoyment in a game comes from only winning or losing, not what you do within the game, you may want to try changing your mindset
@@ARandomClown It would be a lot easier to do that if the game systems weren't designed to reinforce that mindset.
The only reason why it would "force everyone to play optimally" is if you enforce the mentality of "forcing everyone to play optimally"... in a casual mode, where the rank is not visible, then... trying to "preserve your mmr" is irrelevant.
"But lastly, the biggest problem is that losing isn't fun, and on top of that you're punished for it. You get less rewards, your number goes down, and it's just all around a negativity spike. In a sense, losing a game is a service you provide. You lose so someone else can win. While being rewarded for losing can quickly lead to people intentionally tanking games, it is a mechanism that has to be investigated nonetheless. Losing has to feel as good as winning. "
THIS!!! is a YOU problem...
And most people, even if they complain, would rather have skillbased matchmaking than not.
Imagine playing... let's say, Age of Empires... Let's imagine they didn't have a skillbased matchmaking system, and you just start a game... you are a mid level player... most games for you will either be a stomp, or you will get stomped...
The low level player will almost always be stomped... and the high level player will almost always stomp... that's not fun either... even when playing casually, that stops being fun really quickly...
The pitfalls you mention are rather player attitudes that have evolved, rather than intrinsic properties of the matchmaking systems, especially because games having various matchmaking systems are not complained about like that.
Chess (which most systems have been based around in one form or another) has had these systems for a long time, and doesn't have that toxic culture itself... and the ranking system is used heavily so you can find interesting chess games. It's not particularly fun playing against someone that is far worse than you, because you can counter their moves without even thinking, in a game about thinking.
It's not going to be easy to find ways to make losing fun... since winning is kind of the reason why competitive games are fun. There are party games that have been made that can be fun to lose in, because you end up engaging with the game while you have lost, until the final winner has been declared, but that doesn't make it more "fun" losing... regardless of that, those party games that are wacky and random that makes it less frustrating to lose, those... they have a tendency to be... well... not really played en masse.
The Battle Royale genre kind of did something like that. If you lose, you can start again very quickly, and the whole looting part of it is also a fun part, so even if you die, it doesn't feel like wasted time, and due to the random nature, you will inevitably get high in some games, regardless of your skill level. It gives engagement from the feeling of "I almost won!", and since there are so many teams/players, and the likelihood of winning is lower, then losing at the higher levels can be quite engaging.
But that doesn't really work for a MOBA, does it? Or a general team shooter? Or a 1v1 fighting game?
To make "losing fun" you have to make a significantly different game...
@@billtodd2194 I think the primary reason for your "stress" is not the numbers, or anything... it's rather a change in attitude.
Totally agree with the video! One small thing to clarify though, when you mention flaming teammates for playing off meta picks (specifically in MOBAs), it's usually because players are at their peak elo on a specific role/champ. I do not mind having a long time off meta player on my team, but it does tilt me when my Janna OTP first times their AP Shaco top lane because they watched one too many Pink Ward montages.
"but it does tilt me when..." that's a you problem then. You cannot control what your team is doing, you can only control your own reactions. Game is already hard because X person is playing "bad", you don't want to make it even harder by playing on tilt yourself.
@@Lightn0xWhy put in effort to win when your teammate plays 4fun and soft ints and griefs his way down, ruining games as he tumbles down to his deserved rank?
It's not enjoyable to have one guy sacrifice his teammates ability to have fun for his own.
In Crusade Dota 2 most of the off-meta flaming I've seen wasn't warranted. For example I played support Weaver for a while not long ago, despite it being very strong I had issues with my team when I picked it because they didn't understand it.
But I think the real issue you're saying is that they're picking a new hero they don't play well. I'd say that's a tangential issue from the meta or which heroes are actually good. It's more about specialization and learning heroes inside the ranked mode when you shouldn't be.
@@famulanrevengeance3044 You missing the point that the game always set you up to lose. Simply for profits and statistic for higher active player, but also introduce stockholm syndrome.
The more rich/succesful people actually have higher stockholm syndrome rate, than someone who always chill and feel 'enough'. To be fair, everyone goin to have their sinful pleasure onto something else.
@@michaelbuto305 I don't think so, I think that's just your perception. You must realize that you won't win all your games
I'd love to have a real ranked system like you're describing.
Problem is, I just don't have the trust in game devs these days that they aren't "cooking the books" behind the scenes and manipulating the matchmaking to create a certain experience that they've found retains players the best.
Idk how exactly to fix a trust issue, but I'd start by demanding total transparency of how the system works and see where things go.
Cooking the books could definitely happen. I think it's really common in Gacha p2w mobile games, and AAA game devs may take note, when infested with money hungry stockholder lap dogs. I think many devs still rail against that ethic, though
@GrubbyTalks even AAA commonly use scummy tactics like shadow pools or weighted pools to improve matchmaking in certain pools/regions which hurts people individually.
Grubby is so innocent, surely, no game would ever secretly prioritize players who spend more money
Matchmaking is garbage. Back in my day we just played games for fun. Now the only reason people play games is sweat, meta, ego stroking. We can't just goof off and have fun because everyone is in matchmaking.... ffs
It's impossible to just dick around and have fun when you're constantly being nagged at about your lifetime performance...
Seriously, right?
This is why I still play on community TF2 servers to this day… I have literally no interest in being “ranked” in an FPS game (or any other type of game, for that matter)
I just want to hop online and shoot at some people for fun. I can spend hours just “scrimming” and enjoying the experience without stressing about anything 🤷
I'd love to see more experimentation with PvP multiplayer that doesn't care about skill. Most games are made to work around this model, but I'd like to see some that are more experimental in nature. Imagine if Starcraft didn't care who won, you could add a lot crazier systems in because strict balance is less important. Then you could tie meta progression to games played instead of ladder advancement through winrate, and suddenly you have a system where you can introduce new unlocks that everyone gets to partake in.
EOMM is definitely being used in AAA games, one example being TrueSkill 2.0 which Microsoft uses for halo infinite. TS2 can predict which team will win a given match with 68% accuracy which if you think about for 3 seconds you’ll recognize why that’s a bad thing. If two teams are of equal skill then you should not be able to predict who will win because they should both have a 50% win rate. I recommend watching Favyn’s video on sbmm because it coincides with a lot of the ideas in this video but from a fps perspective.
But the hability to predict match outcome doesn't mean it tries to screw with players, it just means it is good at evaluating their respective skill level. There's nothing that garantees there will always be enough players available to create perfectly balanced matches.
@@theslay66 That's the thing about Engagement Optimized Matchmaking - it doesn't care if the match is perfectly balanced, or even close to fair. It just cares what the chances of you quitting after the match is over are. The longer it keeps you playing, the better it believes to have done its job, so if you're a player who tends to keep playing on tilt when presented with a 10 game loss streak, the algorithm is actually going to _favor_ putting you in that position. Anything to get you to play for longer.
@@scrapbotcommander And like explained in the video, this argument doesn't hold.
If you have fun, you play. If you don't, you quit. Maximizing engagement is just about maximizing fun for the players, and having balanced matches is generally what is overall better for that.
Nobody plays more when they are presented with a long loss streak. And even if there are, they are outliers, not your average player. There's no sense in building your system around the behavior of a minority.
As there is no sense building a system with purposefully unbalanced matches to screw with the players. There's no need for that. Randomness of available player, and a good ranking system, is all you need. This kind of streaks will naturally emerge from the system. Random is random, streaks will happen.
Seriously, most people complaining about how the matchmaker makes them lose on purpose are just biased. We tend to attribute success on ourselves, and failures to external causes, that's natural.
But why the system would target any player specifically ? It must work with all players simultanously, while working with a limited pool of available players at any time. Making a system that try to decide in advance who should win and who should lose is just senseless, and far more difficult than simply trying to make balanced matches.
The rest is just speculation, complotism and persecution syndrom.
Eomm straight up rig the games using aim assist aim resist, the strong players all have aim resist, except when they are meant to win the game for the noobs on their team, the noobs all have aim assist, the lobbies are rigged to generate dopamine in the noobs, get them addicted to being better than they are, making them believe they are good, and thus spending money, rewarding the company for giving them a fantastic gaming experience where they were made to be the hero. It's somewhat of a twisted business model but it is the best business model you can have in a shooter environment, it's what makes the most money, it preys on the weak. So there is two ladders juxtaposed, one real, that is hidden, one manipulated in the open where they make people who spends climb and the people who don't stall, with the caveat that depending on who you are they may model their abuse of you based on you.
Other way to look at it, imagine you have a garden, the weeds are the good players, they perform well in the environment, and the noobs are your fruits and veggies that needs nurturing but produce food. So you use the weeds as biomass to fuel your production, perfectly logical.
It's perfectly fine in nature, not so much in a society.
Tl:dr: eomm is a con.
@@Deucely I mean, the matchmaking system can't control the presence or absence of aim assist, it just pairs groups of players together. That isn't the problem with EOMM, the problem is that it will sometimes match you with an opponent of higher or lower skill than you, like matching a platinum player with a diamond player.
In the case of team games, that would appear as one team's average rank being somewhat higher than the other team's average rank. Not enough to be blatant, but enough to skew the results in favor of one team, by maybe a 60%/40% chance. You should still absolutely take responsibility for your own actions as a player.
It also won't do it constantly, @theslay66 is correct that this would make most players quit, and my example was admittedly absurd. But the problem is that EOMM may sometimes decide that skewing a match is the best thing to do based on how much it values the outcome of a game for a specific player.
And yes, unlike SBMM, EOMM does actually value using the predicted outcome as a metric for optimizing player engagement. If you can predict the chances of a match's outcome, you can control how often a person will win or lose. You can break up lose streaks by handing out pity matches with slightly lower MMR opponents at key timings designed to keep you from quitting in frustration, then throw in a close game after feeding an easy win so you still feel challenged as a player, then throw in a clear losing scenario so someone else can have their pity match as well once the system thinks you can tolerate it without calling it for the day.
That does not mean it is necessarily a fun or fair method to use. While fun always creates engagement, it is not the _only_ source of engagement. (I would not consider getting flamed for 30 minutes straight in a losing match of LoL to be fun, for instance.) And EOMM, sadly, does not discriminate: To it, engagement is engagement, no matter what that looks like in practice.
That, however, also _doesn't_ mean the algorithm has some personal vendetta against you specifically, or that it will always disrupt a streak. That's not true either, it's just trying to maximize the average engagement among all players. And sometimes that just ends up involving unfair matches... the only time it doesn't, it matches SBMM anyways, and in my opinion SBMM seems to be to just be a better system overall, at least for 1v1.
it's really hard to balance any game, so that you both have a clear progression of strength that you feel stronger all the time, and also aren't bored by the lack of challenge once you've become strong
12:00 - People that defend 50% completely missing the point every time. It's not about "we can't get to where we belong". it's about: "Why do I need to play 300 games instead of 100 to get where I belong?". LoL players would I understand what I am talking about. For everyone else, imagine if Happy/Sok/Lyn/Hawk would've lost 3 games out 10 when climbing through adept/diamond leagues towards the grandmaster.
I was stuck on a low elo on leage recently. well am but not played lately. Either you completely stomp your lane win and go to the next game and get a free lose and get right back to where you were. Or you completely stomp your lane and still lose because too many of your team mates are completely uncarryable. even if the other players get fed they still manage to throw. there comes a point where it is so rigged it stops being fun. why do you need to grind through 100s of games like that just to climb one division closer to where you were.
@lightlayagajoie5739 then why do smurfs achieve 80 to 90% winrates? You just aren't good enough to climb fast
@@elvis4868Smurf is basically a skilled player that plays on newb matchmaking
*Obviously the skilled player will win evertime against someone that's entirely new*
@@lightlayagajoie5739 There is a solution, which is to be less secretive about mmr and perceived skill.
Tell the people playing who's getting matched up and who's getting matched down. The expectation of being matched with equals breeds resentment and toxicity when you get people way below your skill.
It also allows the game to be more transparent and say that you get a big mmr boost for pulling through and winning this game despite the odds, while you only get a small hit if you lose. Vice versa as well.
Sure, those games will still suck, but being clear about that helps players rationalize wins/losses rather than bemoaning "elo hell"
@@antarath517 the problem is not only lack of transparency but also that some games are clearly intended to be predetermined to be a win or a lose. because somehow that makes it more addictive I think. If I win one game I don't need to be tested if I can still win against 2 smurfs and 3 useless allies. yes sure I'm good but I'm not that good, or whatever. This is just another game you have to sit out before you can continue to climb.
The problem with MMR that you have ignored is that you dont feel progress. The only thing you have to go on is the rank up because the matches are overtuned.
The sensation of being good cannot be felt because your opponent is always too close. You never get matches against 'your old self' to showcase how far you have come, you are constantly locked in cuthroat matches and can never relax and try new things.
In 1v1 it's nonsense, but in team based games there is a nugget of truth to it. Not in that it matches you against harder opponents - that makes sense. The problem is when the matchmaking does things like put you in a group where people main the same role (e.g. 3-5x jungle mains) so that people on your team are forced to off-role, just because the people in your team are on win streaks. While the other team is populated by an even spread of main-rolers (e.g. top/mid/bottom/jungle), thus inflating their chance of victory.
I'll watch and see if you're talking about the same thing now : D
That is a real problem, but it's not really related to the win streak idea. Matchmaking rating in team games is just noisier, because there's so many things that are outside of your control during a game with 10 people compared to 2 people. Over a theoretical infinite game sample, you should still arrive at your true MMR eventually. But over a realistic sample of 50 games or 100 games, you could easily win or lose an extra 10 games due to this sort of what-random-teammates-did-I-get-today thing, which makes it much harder for the matchmaker algorithm to understand what your MMR ought to be. Role selection adds another layer of complexity on top of that: For example in Starcraft, it's very reasonable to imagine your "true" MMR with Terran and with Protoss may not be the same, but at least there the matchmaker knows which race you chose to queue with and can adjust appropriately. In DotA, maybe your "true" MMR as main-carry is significantly different than with a support hero, or on a finer detail maybe there's only 2 specific heros out of 100-however-many that you have learned to play efficiently, and the matchmaker can't really know in advance which hero you will select this game, or whether it will mesh will with the limited hero set of the rest of your team. Arguably having a wide selection of potential heros is itself a skill that makes you better, but all of this adds noise to the matchmaker.
@@Kerostasis I don't think you quite understood. Imagine if in SC2 you had to play 3v3, and each player on each team must play a different race. Then one team gets a terran main, a protoss main, and a zerg main. While the other team gets three zerg mains and two of them have to pick protoss/terran. Effectively basically deciding the victors of the game before it even started. In a sane world, each team would get two zerg players so one player on each team had to offrace.
I mean, if you're going into a moba solo queue (without role-based queuing) and you can only play one role well...that's a skill issue. Your skillset is limited, so you're going have a rough time when a change of plans is called for.
@@thekingofcheese9005 You don't get it. You select your role when you queue. But it puts main rolers of diverse roles on one team, while the other team gets stacked by redundant main rolers. We're not talking about having to offrole, we're talking about one team having to offrole massively more than the other team (and the matchmaker knows the main role of every player).
Finished watching the video now. Grubby didn't really touch on what I was talking about. But he got into the periphery regarding team games with the 1 champ + 4 mid rankers vs 5 high rankers. To add to that, there's more complexity than you'd think. For example, in this matchup the champion player is obviously intended to hard carry, that's the only way it can be fair. But if the champion is forced to pick his hero before his direct opponent, it means he can be hardcounter-picked which makes that extremely difficult when you can't rely on your teammates to cover for you. Stuff like that.
As usual, a reasonable take from the most level-headed guy around. Looking at the history of matchmaking, I love the "psychological arms race against ego damage" the devs have been at for decades.
That being said, the difficulties with team matchmaking are the second biggest reason (after game length) why I left MOBAs and got into RTS and later into fighting games. It gets easier to see rank for what it is, and a combination of ranked and casual games (with looser MMR matching) has been the best way for me to feel the improvement, and enjoy the journey.
So essentially the result of a system that has matchmaking (mmr,elo) is that everyone will have 50% win rate, but they will be assigned a different badge of honor (silver, gold, etc) and the result of a system that has no matchmaking is that your badge of honor will be your win rate - lower skilled players will have lower win rate, and the best - the highest.
Given that overview which do you like more? All your games are a coin flip but you get a different colored thing in game, or winning/losing proportional to your skill? I would take the second. Being one of the best players, but still getting 50% win rate is not something I would consider fun. Why did I become better after so much hard work if I am not going to win more? And losing when you are the worst should be something people understand as normal and natural. Something to overcome, not to avoid or mask with a 50% win rate.
If I perform badly I want to lose, If I perform good I want to win. And I am talking about actually winning games, not winning more mmr points.
I have played a game with no rating and it was damn fun progressing from the bottom to someone that can kill almost everyone. As opposed to a different game with mmr that always feels medium levels of rewarding no matter how much you improve.
There are other considerations: if you play at your skill level you MUST always use the best performing strategies/other best choices or you risk performing badly. It ruins fun even more by making things stale. Not to mention that every game is the hardest. There are no breaks. I don't want to do 100% effort 100% of the time. It's a video game, not life or death. Some times I want to chill with something goofy. The most fun I had is when I was a good player, but used shitty, weak, fun strategies against worse players without sacrificing winning. Do that when you are in an mmr game at your skill level and you will have a bad time. Don't tell me that is OK. It ruins diversity and fun.
But there is something even worse than this topic, it is when a game is a team game (like league or dota). That is truly something that would ruin your fun. You know if you played. I avoid games like that now.
Man I feel this so hard. Moving through the sludge of modern mmr systems is slow and unrewarding. You always feel like you're getting nowhere fast even if you know logically you are in fact getting better.
You're right too - that winning more is literally what we're trying to do when we improve at something competitive. When you rob people of that catharsis is when they start getting frustrated and quitting, blaming "elo hell" and other things, with most people not even realizing what's happening or that they are actually improving. In reality it's a system that feels stagnant and is actively causing the frustration.
If you are scared of losing and need to have your hand held and maintain a 50% win rate no matter what to stay engaged with the game then i'm sorry but you are not a competitive person at heart. Unfortunately capitalism could never allow for such an unexploited demographic and leave potential customers on the table however.
@@coreyrachar9694 Yep, it's capitalism at the end of the day.
@@coreyrachar9694 i mean... you want to have a really high win rate that comes from smashing noobs and people you KNOW youre better than, and you claim to be competitive? thats just jerkin yourself off
@coreyrachar9694 tf? Poor players in a sbmm would eventually lose at below 50% winrate until they are matched against other equally poor players, then if they improve they will win more and climb mmr until they reach a higher skill lvl whats the problem with that? If your motivation to improve is to smurf on worse players then thats just incredibly selfish, i play badminton and when much better players play agaisnt weaker oppos they know to chill cuz otherwise its just bullying their opponent
If performing at your best to have a 50% chance of winning isnt enjoyable to you then yes sbmm is not for you, but some people to enjoy the challenge and strive to get better still as they can find enjoyment in improving game to game to climb to a higher rank, you want to be allowed to destroy people worse than you which would actually kill the game as the worse players wont even be able to have a game experience to learn and improve
THANK YOU GRUBBY! I had a friend who complained so much about 50% winrate being BS and as someone who has studied statistics I found his arguments infuriating. Especially for 1v1 games, which we was playing, when you drill down into their arguments it is all about 'oh I think I am good and I want to win more.' It is never from the perspective that finding challenging opponents is actually a virtue of a good system and the quickest way to improve. No, it's just 'I want to feel good and curb stomp noobs' deep down.
There's a quote on the wall of my dojo that says, "Your opponent is your greatest teacher. Your ego is your greatest enemy." People would hate it but that should be on the loading screen of all competitive games.
That last example of grandmaster with diamond teammates is exactly why I stopped playing HOTS and Overwatch. I put in a bunch of time and learning only to get punished for slowly building up my skills. It's so exhausting and frustrating knowing that if you don't HARD CARRY, you WILL lose badly and it just gets worse over time because no one wants to be a perfect robot tryharding every single game just to have a chance at winning.
I mean, if you were only put in games with other people as good as you, you would still have to tryhard every game to have a chance at winning, because you have to tryhard to get 50% winrate at the highest level. Thats how competitive multiplayer games work. If you dont want that, just intentionally derank to diamond and tryhard when you feel like it and throw games to derank back if you win too many times. That is called smurfing.
>ranked
>waaa waaa why do I have to tryhard
I don't know man, geez, no clue whatsoever why people care about rank and actually try to win games and won't let you rank up for free.
Once I played like 200 games of Hots over a week or 2. During that time I became a very good player, as with that many games you see the patterns develop quickly. You are most likely to lose around minute 17 when one of your teammates tries to solo a camp, doesnt watch the map and get ganked by the enemy team, and you cant recover the momentum. I would argue the fact that I understand this and my teammates dont means that the matchmaking is flawed as I am paired with teammates who are less skillful. Anyway my rank barely moved over those games and played my heart out tanking for teammates....still doesnt matter.
@@CommanderReplay its a team game. Unless you are extremely more skilled than the opponents at your rank, you won't be able to significantly carry randoms. You would need to find a team with 4 friends who are at the same level of cognitive abilities.
@@TheSuperappelflap Right I get that, but my point is, shouldnt the MMR system be placing me with people of similar abilities?
I had a thought resently, that MOBAs shoutn play with individual Hero/Champion MMR. Just like different races in WC3 have their own MMR
People that hate SBMM are all people that are slightly above average and were used to dominate in lan parties, but the moment they go online with a big player pool, they realize they're not that good.
Ie most of this comment section. Filled with delusional players who think the evil MMR is keeping them in silver.
You should start an e-sport team, you are extremely good at viewing learning in a good perspective
Before Apex legends had ranked mode I managed to win 15 games straight as a solo player (20 teams with 3 players in each game). The fun stopped when I played ranked, and I quit shortly there after.
You were probably a top 90% ELO player. The problem is, SBMM extends ELO hell to about the top 95% ELO players. You got placed in top 85% to 95% ELO lobbies after SBMM was implemented. This video fails to mention the extension of ELO hell that is caused by SBMM, which is why SBMM can work in theory but feel bad in practice.
37:45 I remember splitgate doing this. I was mid diamond and ended up playing against 3 people of the highest ranks in the game who had developer tags next to their names and I reconized them from the dev team. They also had a mid diamond on their team and my team had top 100 players on it too. I'd also get paired as a mid diamond with a team of silvers to fight a team of golds with me being the outlier. So this ends up happening relatively frequently. I had to carry lots of gems because I was the highest rank in my team and it was a glorified 1v1 between 2 diamond ranks and a bunch of lower ranked players.
Quake Champions literally just putting the first 8 players in the queue into a game is way better than anything Blizzard has ever coded.
Well, it's easier in FPS games, especially in a game like Quake where one person can literally carry 1v10 if he's so much better.
Can't really do that in Counter Strike for example, just because of how the match, movement, weapons and maps are designed...
Games like dota and lol are especially annoying because of how you can "feed" the enemy. You choose an "isolated" lane. Consider a scenario where, let's say you're the worst in your team and you go into a lane vs the enemy team best player. Your best player will get matched with someone "mediocore" meanwhile in their lane. You're probably going to make the game so much harder for your team because of how easy your enemy will snowball, enabling his rotation and plays through "experience" and "gold" gained in your lane. On average this scenario doesn't matter much, but it will happen pretty often, the opposite scenario will happen too, where you dominate your "opponent" and then lose the game anyway, and we're all ready to blame the rest.
@@grimonce So basically you are arguing that you can't solve bad game design with match making?
Overwatch was supposed to be an FPS but the hero design totally ruined it, even though that was supposed to be the main selling point.
CS seems a bit more feasible, I have barely played that game but brought back a round from 1v4 once.
LoL in particular just really sounds like a snowball fest, idk why you'd want to play that game.
But I can really only compare OW and Quake from the list... Also my experience in CS was very cheater heavy.
Id agree but the playerbase in QC is not great, and half the games are stomps and leavers and people complaining about the sbmm which is quite ironic.
Those eight players are the entire playerbase in Quake Champions
@@PepsiMagt It's still less dead than Concord.
I couldn't agree more, just watching the title made me excited.
This was the most "You didn't actually watch the video" video of all time based on peoples comments saying they agree that SBMM is bad. Never seen so many people get fooled
Yea, there are a lot of enormous dumbasses out there...
Ah yes let me watch a 40 minute video on something stupid
@@Binzobdidn't read the title before u clicked or something? It's an informational video nothing stupid about it. Maybe look internally 😊
@@nicholasmccune7949 look internally cause I mindlessly clicked on a video that is 10x too long? You okay yourself?
@@Binzobwhy on Earth are you even commenting here?
My boy grubby did a great video! To summarize : everyone is silver and mmr is a rat race regardless of game. the points and rewards system work! I play Marvel Snap(card game) with challenges on parallel with the ladder and it is very enjoyable. For Sweaty boys , open tournaments could be a very challenging and rewarding experience than the ladder.
The comments are already a gold mine of folks who won't actually watch the video... But please watch the video (especially you, CoD bros)
Thanks for making this video. All this video did was confirmed that the changes made in Wild Rift was the changes that the community. was asking for when it came to rank. (My TL:DR for the day. The bottom doesn't truly matter.) Note: when I say people or community I mean all of North America. Not Japan, not Europe. Strictly NA.
I remember that points were a thing Diamond and up. The first season I've experienced this was in fact the last season for me. I was stuck in low emerald for the second season in a row. I've played with a friend that was low platinum. 3 straight games in a platinum room that I got wins at, and the system decided to put me in diamond for the first time. Cool. At that time, many people were either climbing or dropping with an average of +10/-15 rating. Riot decides to fix this mid season before the dominoes fell into place later on.
Points were replaced with marks, and Riot made it easier to play with friends in rank regardless of rank. Autofill was already a thing during pc helping to guarantee faster matchmaking. It's still being strongly adapted to this day. So unless you're in legendary queue you now have a system that rewards strictly based off of wins and losses with the performances being measured towards getting your next shield mark protection. Because autofill is there the community now expects you to know how to play all roles really well, and if you can't people are just going to surrender until they're able to find a team that can play perfectly. If the other side is being totally dominated they're going to surrender since they're on the other side of the table. The community enforces this, because most is fighting to be a 1v9 carry for the team.
Whether you want to admit this or not these current systems only benefit the casual playerbase. They also benefit the lazy playerbase: those that don't to take responsibility and would rather buffed and nerf everything to their liking just so that can finally have a shot at winning. This game is for them. After all, Riot themselves wanted the entire playerbase to be at a diamond level. So now we're at a point that since the amount of players at the lower ranks are way too small everyone between iron and gold are now considered one group to make matchmaking easier. That's what happens when it comes to boosting.
While I was on the fence about whether the forced 50% was real this has basically cleared my suspicions. Thank you. The develop.ers are at fault for making the system that some people hate. At the same time, some have been asking for this. Do NOT put all the blame on the developers
If you "can't make it out of silver," then you're a silver player
It only means you dont have friends for group match.
@Glory264 incorrect. And if you actually watched the video, you knew why, too.
@@zerg0s correct, you can climb up pretty quickly in a stack, and then play stably in this rank even in solo. if you're in the broader part of a bell curve, you can be placed whereever and play pretty comfortable.
@@zerg0s If you got a skills and you got a team, you can easily cheat a system and get 55-70% up to 80%. This system sucks but devs are loving it because it forces your engagement. More engagement = more $$$.
@@MrSam1804 you climb faster up until the natural elo of the worst member of your duo/group. Then you climb slower, because that player just became an anchor for the others.
And at that point, you are still a silver player. You're just playing in gold because you got carried and you'll slowly sink back down over time.
Elo kinda sucks because it needs 200+ games to place you (as the video points out), so if you play less games than that per season you're just kinda adrift.
An excellent analysis and explanation of how mmr systems actually work over extended periods of time. While listening to you I caught my past self getting emotional when losing a match after a pride-filled winstreak because of griefing / trolling / whatever. It feels bad, but like you said on the other hand, it isn't nearly as annoying when such a case happens on the enemy team. I feel like I've gained something by watching this video. Thank you!