I'll save you three hours. 1. Kramnik repeatedly says he's not accusing anyone specific of cheating like HIKARU for example, then goes on to discuss Hikaru's games. 2. Eric repeatedly makes solid points refuting each of Kramnik's points, Eric making the case that each of the aspects of success a guy like Hikaru has can be clearly explained. At which Kramnik says, "Yes, sure of course," and then keeps concluding what he already thinks. You're welcome
that's mischaracterizing what he says so much... he literally said that hikaru's streaks show that something is wrong: A) hikaru is cheating, B) his opponents are cheating (in other games, so their rating is inflated), and he says explicitly that "B" is much more likely the case kraminik's communication's skills are trash tier, but if you really pay attention to what he is _trying_ to say, it's very reasonable
not really. he has no data, you are just trying to intuit if it sounds reasonable or not, his data is that hikaru overperforms in blitz matches, no shit that's to be expected from a player who is 300+ elo stronger in classical, now show the data on why hikaru specifically is an anomaly, u cant look at his data and be like "looks funny to me". @@zaftnotameni
@@Papageno123 I think there's a lot being lost in translation here, not just from english but also from cultural differences - I have quite a few eastern european friends that I could see making the same type of extremely specific insistent points about something, even the doubling down on focusing on hikaru when people start questioning is he right in doing it? no is he doing himself any favors? no is he helping cheating be solved? no is he being malicious/crazy? no
they can chop it to five episodes, 40 mins each, add intro put credits and boom they can market it as a series: PS, someone give it a title "_______" season 1 😂😂😂
Somebody should make a short version of this with his main points. I want to give Kramnik an opportunity to explain himself, but at the same time there's no way I can watch 3 and a half hours of him talking
Рік тому+17
100% Agree. I also want to give him the benefit of the doubt. But 3 hours 34 minutes? fuck that. can AI do a TL;DW type of thing?
I'm 30 minutes in and there is nothing new or specific yet. Only the claim that he and his mathematicians checked things but not exactly what they came up with. As for what anti-cheating might look like, he suggests taking a suspected cheater on with an engine.
@ Give him the benefit of the doubt? For what? lol I think he made some really good points. He don't wanna talk about the best methods public, but he suggest a variety of them, and if they all sets off alarms, may want to investigate or just instant ask person to leave etc, have it in the rules to not award such cases. Also, if people who cheat succeed, it's does affect others, it means others don't win, and they either quit or have to cheat to win.
The main point is that he has some data but he hasn't published it. At 31:00 his data shows that his opponents in lost and drawn Titled Tuesday games have better accuracy than Magnus At 44:00 he notes that some players are surprisingly good in the first 4 rounds of Titled Tuesday but bad in the last 5 rounds. At 2:07:00 he doesn't understand that Hikaru can have such a high rating when he plays all the time At 2:12:00 his data suggest that some players start cheating after losing 4 games in a row At 2:25:30 he says that someone should probably be suspended for Hikaru's winning streak. At 2:30:00 He complains that Hikaru is misrepresenting his intentions.
Kramnik: There are only four possibilities: -Cheater vs Cheater -Cheater vs Honest -Honest vs Cheater -Honest vs Honest So the probabilty that someone cheats is 75%.
cheater vs honest and honest vs cheater is the same so there are only 3 possibilities or 66% of cheating out of 3 games ASSUMING every of the 3 items shows up EXACTLY 1 time in sets of 3 games. which is, ffs, highly unlikely and obvious to anyone using basic logic. if you generously assume 10% of players are cheaters, then you can calculate more realistic odds. but saying that in 2 of the 3 games there's cheating is totally unreal. so, vladimir is starting off of completely wrong premises that comfirm his unreal odds.
watch on 1.25 or 1.5 speed, guys. Some timestamps for u guys: 23:45 what can be done against cheating 29:40 about the tests. Test on levitov chess about players performing in last critical rounds, etc Then he talks about accuracy.. i suggest you to skip it, because accuracy is not that relevant, with all the respect. it depends. Depends on what you're playing, how positional your game, how long, etc. Just rememberd how Aman won vs some 500 rated player and they both had 90+ accuracy just because Aman took all his pieces and played with his food for like 200 moves more. So, 500 rated player has played 90+ accurate and better than Magnus in average, eh) with no pieces and chances for the win though, but that accuracy Pog 45:25 another example, how often people have 4 out of 4 from first games 54:37 Kramnik suggests playing TTs only with camera on 58:10 Eric says it's still not 100% proof, Kramnik says that it should be a fight from different angles combined 1:11:10 unless someone beats Hikaru or Magnus lol, yeah need to go
i think kramnik means well and I don't think he's 'gone crazy' like i hear a lot, i just think maybe he's being consumed with confirmation bias or isn't the clearest thinker on this topic. when asked a direct and simple question he often gives a rambling answer that you have to kind of put together an answer from an ocean of words said that relate to the question. i get that english isn't his native language, but he is fluent enough that I don't think it's a good explanation. when I speak in a non native language I tend to be even more direct because I don't have the vocabulary to express more subtle points in great detail. one response i noted was kramnik saying that streaming and being on camera makes "no difference" in regards to cheating...and eric's reaction clearly showed he wanted to push back on that as well. Being a popular streamer definitely makes it less likely that one would cheat. having a humongous audience of people watching every game hikaru plays while he is on camera greatly increases the scrutiny of his games. it's not just scrubs like me that watch hikaru...titled players, GMs, super GMs are watching him and many of these people aren't the biggest fans of hikaru personally and would not hesitate for a moment to bring attention to suspicious gameplay if they saw it.
You are extremely naive on this, do you really believe that streaming and zoom can prevent someone from cheating, really? The temptation to cheat online is at least 100 fold greater than over the table. Which means that for every one top tier player that ever tried to cheat OTB there are at least 100 more that are trying their luck with it online. There are innumerable ways to cheat, even with zoom. For one this person can have an app relaying the game to a friend on a different computer which has an engine running and he could be feeding the moves via a hidden earpiece. That is just one easy way this could be done. Also the mathematics does show that cheating is going on so what exactly makes you think that Kramnik isn't the clearest thinker on this? Is it a matter of magnitude which you think Kramnik is overestimating? Even so, would it not still be a problem, even a serious problem that needed to be addressed? PS: Or is it that Kramnik accused an American player of cheating and you Yanks Don't like that, just like you got all defensive when Hans Neimann was accused of cheating? BTW that meteoric rise Neimann was on up until a year ago seems to have stalled quite aggressively and has gone in reverse. Fancy that, being a below average GM level player for the longest time then suddenly for a two year period start playing better than anyone in history only to have that rocket ship of progress stall and start to dip. Sure, if you believe that Neimann didn't cheat, I have a 24k gold vibrating butt plug to sell you. As for Hikaru, I can't say haven't looked at the math/statistics yet, time will tell.
Do you have a reading comprehension issue? Nowhere in the original comment does it say streaming makes cheating impossible or that it is prevented in anyway, It simply draws the clear conclusion that a bigger audience seeing the games would make doing so much harder and the games are under great scrutiny. I have no clue what your source is for the temptation to cheat online being a 100 fold but I will assume you made it up or are self reflecting. For the last comment mathematics can show ANYTHING if you pick and chose samples. Companies do this all of the time to make statements about a product. The reason why Kramnik isn't the clearest thinker on this? because he has shown repeatedly through his own comments that he does not understand online chess or statistics at all.
"titled players, GMs, super GMs are watching him" - agreed. At a different level I have heard it is amazing what can be picked out from a video stream by a professional, for example screen reflections from a pair of glasses.
@@Crypt0n1an Who got defensive when Niemann was accused of cheating? You’re acting like most of the US was on his side when in reality he was denounced by pretty much everyone possible.
Thank you chessbrah for taking the time to have a full conversation with Kramnik and giving him the opportunity to share his thoughts and concerns regarding cheating. I know it was a long video (I'm still up to 1:09 so far) but I think you are doing a public service for the benefit of the chess world.
Yes. It's clear that Kramnik thinks that anticheats are integral to the future of chess. With this as a first principle it follows that companies and different actors/entities in the chess world has to get in line with this principle for the survivability of chess. It's his entire legacy that's at stake (and everyone else prominent in the chess world) Just compare it to 80s-00s bicycle racing, where the entirety of competitors have zero credibility when it comes to fairplay.
I met Hambo once, well I say met, I was at a party and had to use the bathroom and who walked out but famous online chess personality Aman Hambleton. As he breezed past me I blurted out "hey you're the chessbrah guy!" bc in my excitement his name escaped me at that moment. But he didn't even aknowledge my existence and I had too pee so bad I could not follow up to try insist some interaction. When I got into the bathroom somebody had take the worse dump I ever experienced in my life. I don't mean there was unflushed or visable mess to be found, but the smell was absolutely god awful. I swear I had olfactory nerve damage from the short time of exposure while I peed as quickly as I could, and despite my habit to do so it was so bad I left the restroom without even considering washing and drying my hands bc the stench was so overwhelming it was very uncomfortable to breathe. I am not saying for sure it was Hambo that did the deed, I can't rule out that there was somebody that was in turn before him that cursed that bathroom that night. But I will say if it was him, then he probably has something medically wrong with him, bc although I am not professional I can think of no other reason why somebody could have such a horrible odor to their bowl movements. While my sense of smell was disrupted for about an hour or so, I still tried to find Aman at the party, but one of my friends said he had retired upstairs to an "invite only" section of the party that was being gatekept by two dudes that would probably get serious about it you tried to insist upon inviting yourself. Anyway, the moral of the story is...Don't support these guys.
He definitely makes more money then before all these accusations. People just love drama. We are greedy little monkeys that absorb all kind of drama about people we don't know because our own lives must be so lousy I guess.
Kramnik should have opened his UA-cam account and post these statistics there. He would still lose his credibility. But at least, he would make some money out of it. Right now, the only people who make money out of this drama are Hikaru and Levy.
Kramnik's entire argument basically sums up to "Online ELO and FIDE ratings aren't the same for most players, this means that people are cheating!" And he completely disregards all other variables that impact OTB and online play, including stress, tilt, the ability to select opponents, the non-standard game modes and time controls, etc. The rest of the video is Kramnik's mind boggling over the concept that FIDE rated 2200s can be rated higher than him on Chesscom, and him giving suggestions for how Chesscom can burn millions of dollars per year giving virtual cavity searches to the entire Chesscom playerbase.
You kind of missed his point. All of those things you mentioned should already be accounted for in the online rating. He's not complaining that the ratings are different, but that there are results that should be nearly statistically impossible which points to something being wrong. Either people are cheating (not necessarily Hikaru, but possibly even his opponents in other games), or the rating system is broken outright.
@@MrFreeGman They are accounted for in online ratings, which is why online ratings higher than FIDE ratings from OTB play. The results are only statistically impossible when you make incorrect and misinformed assumptions about the veracity of Chesscom ELO ratings. Of course the online chess arcade ELO system is "broken", because there's a huge variety of player skill levels, of available game modes, of various time controls, and all the other factors I previously mentioned which allow players to influence their probability of a winning outcome. As a for-instance, there's a gentleman rated 3000+ ELO on Chesscom who only plays the 10 second variant of chess. At that pace, the game ceases to be chess and becomes a contest of reaction time, mouse speed, and accuracy of motion. And it's fine that he has this rating, because Chesscom is a chess *arcade* site. You and Kramnik are making a mistake by taking their ratings so seriously.
dude..hate to tell you this but because of legalities a corporation can not go around accusing pros of cheating...but i can assure you they have a list with many names you know that are more than suspicious....and cheating IS a huge problem ..that is not dependent on your.....belief
1:24:45 Gotta give props to Kramnik at least for that bit. Agree or not with "the procedure", let's be fair. I must admit I was surprised to hear that from him.
Regarding Kramnik’s metric at about the 35min-40min mark about the accuracy of games played by himself or others vs Magnus. This seems like a misunderstanding of what “accuracy” means. It’s a % of how close one plays vs an engine. A higher % doesn’t mean more “accurate” it means more like an engine. So in theory it’s possible to play more like an engine just to game those numbers regardless of winning/losing/cheating. If both players play known lines and quickly get to an end game then the “accuracy” % will be higher. Conversely Magnus who doesn’t want to play like an engine but wants to have fun is known for playing bad openings just to mess with people and create imbalance so more likely to have games that is less engine like aka “less accurate”. (I do accept there is a lot of cheating and I’ll keep watching to see what else he says but much like the cherry picking of statistics Kramnik seems to be working off of a misunderstanding of what the numbers actually mean.)
He trying to find players that using an engine, by comparing their moves to engine moves, whats wrong with this, winning/losing percentage is a different parameter that he also takes into account
accuracy measures consistency or precision in stastics. this is how much variance moves a player makes between games. in philosophy accuracy is the reliability -a measure of the objectivity of each primise. chesscom use of the measure of accuracy is the amount of top engine moves so it changes the fundamental common uses of the tearm accuracy. what i think kramnic means when he says accuracy is its intended meaning which is is the player making common types of moves for that player in all games measured. kramnic is saying that there is a flaw with hikaru opps in that they dieviate with a low reliability measure. niw here is the problem… kramnic needs to test for reliability within group variance but instead he is using validity (logical) measures -comparisons if engine to make these claims and now you have the dreaded type 1 error (accepting the null when its false)
Could you try to explain your position in a different way? From my understanding of the situation, he's accusing quite a few titled players of using an engine. For this reason, you would compare their play to an engine's play. Therefore, utilizing accuracy should be meaningful. Also, although Magnus may play non-standard openings on rare occasion, the accuracy metric takes into account all moves and how they compare to the engine's preferred moves. While the beginning of the match may not be "accurate," the middle and end game will raise the "accuracy" as you have to play more and more precisely to gain an advantage. This is why he can still have accuracies near the lower 90s as GM Kramnik said. While knowing 10 - 15 moves in main lines will definitely contribute to accuracy, it's ultimately a small component of the measurement over the course of a 45 - 60 move game. It sounds to me like he's done what I would do, and targeted the middle and end game, where the play should be well outside of the prepared lines. Just my two cents. I'm also only 1/3rd of the way through the video, so there's that.
What? A higher accuracy doesn't mean more accurate it means more like an engine? Engines are more accurate than humans. Therefore, a higher % absolutely means more accurate. Like what? And people upvote this ignorant shit. What really gets me is that I can tell already that you go along with the saying that humans don't play like engines. I doubt that you even understand what that means. Because they evaluate positions the same way we do. At least until NNUE became a thing a few years ago as well as Leela. But a lot of engines, and Stockfish up until recently, all evaluate(d) positions based on handcrafted human features. Things that are very easily understandable by people. Material, king safety, piece square tables, space, mobility/freedom, having control over important squares, having outposts, initiative. The only thing a computer can do that we can't, is analyze extremely deep into the game. So humans, as you can imagine, will never be quite as good because it's impossible for us to look that far ahead. But we still look ahead, and we still evaluate positions the same way, some of us better than others. Your whole take and comment is just nonsense, and it amazes me this is the shit people listen to and upvote. I'll never get this world.
@@paulsmith5237 To be clear my position is that one of Kramnik’s (supposedly many) examples that he gives shows that he doesn’t understand the the meaning behind the numbers and he’s just hashing numbers and calling it many to justify the conclusion that there is rampant cheating. I accept that there is cheating but that his numbers when he does share them aren’t the smoking gun he thinks they are because there are many ways to explain them, namely being that people aren’t completely emotionless rational computers that are always trying to follow an algorithm. *** For those that are too lazy to hear Kramnik’s explanation that I think is particularly ridiculous here is a summary of what he’s saying starting at 31:00: At some point about “half a year ago” he decided to look at just the last 100 games during Titled Tuesday played by himself, Nakamura, Magus, Grishuk, Kamsky, and “several players” where they lost or ended in a draw. His average accuracy in that set of losing/drawing games was 86.3%. Of this 100, 31 in which he lost he had an accuracy of over 90%. His average rating in those 100 games was 2834. Fabiano won titled Tuesday in the past with an accuracy of 86.1 (average over the tournament? last game? he doesn’t specify). This one data point leads him to believe that his score of 86.3 is therefore good enough to possibly win Titled Tuesday. Magnus’s accuracy in his data set (though he admits that Magnus doesn’t lose much thetr he only had like 50 games to analyze) was 86.8% and Magnus’ average rating was over 3000 in these losing/drawing games. He then looked at… his? (or maybe all of the players he named’ opponents? he doesn’t specify) 28-34? players across, across 317 games, their average level was 89.8%. He then looked at Magnus’s last 100 winning games and saw Magnus only scored 89.3. Therefore according to him 1/3rd of his opponents when he losses or draws is playing “better than Magnus” Magnus which means they must be cheating. *** Besides the fact that because he doesn’t actually share his dataset or equations it’s impossible to see what he’s actually talking about. Even if we take everything he says at face value this is a nonsensical way to prove that cheating exists and it’s as bad as 1/3rd of all the players he’s losing/drawing against.
At 2:10:00 or so Eric is arguing based on Hikaru's strength and Kramnik keeps referencing sttatistics to justify calling Hikaru's streaks unlikely. But it's precisely a statistical analysis that shows that these are not unlikely. Someone please have Kramnik interviewed by someone who understands stats and can explain it to him
Is is precisely the known win/loss/draw percentages that shows such streaks are *very* unlikely. Someone can certainly interview me because I understand stats, and it all begins with the win%. Kramnik is spot-on wen he speaks of Elo and the baked-in stats, and one must start with a win% that does not include the draw%. The streaks are statistically very unlikely. Some ways around this are 1) to have an explanation as to why Elo breaks down near 3000, or 2) an explanation as to why Elo stats are not reflective of outcomes for blitz games. There may be other reasons, but if we stick to ONLY the known math, Kramnik is correct in the term "interesting". The win% presented by some "mathematicians" started with "let's assume Hikaru wins 90% of his games against such players" and such statements are total garbage. Why assume? The win% os between 73 and 80, which makes a huge difference. The math after determining this value is all that is needed for any and all of us to use an online calculator to determine likelyhood of a streak after x trials. So it is precisely a statistical analysis starting with known win percentages that shows such streaks are very unlikely.
@@scottekoontz did you read my reply to a different post of yours in another threat? I stated there that the win probability of Hikaru in Blitz against 2850-2950 opponents is 85% and against 2650--2750 it's 91%. This follows from the data: number of wins divided by number of games. No simplified Elo computation (and cc doesn't use Elo in any case). What's interesting is that these probabilities fall into a narrow sweet spot: a bit lower, and we'd never see such streaks, a bit higher, and we'd see them all the time. Very refined cheating indeed. As for an explanation why Elo breaks down / is not accurate in online blitz, see the C2 podcast where Caruana talks about it, but that's a different issue...
@@martinpaddleI thought a lot of games that use any kind of “elo” actually just use either Glicko, or Glicko-2 or some other related system. I know chesscom doesn’t use elo because it’s a worse rating system. While only anecdotally true, all rating systems have some kind of breakdown at the highest ratings from what I’ve seen across all games that I know use similar rating systems.
@@martinpaddle Yes, you are correct about actual Hikaru win percentages, although using the subject's games that created the stats to prove something is likely or not is useful but does not paint the entire picture. Hikaru really does have a high win% that exceeds Elo calculations (which are valid if rating system is using Glicko). And blitz games do appear to have fewer draws and they are funneled to both players as additional wins. An extreme example: If Hikaru were to spend the year winning 92% of all blitz games against 2850-2950 players, we would not use 92% to determine is a streak of 55/55 was likely within x games. The idea is to see what is 'interesting' in Kramniks words, and 'sus' in Hikaru's words. Not saying Hikaru cheats, but there must be break points at which we cannot say "Player A is not cheating, we can use his own win% and streaks to prove it can happen..." So we end up with a general agreement that Hikaru (possibly others) gleans a lot of what we expect as draws into wins, and/or that Elo calculations have issues at extreme differences. So the next questions are: 1) Is this true for other players? Using only Hikaru's existing stats to prove Hikaru's streak stats are likely is not proper data science. If other 3200 players have a similar win% against the same range opponents, then we need to adjust calculation for blitz especially with a wide Elo difference. 2) Why are some papers using Elo stats incorrectly? Most (one often quoted paper in particular) have used the Expected Score and not the Win Probability to show that even Elo math agrees with such streaks being likely, which is clearly an incorrect first step. Using Hikaru's actual stats is not completely proper, but much better than some of the "I'm a data scientist, and here is what I found..." papers which misuse Elo math.
Kramnik fundamentally misunderstands statistics. He seems to think just because something unlikely happened, it's proof that there was fouplay involved. If that were the case, winning the lottery should land you in jail. He needs someone who understands the law of big numbers to explain this to him. He's embarrassing himself. He was a brilliant player, it's sad that this is how he'll be seen if he doesn't stop soon.
Kramnik is fooled by randomness and basic stats. Stop giving him a platform to spew his nonsense and pretend like he's doing something mathematically rigorous ... It's too cringey to keep listening
Not that I've watched the video or followed the controversy too much, but I tend to disagree with the claim that Kramnik is completely inept. Hikaru's results are so statistically improbable that they are essentially impossible according to statistics. Sure, statistically speaking, such a winning streak can happen once or twice in someone's entire career, but Kramnik's point was that Hikaru managed to have similar streaks in relatively short succession. The problem is that statistics is not something that necessarily accurately reflects reality. Unlike statistics' typical assumption of complete randomness other than one variable, for example, there may exist factors that have not been accounted for. For starters, during Hikaru's 45.5/46 game streak, he played against only 5 opponents during these 46 games. There is absolutely nothing random about that, ergo using statistics to predict the likelihood of certain outcomes is already ill-advised (other than to compare the observed results to the likelihood of such an outcome in a genuinely random statistical model.) The lack of random opponents introduces otherwise relatively negligible variables, such as the quality of one opponents' internet connection at the time, whether or not they were playing on their phones, whether or not they were playing in a noisy Starbucks, whether or not they were playing during school classes, whether or not they were playing on a bus, whether or not they were feeling ill at the time, etc. (One of Hikaru's opponents during this streak was a 15-year-old. For all we know, he was playing on his phone while riding on a school bus. Is statistical analysis going to account for that? No, it's absolutely not.) Kramnik's mistake (assuming that he is wrong) is in having too much faith in the idea that simple statistics can be used to make accurate predictions about the likelihood of events that have multiple variables that are not accounted for.
@@miguelito2361 ''Hikaru's results are so statistically improbable that they are essentially impossible according to statistics.'' simply untrue, this has been debunked already
@@miguelito2361 Statisticians unanimously disagree with the statement that "Hikaru's results are so statistically improbable that they are essentially impossible", so why do YOU continue to say that?
Kramnik is a good person. He's taking an insane amount of abuse and straw-manning from the mob for sharing some basic and obviously reasonable concerns, and he is still so calm and polite, and nice enough to spend literally hours explaining the obvious over and over. Most people would crack in 5 minutes, but he doesn't even seem all that annoyed
I am really pissed of because he makes joke of it ? (or not?) - accusing of cheating just by accuracy rating ... thats not how statistics work and he shoud know it :(
7:23 Eric: Are you relieved to not face cheaters? Kramnik: 15 paragraphs discussing how he started his analysis This was maybe one of the funniest responses to a question in an interview ever
@@FemtoSecondSpectroscopy Idk man you cared enough to reply to my comment, and there are 31 likes on my comment. Both of these things lead me to believe that, contrary to your assertion, it is not the case that nobody cares what I do or don't do
Eric, why don't you interview Lichess tech-guy to explain us their anti-cheating system, since they are open source, it would be nice to hear their perspective
To be clear, you are not going to get a clear understanding of the anticheating methods a platform uses, or at least they are incentivised to not let that be disclosed bc with that information a would be cheater would no exactly how to avoid detection.
Kramik's statistics hold about much merit as me seeing he cheated his way to world champion because it is so statistically unlikely for anyone to be world champion. He needs to release all data, procedure, results, and collaborators or people need to ignore him completely .
Exactly. What kind of accuracy he refers to? How reliable is that? How much data he has to prove anything? One thing he mentioned: at 1 of 3 games high level players who can beat him play on a similar level as Magnus on that game. I don't find this strange at all. 2600+ players are playing extremely well.
With the 20 DQs, I believe Kramnik is talking about the 2021 FIDE World University Rapid Championships. 20 players there were disqualified by the Fair Play Panel (under advisement from Kenneth Regan) with the disclaimer: "Neither FIDE, nor the Hosting Internet Platform claims that the determination of a suspected fair play violation is proof of actual cheating or an admission of guilt by the disqualified player." Rather than 100 competitors, there almost 900. This kind of vagueness of scale typifies Kramnik's grasp of statistics. There is a big difference between 20/100 (with clear proof) and 20/900 (with no proof), just as there is a big difference between winning 45 games in a row from a selection of 1000 games versus a pool of 10000. He needs to understand that there are real mathematicians around who can help him on these points. Apparently "the people he works with" have no idea about statistics or probability.
@@Momus2024 The question is not whether Kramnik can play chess so much as whether he (or "the people he works with") have even a basic understanding of statistics. I am quite confident my understanding in this specific area is well above his and many professional statisticians have already explained why his numbers are flawed, which is presumably why he has become so reticent in publishing any real analysis and simply states his conclusions without addressing any of the issues.
@@zelandakhniteblade5436 you're a patient fella, trying to explain statistics to people who don't even know enough to know that they don't know is miserable
@@FemtoSecondSpectroscopy if you listen him he is mostly repeating himself. Doesn't take many factors into account. He is amazing I will always respect him. I'm sure he's genuinely concerned about the cheaters, I lovr Kramnik But I will not agree with his many opinions.
I lost all respect for Kramnik. Initially i appreciated he was looking into cheating and taking it seriously, and he had some good ideas, but in the face of so much evidence feom experts in their fields and he just keeps making up new ways his feelings matter more than facts. The win streak is such a non-issue. He is not only wrong but when it was proven he just lied He also showed a complete lack of integrity when he pretended that he never pointed a finger at Naka...pathetic.
At 31:00 Kramnik discusses the average accuracy of his opponents, and he complains that his opponents have a better accuracy than Magnus. One explanation is that Magnus plays a lot of random openings, and therefore both players in Magnus's games will be less accurate. From comparison Kramnik's games probably start with a lot of well known theory, so in Kramnik's games both players will have an accuracy close to 100% after 10 moves. So accuracy is probably not a good metric.
dorianquelle just posted a new blog with dataanalysis of Titled Tuesday. He shows that Kramnik has a high "average accuracy", but he is prone to making big blunders. So perhaps Kramnik is the special case here.
It looks like Kramnik is coming from a good place, and I would like to see his actual data. But instead of showing his data, he wrote a blog post wrongly accusing Hikaru of cheating. How can he be surprised that Hikaru answered?
Kramnik says it's unlikely to have many streaks like that and it means that probably his opponents had inflated ratings (higher than they should have). Streaks against 2900 players - not likely. Streaks against inflated 2900 players (e.g. 2600 true not cheated rating) - very likely.
@@MMasterDE He may be right about that, it's certainly possible and I'm sure true of some. Not sure if it's true of enough people to make a big difference but maybe it is.
The "every third opponent in Titled Tuesday plays better than Magnus" claim is wrong. Every 3rd opponent has a higher chesscom accuracy score, possibly, but the accuracy score is a very gimmicky measure that should not be used for such comparisons (because for starters it is a relative rather than absolute measure, meaning it is adjusted based on your oponnent, but also impacted by a host of other factors). These accuracy scores just shouldn't be part of the conversation about cheating at all.
He’s doing nothing for the legitimacy of actual cheat detection, they’re so emotionally fuelled. Kramnik’s “interesting” investigations so far can all be boiled down to his failing ability and an unwillingness to accept it. The younger players are just better, just like it’s always been, he’s no exception. This is a frail ego throwing a tantrum.
Kramnik is a living legend. All his points were well explained with valid arguments. Sad to see how Hikaru reacted to the valid statistics on his performance. Probably should call Kramnik and try together investigate the numbers. If the person never cheated and get such a high performance rate, he probably should share this statistics and be proud of it. Calling Kramnik crazy and making jokes of everything that Kramnik says is a red flag for me.
At 44:00 Kramnik compares the performance in the first four rounds of Titled Tuesday with the performance in the last 5 rounds of Titled Tuesday. This is actually an interesting idea. I guess that some of the best players are less concentrated during the first rounds, or perhaps they will make silly moves for content, but it could be nice to see some data.
The platform establishes the criteria for anti-cheating publicly and beforehand. They make it clear that it's not 100%, but robust enough, and then, make every user sign a term of service where they agree with it. Every player who fits the criteria will automatically be banned. This way they avoid lawsuits.
feel like eric doesnt really get that he is saying he thinks people cheated to get to 2950 to be in a position to give hikaru the streak, not specifically that hikaru cheated.
The root of the problem is that Kramnik is imagining online ratings are as reliable as official FIDE ratings. When they are much less reliable for several reasons.
How many times is Vlad going to start a sentence, then start on something else instead of finishing his thought? He never lets Eric finish a sentence. He's a terrible communicator
thank you for doing this. Although at first it's easy to dismiss Kramnik, after hearing him out i tend to agree that cheating may be much more widespread than we think.
Eric kept saying that "some" of Kramnik's suggestions were hard to implement but 1) he doesn't know exactly the stats that Kramnik is proposing and 2) what does he know about how hard it is to implement stats in software? We should have access to Kramnik's stats to be checked.
Around 30 minutes in, Vlad repeats the phony "evidence" that players in important final rounds perform better than they do in unimportant rounds. Even Caruana fell for this tautology. Of course players who reach finals in tournaments are going to have higher performance ratings because they are playing higher-rated, more competitive players in a situation when both players are playing well.
At 2:07:00 Kramnik discusses Hikarus rating. He doesn't understand how Hikarus can have such a high rating when he plays all the time. He doesn't agree that Hikaru gains rating from farming low rated players. So for each player he should examine two numbers. * The ELO points gained from opponents with equal rating * The ELO points gained from opponents with much lower equal rating
At 2:25:30 Kramnik repeats that someone should probably be suspended for Hikaru's winning streak. At 2:30:00 He complains that Hikaru is misrepresenting his intentions.
Your motive turned from cynically attacking a particular player to the benevolent motive of addressing cheating as a widespread problem. You did this because you got called out and ridiculed, and now it's the only way to save face and make yourself look good. At least recognize what you're doing, Kramsack.
Kramnik is the oldest 48 year old I know of. Who leaves the green screen on default like that. I think Viktor Korchnoi would be more comfortable with modern technology.
At 2:12:00 Kramnik just rambles in denial, and it is weird. For anyone who has seen Hikaru winning several games in a row for years like me knows that what Eric is saying is obvious - Hikaru wins lost positions on time so many times it is not funny, and the assumption that his opponents won't blunder or get nervous while having to deal with very complicated positions in time trouble just because of their ratings shows a total lack of understanding of the dynamics of online Blitz with no increment. I even have games with 2750 estimated rated at 94% that should be statistically impossible for Kramnik. He assumes that people should play the same way all the time, that their ratings should be perfectly stable, and that it should all be ruled by some program that can simply declared someone guilty. Maybe he's been watching too much Minority Report, to say the least.
Gotta give respect to a world champ. Also kramnik, maybe incorrect in this instance, is well intended. He wants to protect the game he loves. I don't mind hearing his viewpoint at all.
@@kunalsingh4418 Agreed. Even if we do not agree with what he is saying, you listen out of respect, especially if you are a fellow GM. Almost all of the chess greats eventually struggle in some way as they become older. But this is a part of life and times will always be changing.
2:06:06 he said they can check who is guilty or not, but they can't say it's "normal" after they check, which means they have to keep checking until they find guilty. in fact, chesscom and other professors and professionals already checked, and he said they were all wrong. btw if u read his posts, he is always implying they are guilty. 2:06:36 ok is Eric just not up to date or is he helping Kramnik on purpose? the whole point is kramnik doesn't understand how the ratings are inaccurate and how that makes all his points bogus.
Grischuk gave a very interesting interview to Levitov, where he gave a lecture on basics of theory of probability in context of recent events with cheating. Very interesting. The conclusion he gave is that, in theory, we could count the probability of cheating with a 100% accuracy, if we have values for all variables in equation, that he presented in that video, but the problem is, there are couple of variables, that are impossible to figure out with current methods. Then he just shows how it works by giving some numbers that he randomly came up with.
@@pat_athens Well, fine ... with 'almost' 100% accuracy. The point is, it would be pretty clear, if the player was cheating or not, if it wasn't for one thing ...
without having watched the interview, you can cumpute the probabilities of streaks exactly if youbknew the winning probabilities of each game. But the best you can do is estimate these probabilities from past data, which is how statistics works. This has been done. None of this proves cheating though, it can just give you a way of quantifying the likelihood of cheating.
It is very clear, that while cheating does take place (being a reality) at various levels, what Kramnik (and Nepo) really object to is the "Farming"...as they and other sporadic high level players that only play key online events, do not do this. Farming has occurred at all levels for many many years, whether at the very top (3000 plus players farming 2700-2800s), 2800 players farming 2500, 2500 farming 2200, and so on. Players do this to increase their ratings for a variety of reason, and I am speaking now specifically for online platforms. It doesn't matter to Kramnik/Nepo that some GM's do it for content or not, and of course it can equate to $$$, which again, Kramnik and Nepo do not enjoy having, or player Farm for other reasons such as 'status' and respect..I guess. While Cheating is an underlying issue that certainly needs to be constantly addressed and I believe is, to some regard, the Kramnik's and Nepo's (etc) think farming, if not cheating, is unethical.
He's talking more about a variety of parameters he has assumedly tested against players, mainly performance in lost and drew games player vs player which is a big thing if certain patterns are found. data is very interesting tbh.
I checked a couple of my recent games. I lost one with 30% accuracy and my opponent had 36%. Another, I won with 85% and my opponent had 72%. Both games over 20 moves. Did I cheat?
One the one hand, it is truly impossible to imagine people like Hikaru, Danya, Bortnyk cheating. On the other hand, It's sad that Kramnik chose Hikaru as his target of all people, because cheating has become more and more frequent in every format of the game. Even in regular bullet and casual games, without any money rewards, somehow people still manage to, and are inclined to, cheat. Just today I played 5 1700 rated bullets, 3 were fair and I lost; 2 of them were played by 800 Clasically rated players, with 97 percent accuracy who played precisely the same opening moves as each other,(Caro exchange) and had a similar profile. I wish the former world champion would go after the real problem, that does exist, not a phantom.
Before I comment: 1) It's Xmas eve and I have nothing better to do. 2) I'm a Hikaru fan. I know most of y'all hate the guy, but I'm curious to hear what Kramnik thinks. So after watching the entire video, it appears to me that the man wants to launch his own anti-cheating business and this is his way of marketing it - by using the biggest chess platform and sending accusations at the biggest chess streamer out there. I wish they'd dug deeper into Hikaru instead of saying it's just mathematics and nothing more. Kramnik himself admits that Magnus and Hikaru are capable of that level of play. So why accuse Hikaru of cheating if there's nothing personal about it, when Magnus also demonstrates weird stats.
He discredits himself within his first thought, "I've only started playing online chess maybe a year ago"... He clearly has no experience in online chess and is not aware of the difference between OTB and Online chess. Hikaru has pointed this out adnauseam. Krammnik is desperate for attention hence the pink shirt. Good day.
It seems like something Kramnik totally misses is not just the chess skill, but different aspects like simply how quick/accurate someone is with a mouse. It's got to be nearly impossible to cheat in any way at 1 0 chess for example, but players like Naroditsky, Andrew Tang, Minh Le, Bortnyk and others maybe 2600 ish OTB players can hang with the like of Hikaru and Magnus in some cases. Even Eric Hansen would likely crush Kramnik in bullet chess, but might get beat regularly once time controls get above 20 minutes or something.
Nobody said strong gms don't have a chance to beat top gms. Kramnik claimed rather the opposite that streaks as Nakamuras against good players should be impossible.
@@paulgoogol2652 But the same people that are strong online bullet players are not the people that are top GMs (usually). Andrew Tang for example is an exceptional online bullet player, he has incredible speed and flagging potential. But he's not a top GM. The skillsets are very different, there are mind tricks at play in online bullet that isn't there OTB or in slower time controls, mouse speed is also a hugely important factor. So who's a top GM and not is kind of irrelevant, you'll have IMs that are much stronger online bullet players than some top GMs.
I'm about to watch the entire 3 hours and a half of this video. I will do it while playing some game to avoid suicidal thoughts. Wish me luck, guys. See you on the other side.
listen up yu nakamura fans. kramnik was champion of the world with IQ 19o. he would not talk rubbish.and of course he is not alone if Kramnik says that Nakamura cheats i belive him
A couple of things really stood out to me. First, Kramnik saying that there is no need to prove cheating beyond a reasonable doubt which is just insane to me. The other is his claim that all the backlash he has received means to him that he is on the right track. In terms of Hikaru, he is completely missing the real point of the backlash and that is the fact that his argument is so clearly and blatantly wrong that people can’t help but try to tell him this. Obviously any push against cheating is a good thing but the way Kramnik is doing it is just wrong. His complete lack of understanding of the online chess world and his willingness to call out specific chess players without giving any actual concrete proof is ridiculous. He keeps saying that statisticians and mathematicians tell him this that and the other which is either a lie or the people he is consulting are just like him, without any knowledge in this field at all. When will he realize or when will someone tell him that his logic is a fallacy?
People in comments are insulting and attacking Kramnik personal but cant show anything wrong in his ideas, as he is describing an AI anticheating system which is now used in many games. But who are those people which are attacking him and the whole anticheat idea?
43:35 "you don't even need to look at games" - oof I wonder if he bothered to look at those 89% accuracy wins by magnus where he opened a3 or h3 My feeling is that good cheat detection will be looking at players who find genius moves in 5 seconds and then take the same time to play obvious moves
At 2:12:15 Kramnik mentions that whenever a player loses 4 games in a row, then the chance of winning the 5th game is surprisingly high. This is actually interesting. Apparently this means that players start cheating after losing 4 games in a row. (or perhaps they leave the computer and continue playing the next day)
2:10:00 Kramnik is correct here. Elo does have percentages baked-in, and we should use them. Also he alluded to the fact that even if you play against an average of -300 Elo players, winning a long streak (no draws) is very difficult. 2:11:00 Psychological issues go both ways, and one cannot assume they ALL favor the better player. In fact studies show farming is generally not worth the better player's time because on average the lower rated players tend to play slightly above their level. That's the "psychological" issue with some studies to back it and not some "well, as we all know" comments. All other comments that make assumptions are not part of any statistics. You either trust Elo percentages of you do not. If not, then you're not an important part of the statistics conversation. If you do understand/trust Elo, then Hikaru's win% for the 45.5/46 streak was not 90% as some "mathematicians" claimed, but between 73 and 80. This makes all the difference in the world. Even using 80% (helping Hikaru out as much as possible, i.e., assuming his highest rating and keeping his opponents' ratings static) it would be very unlikely to achieve 45.5/46 even after 10,000 trials/games. If Kramnik would simply give what he found to be the assumed win% for the Hikaru streaks, then all of us (you don't need to be a mathematician) can calculate the likelihood after x trials.
2:13:00 A camera, sure. But note that according to Hikaru is would be easy to cheat OTB, but that is coming from a guy who wears headphones while streaming. Headphones would be the easiest way to cheat. No, I'm not saying Hikaru cheats, but given the two situations of OTB and headphones, of course the easiest way by far would be headphones. So very very simple to cheat that way. So incredibly difficult OTB. After Hikaru's uneducated comments about Hans' OTB play there's a bit of schadenfreude here. 2:16:00 "FIDE blitz ratings are not very accurate because..." OK, so do we assume Hikaru's rating is too high or low? Do we assume his opponents' are too high or low? This gets us nowhere except to say we cannot be certain Elo is a good representation of the player's abilities. That may be so, but until we know more we use the known ratings. The 45.5/46 streak had known blitz ratings. We have known/accepted win/loss/draw percentages for such games. Why don't we use those Elo ratings and the resulting win% for the probabilities of such streaks? Otherwise we are left with "it's possible to have such streaks because it is possible".
Hikaru has a pretty good record against Kramnik, just off the top of my head, Hikaru beat Kramnik in the final round of titled Tuesday 3 months ago for 11/11
accuracy measures consistency or precision in stastics. this is how much variance moves a player makes between games. in philosophy accuracy is the reliability -a measure of the objectivity of each primise. chesscom use of the measure of accuracy is the amount of top engine moves so it changes the fundamental common uses of the tearm accuracy. what i think kramnic means when he says accuracy is its intended meaning which is is the player making common types of moves for that player in all games measured. kramnic is saying that there is a flaw with hikaru opps in that they dieviate with a low reliability measure. niw here is the problem… kramnic needs to test for reliability within group variance but instead he is using validity (logical) measures -comparisons if engine to make these claims and now you have the dreaded type 1 error (accepting the null when the hypothesis its false)
Magnus accuracy less than 90 is cause he trolls with bong cloud, A4 and H4 openings lol. Magnus is so good that he can still beat top gms even if he makes 10% inaccuracies in the openings lol
No, cheating is good ... I have no incentive or ambition to get good at chess and neither do you or other 99.9999% of players ... Coz none of you are going to be GM ... If you are above 8 years of age
Hence cheating is good ... Chess is not my craft , i don't cheat in my craft , there i try to improve my skills ... But chess is not ... Hence I don't care ,i just enjoy chess & of opponent is talking trash to me or is showing off , then I show him his place 😅
@@speed999-uj5kr Just shows how emotionally weak you are that a little trash talk and you have a tantrum flip the table and ruin the game like a child, doesn't prove anything about your opponent. And whether or not you make a career of chess if you're not striving to improve (which by itself will improve you as a person and in other aspects of life) you're wasting yours and whoever you're playing against time.
@@mendikk6386 doesn't matter ... No one in this world is 100% truthful ... Everyone cheats , if you go office late by 2 mins ur cheating, if u lie to your partner u r cheating ... So everyone cheats ... There's nothing wrong in it ...
kraminik shows a WGM level of communication "skill", but his points are real GM level tl/dr: he wants chesscom to be more aggressive and admit there is a big problem. his hikaru streak point raises 2 hypothesis: A) hikaru is cheating, or B) hikaru's opponents have cheated at some point to get such high wrong ratings... and he _explicitly_ says that B is much more likely the case
Thank you for bringing Kramnik on. I think he is being treated to harshly for trying to solve a problem. His ideas aren't perfect but he is well meaning. After watching the Dubov and Caruana interviews they see the problem as well but haven't really brought any good ideas to the table to solve the problem. Kramnik is taking notes, dealing with statisticians, and has some ideas. His mistake was posting about some players and their rating performances and a lot of players took it the wrong way (For good Reason). I also don't like the idea of anyone calling out minors when they don't have a way to defend themselves like Naka was saying but Kramnik, Dubov, & Caruana are more RIGHT then wrong in my opinion.
He said he sent the data to chessKom! And they didn't agree with him even though an average Joe person could see he is right mathematically.. while he never went to high school himself
@@angosalvo5734 he is not right mathematically, there was a DOCTOR IN STATISTICS of the university of Chicago that made a while paper into that, Kramnik has no right talking about stats over a freaking PhD in Statistics lol
I'm also a maths major, Kramnik as such a Simplist view on math and stats that only a high schooler could, he should keep to chess and let actual mathematicians handle the math
I like the idea of secretly matching someone suspected of cheating against an engine (without them knowing, of course). If they manage to draw against Stockfish then that would be evidence, would it not?
@@Monika77ful If a person is suspected of cheating AND also happens to "hold" against stockfish (once, twice, three times?) then that would be good but not perfect evidence of cheating.
100% spot on the over performing. If you are performing better than 99% of people and actively compete for money prizes then you 100% should be checked!
flip 1 million fair coins, somewhere in there there will be a streak of 20 heads in a row, exclude 999000 flips, 20 heads in a row in 100 flips? HMM INTERESTING
I find Vladimir's explanation for his post about Nakamura's streak interesting (no pun intended). I always thought that it was inevitably a direct accusation or at the very least an expression of suspicion against Nakamura. However it does make sense, like he says, that if many ratings are inflated through cheating, Nakamura would get much longer winning streaks than usual and much more often. This is simply because cheaters would have such high ratings from playing against other people, while they would not dare to cheat against Nakamura.
16 minutes on the video and I wish you asked what statistical analysis he used, what kind of data structure he used to run the statistics because, I don't believe he actually used statistics. He would need a big database of results and typing these into the computer would take whole day if not more. He looks like an old-school dude who doesn't use the computer much as he recently met with online chess after hearing it from someone. If iI could just download Hikarus data I would have run the test myself. There is a high chance he was biased, he used inadequate data, or his data was falsely structured. Anyways, anyone who watches Hikaru would know he is not cheating, I think this is an unnecessary debate. Also, humans are not machines, there may be good days or bad days.
I caught bits and pieces yesterday on stream and he said he was planning to make his own anti cheat system. So until that or of that ever develops, he has shown zero serious analysis. I was also curious as someone with a Mathematics degree but he said something about 5 pages of calculations no one would read. Tldr he has produced nothing rigorous at this point
I actually understand Kramnik's point of view and I agree with his sentiment! There seems to be a lot of hate from fanboys who clearly want to misinterpret what he's trying to say. Kramnik wants to use statistical anomalies in play to categorize players so you can closely monitor them. IF someone has a crazy streak, they are more likely to cheat, so it makes sense to monitor their play. It's not accusing them of cheating, it's saying "You did exceptionally well, you played way above your usual level, so we'll secretly more closely monitor your play to make sure you're not actually cheating". Also anti-cheating requirements for online tournaments is also a good idea, especially having to download a program. Most other online sports require PC checks for all players who are accused of cheating, so it makes sense.
Not sure I understand. If Hikaru goes playing against lower rated players, I guess he's likely able to go on a very long streak, it's is legit. Hikaru cannot have a such streak in titled Tuesday against strong players like he did admit it.
His approach is plain wrong and naive. Anomaly detection by itself is too naive and noisy. Real life is full of anomalies, so today ML is used in conjunction with other strategies that can more definitely prove something. For example, just try building a stock trading system using only anomaly detection, and sit back and watch all your money dissappear (which ironically is what is happening with Kramnik's reputation)
30 min in - yes even 1500 elo player are having accuracy at 90+ at some times, especially if opponent plays into in (that is make easy game for him) but that does not means he is Magnus. Is he really confused about what high accuracy game means (that is not making mistakes) or am I missing something there?
I think Kramnik has a very great perspective for online chess because of the amount of time he has spent playing offline and getting introduced online so late in his career. He clearly sees a level of play that isn't consistent with playing at live tournaments so he's aware there's a lot of cheating when he experiences level of play he hasn't throughout his career. I think online chess opens the door to a lot of talented players who otherwise cannot participate in live over the board chess. Also obviously the online format favors far faster time variants like blitz or bullet that is a different style from a classical player who spends less time playing blitz.
Kramnik ruined his reputation buy accusing Hikaru of cheating. He brought this on himself. On a positive note, he should be awarded "Chess Content Creator of 2023".
I feel this video is too short. I'm going to wait for Hikaru's 4 hour reaction video.
The video is 3h34 minutes. We need at least 8 hours reaction.
Doesn't matter since you probably can't even focus for 2 min straight
I'll save you three hours. 1. Kramnik repeatedly says he's not accusing anyone specific of cheating like HIKARU for example, then goes on to discuss Hikaru's games. 2. Eric repeatedly makes solid points refuting each of Kramnik's points, Eric making the case that each of the aspects of success a guy like Hikaru has can be clearly explained. At which Kramnik says, "Yes, sure of course," and then keeps concluding what he already thinks.
You're welcome
Any good bits you can timestamp? ty!
that's mischaracterizing what he says so much... he literally said that hikaru's streaks show that something is wrong: A) hikaru is cheating, B) his opponents are cheating (in other games, so their rating is inflated), and he says explicitly that "B" is much more likely the case
kraminik's communication's skills are trash tier, but if you really pay attention to what he is _trying_ to say, it's very reasonable
You da man. Thanks for saving me 3.5 hours.
not really. he has no data, you are just trying to intuit if it sounds reasonable or not, his data is that hikaru overperforms in blitz matches, no shit that's to be expected from a player who is 300+ elo stronger in classical, now show the data on why hikaru specifically is an anomaly, u cant look at his data and be like "looks funny to me". @@zaftnotameni
@@Papageno123 I think there's a lot being lost in translation here, not just from english but also from cultural differences - I have quite a few eastern european friends that I could see making the same type of extremely specific insistent points about something, even the doubling down on focusing on hikaru when people start questioning
is he right in doing it? no
is he doing himself any favors? no
is he helping cheating be solved? no
is he being malicious/crazy? no
Chessbrah managed to release a movie on cheating in chess before Netflix 😂
they can chop it to five episodes, 40 mins each, add intro put credits and boom they can market it as a series: PS, someone give it a title "_______" season 1 😂😂😂
lol
Somebody should make a short version of this with his main points. I want to give Kramnik an opportunity to explain himself, but at the same time there's no way I can watch 3 and a half hours of him talking
100% Agree. I also want to give him the benefit of the doubt. But 3 hours 34 minutes? fuck that. can AI do a TL;DW type of thing?
I'm 30 minutes in and there is nothing new or specific yet. Only the claim that he and his mathematicians checked things but not exactly what they came up with. As for what anti-cheating might look like, he suggests taking a suspected cheater on with an engine.
@ Give him the benefit of the doubt? For what? lol I think he made some really good points. He don't wanna talk about the best methods public, but he suggest a variety of them, and if they all sets off alarms, may want to investigate or just instant ask person to leave etc, have it in the rules to not award such cases. Also, if people who cheat succeed, it's does affect others, it means others don't win, and they either quit or have to cheat to win.
The main point is that he has some data but he hasn't published it.
At 31:00 his data shows that his opponents in lost and drawn Titled Tuesday games have better accuracy than Magnus
At 44:00 he notes that some players are surprisingly good in the first 4 rounds of Titled Tuesday but bad in the last 5 rounds.
At 2:07:00 he doesn't understand that Hikaru can have such a high rating when he plays all the time
At 2:12:00 his data suggest that some players start cheating after losing 4 games in a row
At 2:25:30 he says that someone should probably be suspended for Hikaru's winning streak.
At 2:30:00 He complains that Hikaru is misrepresenting his intentions.
I listened to it it's 100% valid if you are able to follow the reasoning and the multiple indicators he used.
Kramnik: There are only four possibilities:
-Cheater vs Cheater
-Cheater vs Honest
-Honest vs Cheater
-Honest vs Honest
So the probabilty that someone cheats is 75%.
one of the maffs of all time
what an utter nonsense 😆
Abusing/misusing stats is how they did the mass covld hysteria 2020-2021
And his solution is for the cheater to play against a cheater. Bizarre.
cheater vs honest and honest vs cheater is the same so there are only 3 possibilities or 66% of cheating out of 3 games ASSUMING every of the 3 items shows up EXACTLY 1 time in sets of 3 games. which is, ffs, highly unlikely and obvious to anyone using basic logic.
if you generously assume 10% of players are cheaters, then you can calculate more realistic odds. but saying that in 2 of the 3 games there's cheating is totally unreal.
so, vladimir is starting off of completely wrong premises that comfirm his unreal odds.
Thanks kramnik for your contribution to my sleep playlist.
Kramnick Ph. D. in yapping
Who is going to tell Kramnik what green screen is really for?
now imagine this... what if he simply didnt want to use it? that would be crazy i know..
Buttttt that would be cheating to tell him 😂
He just wants that Cosmo / Wanda color scheme.
this made me laugh more than it should have :D kudos...
@OGRia imagine you have a stick up your...
Starting around 2:05:00 they talk about Hikaru
Buddy saved me two hours of my life what a chad
Chad pro Max
@davidwiederman8841 Yeah, I can see that Eric refuted most of Kramnik's nonsense. Kramnik's accusation is full of loopholes
watch on 1.25 or 1.5 speed, guys. Some timestamps for u guys:
23:45 what can be done against cheating
29:40 about the tests. Test on levitov chess about players performing in last critical rounds, etc
Then he talks about accuracy.. i suggest you to skip it, because accuracy is not that relevant, with all the respect. it depends. Depends on what you're playing, how positional your game, how long, etc. Just rememberd how Aman won vs some 500 rated player and they both had 90+ accuracy just because Aman took all his pieces and played with his food for like 200 moves more. So, 500 rated player has played 90+ accurate and better than Magnus in average, eh) with no pieces and chances for the win though, but that accuracy Pog
45:25 another example, how often people have 4 out of 4 from first games
54:37 Kramnik suggests playing TTs only with camera on
58:10 Eric says it's still not 100% proof, Kramnik says that it should be a fight from different angles combined
1:11:10 unless someone beats Hikaru or Magnus lol, yeah
need to go
need to go......NOT soon enough
Don't waste your time. He literally says nothing new. I watched it live on twitch.
anything about his Hiki "non" accusations?
Grassy Ass Me Uh Me Go
Do you literally not care?
@@juliusjackson9198he thinks Hikaru is 75% cheating while also saying he is not accusing hikaru of cheating. Bro is waffling hard
@@nadimasaduzzaman😂😂😂 good 1.
Let me save 3 hours for you. Kramnik: "I'm not saying anything. I'm just saying."
i think kramnik means well and I don't think he's 'gone crazy' like i hear a lot, i just think maybe he's being consumed with confirmation bias or isn't the clearest thinker on this topic. when asked a direct and simple question he often gives a rambling answer that you have to kind of put together an answer from an ocean of words said that relate to the question. i get that english isn't his native language, but he is fluent enough that I don't think it's a good explanation. when I speak in a non native language I tend to be even more direct because I don't have the vocabulary to express more subtle points in great detail.
one response i noted was kramnik saying that streaming and being on camera makes "no difference" in regards to cheating...and eric's reaction clearly showed he wanted to push back on that as well. Being a popular streamer definitely makes it less likely that one would cheat. having a humongous audience of people watching every game hikaru plays while he is on camera greatly increases the scrutiny of his games. it's not just scrubs like me that watch hikaru...titled players, GMs, super GMs are watching him and many of these people aren't the biggest fans of hikaru personally and would not hesitate for a moment to bring attention to suspicious gameplay if they saw it.
You are extremely naive on this, do you really believe that streaming and zoom can prevent someone from cheating, really? The temptation to cheat online is at least 100 fold greater than over the table. Which means that for every one top tier player that ever tried to cheat OTB there are at least 100 more that are trying their luck with it online.
There are innumerable ways to cheat, even with zoom. For one this person can have an app relaying the game to a friend on a different computer which has an engine running and he could be feeding the moves via a hidden earpiece. That is just one easy way this could be done.
Also the mathematics does show that cheating is going on so what exactly makes you think that Kramnik isn't the clearest thinker on this? Is it a matter of magnitude which you think Kramnik is overestimating? Even so, would it not still be a problem, even a serious problem that needed to be addressed?
PS: Or is it that Kramnik accused an American player of cheating and you Yanks Don't like that, just like you got all defensive when Hans Neimann was accused of cheating? BTW that meteoric rise Neimann was on up until a year ago seems to have stalled quite aggressively and has gone in reverse. Fancy that, being a below average GM level player for the longest time then suddenly for a two year period start playing better than anyone in history only to have that rocket ship of progress stall and start to dip. Sure, if you believe that Neimann didn't cheat, I have a 24k gold vibrating butt plug to sell you. As for Hikaru, I can't say haven't looked at the math/statistics yet, time will tell.
Do you have a reading comprehension issue? Nowhere in the original comment does it say streaming makes cheating impossible or that it is prevented in anyway, It simply draws the clear conclusion that a bigger audience seeing the games would make doing so much harder and the games are under great scrutiny.
I have no clue what your source is for the temptation to cheat online being a 100 fold but I will assume you made it up or are self reflecting. For the last comment mathematics can show ANYTHING if you pick and chose samples. Companies do this all of the time to make statements about a product. The reason why Kramnik isn't the clearest thinker on this? because he has shown repeatedly through his own comments that he does not understand online chess or statistics at all.
"titled players, GMs, super GMs are watching him" - agreed.
At a different level I have heard it is amazing what can be picked out from a video stream by a professional, for example screen reflections from a pair of glasses.
@@Crypt0n1an
Who got defensive when Niemann was accused of cheating? You’re acting like most of the US was on his side when in reality he was denounced by pretty much everyone possible.
Thank you chessbrah for taking the time to have a full conversation with Kramnik and giving him the opportunity to share his thoughts and concerns regarding cheating. I know it was a long video (I'm still up to 1:09 so far) but I think you are doing a public service for the benefit of the chess world.
Yes. It's clear that Kramnik thinks that anticheats are integral to the future of chess. With this as a first principle it follows that companies and different actors/entities in the chess world has to get in line with this principle for the survivability of chess.
It's his entire legacy that's at stake (and everyone else prominent in the chess world) Just compare it to 80s-00s bicycle racing, where the entirety of competitors have zero credibility when it comes to fairplay.
Hambo is getting scary good at impressions
@@edmoala00:00 until the end, guy on the left
funny af
I met Hambo once, well I say met, I was at a party and had to use the bathroom and who walked out but famous online chess personality Aman Hambleton. As he breezed past me I blurted out "hey you're the chessbrah guy!" bc in my excitement his name escaped me at that moment.
But he didn't even aknowledge my existence and I had too pee so bad I could not follow up to try insist some interaction.
When I got into the bathroom somebody had take the worse dump I ever experienced in my life.
I don't mean there was unflushed or visable mess to be found, but the smell was absolutely god awful.
I swear I had olfactory nerve damage from the short time of exposure while I peed as quickly as I could, and despite my habit to do so it was so bad I left the restroom without even considering washing and drying my hands bc the stench was so overwhelming it was very uncomfortable to breathe.
I am not saying for sure it was Hambo that did the deed, I can't rule out that there was somebody that was in turn before him that cursed that bathroom that night.
But I will say if it was him, then he probably has something medically wrong with him, bc although I am not professional I can think of no other reason why somebody could have such a horrible odor to their bowl movements.
While my sense of smell was disrupted for about an hour or so, I still tried to find Aman at the party, but one of my friends said he had retired upstairs to an "invite only" section of the party that was being gatekept by two dudes that would probably get serious about it you tried to insist upon inviting yourself.
Anyway, the moral of the story is...Don't support these guys.
@@memegazer cool story
@@dioright
Don't support these guys...that is the important part of my story.
At this point, I'm starting to think that Kramnik was paid by Hikki to say stupid shit just to generate drama and content.
I think Hans paid Kramnik from the $500 million to deflect attention!
He definitely makes more money then before all these accusations. People just love drama. We are greedy little monkeys that absorb all kind of drama about people we don't know because our own lives must be so lousy I guess.
@antimatter2417 very plausible.
@@hrvat7770 speak for yourself, monkey...some of us are gorilla monks. (I'm here for a psychology assignment)
Kramnik should have opened his UA-cam account and post these statistics there. He would still lose his credibility. But at least, he would make some money out of it. Right now, the only people who make money out of this drama are Hikaru and Levy.
Kramnik's entire argument basically sums up to "Online ELO and FIDE ratings aren't the same for most players, this means that people are cheating!"
And he completely disregards all other variables that impact OTB and online play, including stress, tilt, the ability to select opponents, the non-standard game modes and time controls, etc. The rest of the video is Kramnik's mind boggling over the concept that FIDE rated 2200s can be rated higher than him on Chesscom, and him giving suggestions for how Chesscom can burn millions of dollars per year giving virtual cavity searches to the entire Chesscom playerbase.
You kind of missed his point. All of those things you mentioned should already be accounted for in the online rating. He's not complaining that the ratings are different, but that there are results that should be nearly statistically impossible which points to something being wrong. Either people are cheating (not necessarily Hikaru, but possibly even his opponents in other games), or the rating system is broken outright.
@@MrFreeGman They are accounted for in online ratings, which is why online ratings higher than FIDE ratings from OTB play. The results are only statistically impossible when you make incorrect and misinformed assumptions about the veracity of Chesscom ELO ratings. Of course the online chess arcade ELO system is "broken", because there's a huge variety of player skill levels, of available game modes, of various time controls, and all the other factors I previously mentioned which allow players to influence their probability of a winning outcome.
As a for-instance, there's a gentleman rated 3000+ ELO on Chesscom who only plays the 10 second variant of chess. At that pace, the game ceases to be chess and becomes a contest of reaction time, mouse speed, and accuracy of motion. And it's fine that he has this rating, because Chesscom is a chess *arcade* site. You and Kramnik are making a mistake by taking their ratings so seriously.
dude..hate to tell you this but because of legalities a corporation can not go around accusing pros of cheating...but i can assure you they have a list with many names you know that are more than suspicious....and cheating IS a huge problem ..that is not dependent on your.....belief
1:24:45 Gotta give props to Kramnik at least for that bit. Agree or not with "the procedure", let's be fair. I must admit I was surprised to hear that from him.
Regarding Kramnik’s metric at about the 35min-40min mark about the accuracy of games played by himself or others vs Magnus. This seems like a misunderstanding of what “accuracy” means. It’s a % of how close one plays vs an engine. A higher % doesn’t mean more “accurate” it means more like an engine. So in theory it’s possible to play more like an engine just to game those numbers regardless of winning/losing/cheating. If both players play known lines and quickly get to an end game then the “accuracy” % will be higher. Conversely Magnus who doesn’t want to play like an engine but wants to have fun is known for playing bad openings just to mess with people and create imbalance so more likely to have games that is less engine like aka “less accurate”.
(I do accept there is a lot of cheating and I’ll keep watching to see what else he says but much like the cherry picking of statistics Kramnik seems to be working off of a misunderstanding of what the numbers actually mean.)
He trying to find players that using an engine, by comparing their moves to engine moves, whats wrong with this, winning/losing percentage is a different parameter that he also takes into account
accuracy measures consistency or precision in stastics. this is how much variance moves a player makes between games. in philosophy accuracy is the reliability -a measure of the objectivity of each primise.
chesscom use of the measure of accuracy is the amount of top engine moves so it changes the fundamental common uses of the tearm accuracy.
what i think kramnic means when he says accuracy is its intended meaning which is is the player making common types of moves for that player in all games measured. kramnic is saying that there is a flaw with hikaru opps in that they dieviate with a low reliability measure.
niw here is the problem… kramnic needs to test for reliability within group variance but instead he is using validity (logical) measures -comparisons if engine to make these claims and now you have the dreaded type 1 error (accepting the null when its false)
Could you try to explain your position in a different way?
From my understanding of the situation, he's accusing quite a few titled players of using an engine. For this reason, you would compare their play to an engine's play. Therefore, utilizing accuracy should be meaningful.
Also, although Magnus may play non-standard openings on rare occasion, the accuracy metric takes into account all moves and how they compare to the engine's preferred moves. While the beginning of the match may not be "accurate," the middle and end game will raise the "accuracy" as you have to play more and more precisely to gain an advantage. This is why he can still have accuracies near the lower 90s as GM Kramnik said. While knowing 10 - 15 moves in main lines will definitely contribute to accuracy, it's ultimately a small component of the measurement over the course of a 45 - 60 move game.
It sounds to me like he's done what I would do, and targeted the middle and end game, where the play should be well outside of the prepared lines. Just my two cents. I'm also only 1/3rd of the way through the video, so there's that.
What? A higher accuracy doesn't mean more accurate it means more like an engine? Engines are more accurate than humans. Therefore, a higher % absolutely means more accurate. Like what? And people upvote this ignorant shit.
What really gets me is that I can tell already that you go along with the saying that humans don't play like engines. I doubt that you even understand what that means. Because they evaluate positions the same way we do. At least until NNUE became a thing a few years ago as well as Leela. But a lot of engines, and Stockfish up until recently, all evaluate(d) positions based on handcrafted human features. Things that are very easily understandable by people. Material, king safety, piece square tables, space, mobility/freedom, having control over important squares, having outposts, initiative.
The only thing a computer can do that we can't, is analyze extremely deep into the game. So humans, as you can imagine, will never be quite as good because it's impossible for us to look that far ahead. But we still look ahead, and we still evaluate positions the same way, some of us better than others.
Your whole take and comment is just nonsense, and it amazes me this is the shit people listen to and upvote. I'll never get this world.
@@paulsmith5237 To be clear my position is that one of Kramnik’s (supposedly many) examples that he gives shows that he doesn’t understand the the meaning behind the numbers and he’s just hashing numbers and calling it many to justify the conclusion that there is rampant cheating. I accept that there is cheating but that his numbers when he does share them aren’t the smoking gun he thinks they are because there are many ways to explain them, namely being that people aren’t completely emotionless rational computers that are always trying to follow an algorithm.
***
For those that are too lazy to hear Kramnik’s explanation that I think is particularly ridiculous here is a summary of what he’s saying starting at 31:00:
At some point about “half a year ago” he decided to look at just the last 100 games during Titled Tuesday played by himself, Nakamura, Magus, Grishuk, Kamsky, and “several players” where they lost or ended in a draw.
His average accuracy in that set of losing/drawing games was 86.3%. Of this 100, 31 in which he lost he had an accuracy of over 90%. His average rating in those 100 games was 2834.
Fabiano won titled Tuesday in the past with an accuracy of 86.1 (average over the tournament? last game? he doesn’t specify). This one data point leads him to believe that his score of 86.3 is therefore good enough to possibly win Titled Tuesday.
Magnus’s accuracy in his data set (though he admits that Magnus doesn’t lose much thetr he only had like 50 games to analyze) was 86.8% and Magnus’ average rating was over 3000 in these losing/drawing games.
He then looked at… his? (or maybe all of the players he named’ opponents? he doesn’t specify) 28-34? players across, across 317 games, their average level was 89.8%.
He then looked at Magnus’s last 100 winning games and saw Magnus only scored 89.3.
Therefore according to him 1/3rd of his opponents when he losses or draws is playing “better than Magnus” Magnus which means they must be cheating.
***
Besides the fact that because he doesn’t actually share his dataset or equations it’s impossible to see what he’s actually talking about. Even if we take everything he says at face value this is a nonsensical way to prove that cheating exists and it’s as bad as 1/3rd of all the players he’s losing/drawing against.
At 2:10:00 or so Eric is arguing based on Hikaru's strength and Kramnik keeps referencing sttatistics to justify calling Hikaru's streaks unlikely. But it's precisely a statistical analysis that shows that these are not unlikely. Someone please have Kramnik interviewed by someone who understands stats and can explain it to him
Is is precisely the known win/loss/draw percentages that shows such streaks are *very* unlikely. Someone can certainly interview me because I understand stats, and it all begins with the win%. Kramnik is spot-on wen he speaks of Elo and the baked-in stats, and one must start with a win% that does not include the draw%.
The streaks are statistically very unlikely. Some ways around this are 1) to have an explanation as to why Elo breaks down near 3000, or 2) an explanation as to why Elo stats are not reflective of outcomes for blitz games. There may be other reasons, but if we stick to ONLY the known math, Kramnik is correct in the term "interesting".
The win% presented by some "mathematicians" started with "let's assume Hikaru wins 90% of his games against such players" and such statements are total garbage. Why assume? The win% os between 73 and 80, which makes a huge difference. The math after determining this value is all that is needed for any and all of us to use an online calculator to determine likelyhood of a streak after x trials.
So it is precisely a statistical analysis starting with known win percentages that shows such streaks are very unlikely.
@@scottekoontz did you read my reply to a different post of yours in another threat? I stated there that the win probability of Hikaru in Blitz against 2850-2950 opponents is 85% and against 2650--2750 it's 91%. This follows from the data: number of wins divided by number of games. No simplified Elo computation (and cc doesn't use Elo in any case). What's interesting is that these probabilities fall into a narrow sweet spot: a bit lower, and we'd never see such streaks, a bit higher, and we'd see them all the time. Very refined cheating indeed.
As for an explanation why Elo breaks down / is not accurate in online blitz, see the C2 podcast where Caruana talks about it, but that's a different issue...
@@martinpaddleI thought a lot of games that use any kind of “elo” actually just use either Glicko, or Glicko-2 or some other related system. I know chesscom doesn’t use elo because it’s a worse rating system.
While only anecdotally true, all rating systems have some kind of breakdown at the highest ratings from what I’ve seen across all games that I know use similar rating systems.
@@scottekoontzno that shows they are uncommon not unlikely. Big difference.
@@martinpaddle Yes, you are correct about actual Hikaru win percentages, although using the subject's games that created the stats to prove something is likely or not is useful but does not paint the entire picture. Hikaru really does have a high win% that exceeds Elo calculations (which are valid if rating system is using Glicko). And blitz games do appear to have fewer draws and they are funneled to both players as additional wins.
An extreme example: If Hikaru were to spend the year winning 92% of all blitz games against 2850-2950 players, we would not use 92% to determine is a streak of 55/55 was likely within x games. The idea is to see what is 'interesting' in Kramniks words, and 'sus' in Hikaru's words. Not saying Hikaru cheats, but there must be break points at which we cannot say "Player A is not cheating, we can use his own win% and streaks to prove it can happen..."
So we end up with a general agreement that Hikaru (possibly others) gleans a lot of what we expect as draws into wins, and/or that Elo calculations have issues at extreme differences. So the next questions are:
1) Is this true for other players? Using only Hikaru's existing stats to prove Hikaru's streak stats are likely is not proper data science. If other 3200 players have a similar win% against the same range opponents, then we need to adjust calculation for blitz especially with a wide Elo difference.
2) Why are some papers using Elo stats incorrectly? Most (one often quoted paper in particular) have used the Expected Score and not the Win Probability to show that even Elo math agrees with such streaks being likely, which is clearly an incorrect first step. Using Hikaru's actual stats is not completely proper, but much better than some of the "I'm a data scientist, and here is what I found..." papers which misuse Elo math.
Kramnik fundamentally misunderstands statistics. He seems to think just because something unlikely happened, it's proof that there was fouplay involved. If that were the case, winning the lottery should land you in jail.
He needs someone who understands the law of big numbers to explain this to him. He's embarrassing himself. He was a brilliant player, it's sad that this is how he'll be seen if he doesn't stop soon.
Dickeru Fagamura fans out in strength today
Kramnik is fooled by randomness and basic stats. Stop giving him a platform to spew his nonsense and pretend like he's doing something mathematically rigorous ... It's too cringey to keep listening
Not that I've watched the video or followed the controversy too much, but I tend to disagree with the claim that Kramnik is completely inept.
Hikaru's results are so statistically improbable that they are essentially impossible according to statistics. Sure, statistically speaking, such a winning streak can happen once or twice in someone's entire career, but Kramnik's point was that Hikaru managed to have similar streaks in relatively short succession.
The problem is that statistics is not something that necessarily accurately reflects reality. Unlike statistics' typical assumption of complete randomness other than one variable, for example, there may exist factors that have not been accounted for.
For starters, during Hikaru's 45.5/46 game streak, he played against only 5 opponents during these 46 games. There is absolutely nothing random about that, ergo using statistics to predict the likelihood of certain outcomes is already ill-advised (other than to compare the observed results to the likelihood of such an outcome in a genuinely random statistical model.)
The lack of random opponents introduces otherwise relatively negligible variables, such as the quality of one opponents' internet connection at the time, whether or not they were playing on their phones, whether or not they were playing in a noisy Starbucks, whether or not they were playing during school classes, whether or not they were playing on a bus, whether or not they were feeling ill at the time, etc. (One of Hikaru's opponents during this streak was a 15-year-old. For all we know, he was playing on his phone while riding on a school bus. Is statistical analysis going to account for that? No, it's absolutely not.)
Kramnik's mistake (assuming that he is wrong) is in having too much faith in the idea that simple statistics can be used to make accurate predictions about the likelihood of events that have multiple variables that are not accounted for.
@@miguelito2361 ''Hikaru's results are so statistically improbable that they are essentially impossible according to statistics.'' simply untrue, this has been debunked already
@@miguelito2361 Statisticians unanimously disagree with the statement that "Hikaru's results are so statistically improbable that they are essentially impossible", so why do YOU continue to say that?
Very simple explanation for not banning popular streamers. Money. These streamers bring many paying players onto their site
Kramnik is a good person. He's taking an insane amount of abuse and straw-manning from the mob for sharing some basic and obviously reasonable concerns, and he is still so calm and polite, and nice enough to spend literally hours explaining the obvious over and over. Most people would crack in 5 minutes, but he doesn't even seem all that annoyed
I am really pissed of because he makes joke of it ? (or not?) - accusing of cheating just by accuracy rating ... thats not how statistics work and he shoud know it :(
7:23
Eric: Are you relieved to not face cheaters?
Kramnik: 15 paragraphs discussing how he started his analysis
This was maybe one of the funniest responses to a question in an interview ever
Thought I was tripping lol.. 🤔
Respectfully, I am not watching all that. I do appreciate the time to do this interview, though
nobody cares what you do or don't do. (respectfully)
@@FemtoSecondSpectroscopy Idk man you cared enough to reply to my comment, and there are 31 likes on my comment. Both of these things lead me to believe that, contrary to your assertion, it is not the case that nobody cares what I do or don't do
Eric, why don't you interview Lichess tech-guy to explain us their anti-cheating system, since they are open source, it would be nice to hear their perspective
To be clear, you are not going to get a clear understanding of the anticheating methods a platform uses, or at least they are incentivised to not let that be disclosed bc with that information a would be cheater would no exactly how to avoid detection.
Kramik's statistics hold about much merit as me seeing he cheated his way to world champion because it is so statistically unlikely for anyone to be world champion. He needs to release all data, procedure, results, and collaborators or people need to ignore him completely .
hahahahah. good point
Exactly. What kind of accuracy he refers to? How reliable is that? How much data he has to prove anything?
One thing he mentioned: at 1 of 3 games high level players who can beat him play on a similar level as Magnus on that game. I don't find this strange at all. 2600+ players are playing extremely well.
The conversation is so good , especially to have idea about Kramnik point of views . It's always amazing to listen discussion from big Vlad.
Kramnik i applaud your ability to stay patience with this guy
With the 20 DQs, I believe Kramnik is talking about the 2021 FIDE World University Rapid Championships. 20 players there were disqualified by the Fair Play Panel (under advisement from Kenneth Regan) with the disclaimer: "Neither FIDE, nor the Hosting Internet Platform claims that the determination of a suspected fair play violation is proof of actual cheating or an admission of guilt by the disqualified player." Rather than 100 competitors, there almost 900. This kind of vagueness of scale typifies Kramnik's grasp of statistics. There is a big difference between 20/100 (with clear proof) and 20/900 (with no proof), just as there is a big difference between winning 45 games in a row from a selection of 1000 games versus a pool of 10000. He needs to understand that there are real mathematicians around who can help him on these points. Apparently "the people he works with" have no idea about statistics or probability.
i think Kramnik one of the greatest chess players of all time....understands the game a bit better than you
@@Momus2024 The question is not whether Kramnik can play chess so much as whether he (or "the people he works with") have even a basic understanding of statistics. I am quite confident my understanding in this specific area is well above his and many professional statisticians have already explained why his numbers are flawed, which is presumably why he has become so reticent in publishing any real analysis and simply states his conclusions without addressing any of the issues.
@@zelandakhniteblade5436 you're a patient fella, trying to explain statistics to people who don't even know enough to know that they don't know is miserable
Eric is coming up with genuine questions. Kramnik is just answering by saying it's statistically impossible. So frustrating
you aren't even trying to listen or understand him, because if you do you easily get the point he is making.
@@FemtoSecondSpectroscopy if you listen him he is mostly repeating himself. Doesn't take many factors into account. He is amazing I will always respect him. I'm sure he's genuinely concerned about the cheaters, I lovr Kramnik But I will not agree with his many opinions.
@@seya_2 unfortunately he just doesn't understand statistics but he's sure that he does, so he draws the wrong conclusions
I lost all respect for Kramnik.
Initially i appreciated he was looking into cheating and taking it seriously, and he had some good ideas, but in the face of so much evidence feom experts in their fields and he just keeps making up new ways his feelings matter more than facts. The win streak is such a non-issue. He is not only wrong but when it was proven he just lied
He also showed a complete lack of integrity when he pretended that he never pointed a finger at Naka...pathetic.
ALL former world champs have gone cra cra.
( Except Vishy )
How did Kasparov go crazy?@@niagra898
@@niagra898and how did you come to that conclusion? 😂
wth was proven again??? 🤡
@Morphysince94 I'm not supporting him but Bobby fisher has gone crazy actually
Kramnik rocking that cosmo wanda setup. Wishing that hikaru be cheating
Could you have Kramnik on again?
At 31:00 Kramnik discusses the average accuracy of his opponents, and he complains that his opponents have a better accuracy than Magnus.
One explanation is that Magnus plays a lot of random openings, and therefore both players in Magnus's games will be less accurate.
From comparison Kramnik's games probably start with a lot of well known theory, so in Kramnik's games both players will have an accuracy close to 100% after 10 moves.
So accuracy is probably not a good metric.
So in short Kramnik should compare different openings, and examine which openings lead to games with a high accuracy.
Magnus likes to troll his opponents because he is so precise later on and can save losing positions easily. Many times he plays while being drunk also
dorianquelle just posted a new blog with dataanalysis of Titled Tuesday.
He shows that Kramnik has a high "average accuracy", but he is prone to making big blunders.
So perhaps Kramnik is the special case here.
It looks like Kramnik is coming from a good place, and I would like to see his actual data.
But instead of showing his data, he wrote a blog post wrongly accusing Hikaru of cheating.
How can he be surprised that Hikaru answered?
1 minute shorter and it'd be 3:33:33.
I see you know what numbers are. Very good, have a sweetie.
and 1 second *
@@Nick-wp8qy On desktop, I see "3 34 33". On mobile, I see a more pleasant "3 34 34".
@@Truffle_Pup Cheers
@@RG0011003 34 33 is symmetrical though 😮
2:05:50 I love how kramnik says he's not accusing hikaru but in the same breathe says it's impossible for hikaru to have that streak lol
Thanks for the timestamp.
Kramnik says it's unlikely to have many streaks like that and it means that probably his opponents had inflated ratings (higher than they should have).
Streaks against 2900 players - not likely.
Streaks against inflated 2900 players (e.g. 2600 true not cheated rating) - very likely.
He's saying that he thinks his opponents cheat, that they are higher rated than they are supposed to and is afraid to cheat against Hikaru.
@@MMasterDE He may be right about that, it's certainly possible and I'm sure true of some. Not sure if it's true of enough people to make a big difference but maybe it is.
The "every third opponent in Titled Tuesday plays better than Magnus" claim is wrong. Every 3rd opponent has a higher chesscom accuracy score, possibly, but the accuracy score is a very gimmicky measure that should not be used for such comparisons (because for starters it is a relative rather than absolute measure, meaning it is adjusted based on your oponnent, but also impacted by a host of other factors).
These accuracy scores just shouldn't be part of the conversation about cheating at all.
He’s doing nothing for the legitimacy of actual cheat detection, they’re so emotionally fuelled. Kramnik’s “interesting” investigations so far can all be boiled down to his failing ability and an unwillingness to accept it. The younger players are just better, just like it’s always been, he’s no exception. This is a frail ego throwing a tantrum.
Try to listen to his arguments
Kramnik is a living legend. All his points were well explained with valid arguments. Sad to see how Hikaru reacted to the valid statistics on his performance. Probably should call Kramnik and try together investigate the numbers. If the person never cheated and get such a high performance rate, he probably should share this statistics and be proud of it. Calling Kramnik crazy and making jokes of everything that Kramnik says is a red flag for me.
At 44:00 Kramnik compares the performance in the first four rounds of Titled Tuesday with the performance in the last 5 rounds of Titled Tuesday.
This is actually an interesting idea.
I guess that some of the best players are less concentrated during the first rounds, or perhaps they will make silly moves for content, but it could be nice to see some data.
Compare with Steve Levitt's work on sumo.
The platform establishes the criteria for anti-cheating publicly and beforehand. They make it clear that it's not 100%, but robust enough, and then, make every user sign a term of service where they agree with it. Every player who fits the criteria will automatically be banned. This way they avoid lawsuits.
People wont bother because of the length of the video
Au contraire
Good while you are at work
Nah this is great, very good content from chessbrah
I just skipped the first 2 hours and it’s pretty good
Tell me don’t don’t subscribe to the Chessbrah VODs without telling me you don’t subscribe to the VODs…
feel like eric doesnt really get that he is saying he thinks people cheated to get to 2950 to be in a position to give hikaru the streak, not specifically that hikaru cheated.
The root of the problem is that Kramnik is imagining online ratings are as reliable as official FIDE ratings. When they are much less reliable for several reasons.
How many times is Vlad going to start a sentence, then start on something else instead of finishing his thought? He never lets Eric finish a sentence. He's a terrible communicator
thank you for doing this. Although at first it's easy to dismiss Kramnik, after hearing him out i tend to agree that cheating may be much more widespread than we think.
Eric kept saying that "some" of Kramnik's suggestions were hard to implement but 1) he doesn't know exactly the stats that Kramnik is proposing and 2) what does he know about how hard it is to implement stats in software? We should have access to Kramnik's stats to be checked.
Kramnik's deep, droning, voice could be good to fall asleep to.
Babbling ASMR
i like the idea that Hikaru is cheating but he's too dumb to know that he is
I like erics fancy interviewer sweater
Around 30 minutes in, Vlad repeats the phony "evidence" that players in important final rounds perform better than they do in unimportant rounds. Even Caruana fell for this tautology. Of course players who reach finals in tournaments are going to have higher performance ratings because they are playing higher-rated, more competitive players in a situation when both players are playing well.
At 2:07:00 Kramnik discusses Hikarus rating. He doesn't understand how Hikarus can have such a high rating when he plays all the time.
He doesn't agree that Hikaru gains rating from farming low rated players. So for each player he should examine two numbers.
* The ELO points gained from opponents with equal rating
* The ELO points gained from opponents with much lower equal rating
At 2:25:30 Kramnik repeats that someone should probably be suspended for Hikaru's winning streak.
At 2:30:00 He complains that Hikaru is misrepresenting his intentions.
Your motive turned from cynically attacking a particular player to the benevolent motive of addressing cheating as a widespread problem. You did this because you got called out and ridiculed, and now it's the only way to save face and make yourself look good. At least recognize what you're doing, Kramsack.
Kramnik is the oldest 48 year old I know of. Who leaves the green screen on default like that. I think Viktor Korchnoi would be more comfortable with modern technology.
At 2:12:00 Kramnik just rambles in denial, and it is weird. For anyone who has seen Hikaru winning several games in a row for years like me knows that what Eric is saying is obvious - Hikaru wins lost positions on time so many times it is not funny, and the assumption that his opponents won't blunder or get nervous while having to deal with very complicated positions in time trouble just because of their ratings shows a total lack of understanding of the dynamics of online Blitz with no increment. I even have games with 2750 estimated rated at 94% that should be statistically impossible for Kramnik. He assumes that people should play the same way all the time, that their ratings should be perfectly stable, and that it should all be ruled by some program that can simply declared someone guilty. Maybe he's been watching too much Minority Report, to say the least.
Why is he even giving this guy the time of day now? Kramnik has been demonstrated to have been incorrect in his "math" and insisted on continuing.
Respect.
Gotta give respect to a world champ. Also kramnik, maybe incorrect in this instance, is well intended. He wants to protect the game he loves. I don't mind hearing his viewpoint at all.
@@kunalsingh4418 Agreed. Even if we do not agree with what he is saying, you listen out of respect, especially if you are a fellow GM. Almost all of the chess greats eventually struggle in some way as they become older. But this is a part of life and times will always be changing.
2:06:06 he said they can check who is guilty or not, but they can't say it's "normal" after they check, which means they have to keep checking until they find guilty. in fact, chesscom and other professors and professionals already checked, and he said they were all wrong. btw if u read his posts, he is always implying they are guilty.
2:06:36 ok is Eric just not up to date or is he helping Kramnik on purpose? the whole point is kramnik doesn't understand how the ratings are inaccurate and how that makes all his points bogus.
Grischuk gave a very interesting interview to Levitov, where he gave a lecture on basics of theory of probability in context of recent events with cheating. Very interesting. The conclusion he gave is that, in theory, we could count the probability of cheating with a 100% accuracy, if we have values for all variables in equation, that he presented in that video, but the problem is, there are couple of variables, that are impossible to figure out with current methods. Then he just shows how it works by giving some numbers that he randomly came up with.
there's probably no chance this is in english huh. i'd love to see grischuk discussing this tbh.
@@ayden492 yep, unfortunately it’s in russian and no subs
When you say (Grischuk says) "we could count the probability of cheating with a 100% accuracy," what is it you think you're saying?
@@pat_athens Well, fine ... with 'almost' 100% accuracy. The point is, it would be pretty clear, if the player was cheating or not, if it wasn't for one thing ...
without having watched the interview, you can cumpute the probabilities of streaks exactly if youbknew the winning probabilities of each game. But the best you can do is estimate these probabilities from past data, which is how statistics works. This has been done. None of this proves cheating though, it can just give you a way of quantifying the likelihood of cheating.
It is very clear, that while cheating does take place (being a reality) at various levels, what Kramnik (and Nepo) really object to is the "Farming"...as they and other sporadic high level players that only play key online events, do not do this. Farming has occurred at all levels for many many years, whether at the very top (3000 plus players farming 2700-2800s), 2800 players farming 2500, 2500 farming 2200, and so on. Players do this to increase their ratings for a variety of reason, and I am speaking now specifically for online platforms. It doesn't matter to Kramnik/Nepo that some GM's do it for content or not, and of course it can equate to $$$, which again, Kramnik and Nepo do not enjoy having, or player Farm for other reasons such as 'status' and respect..I guess. While Cheating is an underlying issue that certainly needs to be constantly addressed and I believe is, to some regard, the Kramnik's and Nepo's (etc) think farming, if not cheating, is unethical.
He's talking more about a variety of parameters he has assumedly tested against players, mainly performance in lost and drew games player vs player which is a big thing if certain patterns are found. data is very interesting tbh.
I checked a couple of my recent games. I lost one with 30% accuracy and my opponent had 36%. Another, I won with 85% and my opponent had 72%. Both games over 20 moves. Did I cheat?
Kramnik deserves more respect for his work. The Hikaru kids army is one sided and rude
1:52:55 Eric, did you just call Vladimir, "Babe"? 😂
AHAHHAHAHHAHAHhaahhahahhahahahhaHAHHAHHAa
He kinda is being a b tho
One the one hand, it is truly impossible to imagine people like Hikaru, Danya, Bortnyk cheating. On the other hand, It's sad that Kramnik chose Hikaru as his target of all people, because cheating has become more and more frequent in every format of the game. Even in regular bullet and casual games, without any money rewards, somehow people still manage to, and are inclined to, cheat. Just today I played 5 1700 rated bullets, 3 were fair and I lost; 2 of them were played by 800 Clasically rated players, with 97 percent accuracy who played precisely the same opening moves as each other,(Caro exchange) and had a similar profile. I wish the former world champion would go after the real problem, that does exist, not a phantom.
Before I comment: 1) It's Xmas eve and I have nothing better to do. 2) I'm a Hikaru fan. I know most of y'all hate the guy, but I'm curious to hear what Kramnik thinks.
So after watching the entire video, it appears to me that the man wants to launch his own anti-cheating business and this is his way of marketing it - by using the biggest chess platform and sending accusations at the biggest chess streamer out there. I wish they'd dug deeper into Hikaru instead of saying it's just mathematics and nothing more. Kramnik himself admits that Magnus and Hikaru are capable of that level of play. So why accuse Hikaru of cheating if there's nothing personal about it, when Magnus also demonstrates weird stats.
3-0. Kramnik vs Hikaru, 100 games. Let's see how that pans out. Go on Kramnik.....let's see how well you do. Let's see your rating at the end of it.
He discredits himself within his first thought, "I've only started playing online chess maybe a year ago"... He clearly has no experience in online chess and is not aware of the difference between OTB and Online chess. Hikaru has pointed this out adnauseam. Krammnik is desperate for attention hence the pink shirt. Good day.
It seems like something Kramnik totally misses is not just the chess skill, but different aspects like simply how quick/accurate someone is with a mouse. It's got to be nearly impossible to cheat in any way at 1 0 chess for example, but players like Naroditsky, Andrew Tang, Minh Le, Bortnyk and others maybe 2600 ish OTB players can hang with the like of Hikaru and Magnus in some cases.
Even Eric Hansen would likely crush Kramnik in bullet chess, but might get beat regularly once time controls get above 20 minutes or something.
Nobody said strong gms don't have a chance to beat top gms. Kramnik claimed rather the opposite that streaks as Nakamuras against good players should be impossible.
@@paulgoogol2652 But the same people that are strong online bullet players are not the people that are top GMs (usually). Andrew Tang for example is an exceptional online bullet player, he has incredible speed and flagging potential. But he's not a top GM. The skillsets are very different, there are mind tricks at play in online bullet that isn't there OTB or in slower time controls, mouse speed is also a hugely important factor. So who's a top GM and not is kind of irrelevant, you'll have IMs that are much stronger online bullet players than some top GMs.
So ... is he saying his "math" is wrong cuz Elo ratings aren't what he thought they were cuz of farming?
I'm about to watch the entire 3 hours and a half of this video. I will do it while playing some game to avoid suicidal thoughts. Wish me luck, guys. See you on the other side.
how was it, did you survive?
I wonder how long until someone crops Kramnik and uses the green screen for shenanigans.
listen up yu nakamura fans. kramnik was champion of the world with IQ 19o. he would not talk rubbish.and of course he is not alone if Kramnik says that Nakamura cheats i belive him
A couple of things really stood out to me. First, Kramnik saying that there is no need to prove cheating beyond a reasonable doubt which is just insane to me. The other is his claim that all the backlash he has received means to him that he is on the right track. In terms of Hikaru, he is completely missing the real point of the backlash and that is the fact that his argument is so clearly and blatantly wrong that people can’t help but try to tell him this. Obviously any push against cheating is a good thing but the way Kramnik is doing it is just wrong. His complete lack of understanding of the online chess world and his willingness to call out specific chess players without giving any actual concrete proof is ridiculous. He keeps saying that statisticians and mathematicians tell him this that and the other which is either a lie or the people he is consulting are just like him, without any knowledge in this field at all. When will he realize or when will someone tell him that his logic is a fallacy?
People in comments are insulting and attacking Kramnik personal but cant show anything wrong in his ideas, as he is describing an AI anticheating system which is now used in many games. But who are those people which are attacking him and the whole anticheat idea?
43:35 "you don't even need to look at games" - oof
I wonder if he bothered to look at those 89% accuracy wins by magnus where he opened a3 or h3
My feeling is that good cheat detection will be looking at players who find genius moves in 5 seconds and then take the same time to play obvious moves
He did. He explained it.
At 2:12:15 Kramnik mentions that whenever a player loses 4 games in a row, then the chance of winning the 5th game is surprisingly high.
This is actually interesting.
Apparently this means that players start cheating after losing 4 games in a row. (or perhaps they leave the computer and continue playing the next day)
2:10:00 Kramnik is correct here. Elo does have percentages baked-in, and we should use them. Also he alluded to the fact that even if you play against an average of -300 Elo players, winning a long streak (no draws) is very difficult.
2:11:00 Psychological issues go both ways, and one cannot assume they ALL favor the better player. In fact studies show farming is generally not worth the better player's time because on average the lower rated players tend to play slightly above their level. That's the "psychological" issue with some studies to back it and not some "well, as we all know" comments. All other comments that make assumptions are not part of any statistics.
You either trust Elo percentages of you do not. If not, then you're not an important part of the statistics conversation. If you do understand/trust Elo, then Hikaru's win% for the 45.5/46 streak was not 90% as some "mathematicians" claimed, but between 73 and 80. This makes all the difference in the world. Even using 80% (helping Hikaru out as much as possible, i.e., assuming his highest rating and keeping his opponents' ratings static) it would be very unlikely to achieve 45.5/46 even after 10,000 trials/games.
If Kramnik would simply give what he found to be the assumed win% for the Hikaru streaks, then all of us (you don't need to be a mathematician) can calculate the likelihood after x trials.
2:13:00 A camera, sure. But note that according to Hikaru is would be easy to cheat OTB, but that is coming from a guy who wears headphones while streaming. Headphones would be the easiest way to cheat. No, I'm not saying Hikaru cheats, but given the two situations of OTB and headphones, of course the easiest way by far would be headphones. So very very simple to cheat that way. So incredibly difficult OTB. After Hikaru's uneducated comments about Hans' OTB play there's a bit of schadenfreude here.
2:16:00 "FIDE blitz ratings are not very accurate because..." OK, so do we assume Hikaru's rating is too high or low? Do we assume his opponents' are too high or low? This gets us nowhere except to say we cannot be certain Elo is a good representation of the player's abilities. That may be so, but until we know more we use the known ratings.
The 45.5/46 streak had known blitz ratings. We have known/accepted win/loss/draw percentages for such games. Why don't we use those Elo ratings and the resulting win% for the probabilities of such streaks? Otherwise we are left with "it's possible to have such streaks because it is possible".
Hikaru literally shaking rn
Kramnik UNDEFEATED
Hikaru has a pretty good record against Kramnik, just off the top of my head, Hikaru beat Kramnik in the final round of titled Tuesday 3 months ago for 11/11
@@InfinityPowerForts very interesting
we do the procedure
accuracy measures consistency or precision in stastics. this is how much variance moves a player makes between games. in philosophy accuracy is the reliability -a measure of the objectivity of each primise.
chesscom use of the measure of accuracy is the amount of top engine moves so it changes the fundamental common uses of the tearm accuracy.
what i think kramnic means when he says accuracy is its intended meaning which is is the player making common types of moves for that player in all games measured. kramnic is saying that there is a flaw with hikaru opps in that they dieviate with a low reliability measure.
niw here is the problem… kramnic needs to test for reliability within group variance but instead he is using validity (logical) measures -comparisons if engine to make these claims and now you have the dreaded type 1 error (accepting the null when the hypothesis its false)
summary of 3.5 hours: trust me bro...
So, is Hikaru using beads or not?
yes, but he didn't cheat
Only when he streams from his wife's room on behalf of his wife's son request.
Magnus accuracy less than 90 is cause he trolls with bong cloud, A4 and H4 openings lol. Magnus is so good that he can still beat top gms even if he makes 10% inaccuracies in the openings lol
I like that he has raised questions and that no one is above suspicion. Good for him. Please never cheat people. Thanks
No, cheating is good ...
I have no incentive or ambition to get good at chess and neither do you or other 99.9999% of players ... Coz none of you are going to be GM ... If you are above 8 years of age
Hence cheating is good ... Chess is not my craft , i don't cheat in my craft , there i try to improve my skills ... But chess is not ... Hence I don't care ,i just enjoy chess & of opponent is talking trash to me or is showing off , then I show him his place 😅
@@speed999-uj5kr Just shows how emotionally weak you are that a little trash talk and you have a tantrum flip the table and ruin the game like a child, doesn't prove anything about your opponent.
And whether or not you make a career of chess if you're not striving to improve (which by itself will improve you as a person and in other aspects of life) you're wasting yours and whoever you're playing against time.
@@speed999-uj5krwhat tf are you on about? 😂
@@mendikk6386 doesn't matter ... No one in this world is 100% truthful ... Everyone cheats , if you go office late by 2 mins ur cheating, if u lie to your partner u r cheating ... So everyone cheats ... There's nothing wrong in it ...
I feel like I understand his argument even less now.
haha such a good summary for what I've watched so far
more importantly, I feel like HE understands his argument even less the more he talks about it
Three and a half hours? Can someone give me the bullet version?
Bedtime episode
kraminik shows a WGM level of communication "skill", but his points are real GM level
tl/dr: he wants chesscom to be more aggressive and admit there is a big problem. his hikaru streak point raises 2 hypothesis: A) hikaru is cheating, or B) hikaru's opponents have cheated at some point to get such high wrong ratings... and he _explicitly_ says that B is much more likely the case
@@zaftnotameni What is that sexist WGM comment?
@@tsevasait’s not sexist, WGMs just aren’t as strong as GMs. hence the reason why he used WGM to explain his subpar communication skills.
Thank you for bringing Kramnik on. I think he is being treated to harshly for trying to solve a problem. His ideas aren't perfect but he is well meaning. After watching the Dubov and Caruana interviews they see the problem as well but haven't really brought any good ideas to the table to solve the problem. Kramnik is taking notes, dealing with statisticians, and has some ideas.
His mistake was posting about some players and their rating performances and a lot of players took it the wrong way (For good Reason). I also don't like the idea of anyone calling out minors when they don't have a way to defend themselves like Naka was saying but Kramnik, Dubov, & Caruana are more RIGHT then wrong in my opinion.
How many times the word "statistics" was pronounced in this conversation?
I like how Kramnik says take checked the statistics and the numbers don't lie, but he DOESN'T give the numbers
He said he sent the data to chessKom! And they didn't agree with him even though an average Joe person could see he is right mathematically.. while he never went to high school himself
@@angosalvo5734 he is not right mathematically, there was a DOCTOR IN STATISTICS of the university of Chicago that made a while paper into that, Kramnik has no right talking about stats over a freaking PhD in Statistics lol
I'm also a maths major, Kramnik as such a Simplist view on math and stats that only a high schooler could, he should keep to chess and let actual mathematicians handle the math
@@xderyck41 sorry, I was sarcastic
@@angosalvo5734 Sorry, I'm autistic I can't read sarcasm - I TRULY AM SO I'M SORRY AS WELL lol
I like the idea of secretly matching someone suspected of cheating against an engine (without them knowing, of course). If they manage to draw against Stockfish then that would be evidence, would it not?
Depends on the specific game. Some closed games maybe a human can hold.
@@Monika77ful If a person is suspected of cheating AND also happens to "hold" against stockfish (once, twice, three times?) then that would be good but not perfect evidence of cheating.
Why does Kramnik have a green screen behind him?
His polo shirt is green, the screen is pink
What if someone cheats only on a couple of specific key moves every other game? Would it be possible to detect that?
Can a chess computer learn to identify key moments? That could be an interesting challenge.
If you actually listened to the whole interview you would know that a lot of statistics is based on certain game-changing key moves.
This is my new go to video for falling asleep to.
100% spot on the over performing. If you are performing better than 99% of people and actively compete for money prizes then you 100% should be checked!
flip 1 million fair coins, somewhere in there there will be a streak of 20 heads in a row, exclude 999000 flips, 20 heads in a row in 100 flips? HMM INTERESTING
Had this on the Casino. 15 Red in a row out of 16 😂
I find Vladimir's explanation for his post about Nakamura's streak interesting (no pun intended). I always thought that it was inevitably a direct accusation or at the very least an expression of suspicion against Nakamura. However it does make sense, like he says, that if many ratings are inflated through cheating, Nakamura would get much longer winning streaks than usual and much more often. This is simply because cheaters would have such high ratings from playing against other people, while they would not dare to cheat against Nakamura.
Cheaters are afraid to cheat against Nakamura.
Fantastic interview Eric. Very interesting and constructive discussion. Refreshing to see.
Much appreciated!
@@chessbrah make a shorter version covering crucial points
dafuq? no lol
16 minutes on the video and I wish you asked what statistical analysis he used, what kind of data structure he used to run the statistics because, I don't believe he actually used statistics. He would need a big database of results and typing these into the computer would take whole day if not more. He looks like an old-school dude who doesn't use the computer much as he recently met with online chess after hearing it from someone. If iI could just download Hikarus data I would have run the test myself. There is a high chance he was biased, he used inadequate data, or his data was falsely structured. Anyways, anyone who watches Hikaru would know he is not cheating, I think this is an unnecessary debate. Also, humans are not machines, there may be good days or bad days.
I caught bits and pieces yesterday on stream and he said he was planning to make his own anti cheat system. So until that or of that ever develops, he has shown zero serious analysis. I was also curious as someone with a Mathematics degree but he said something about 5 pages of calculations no one would read. Tldr he has produced nothing rigorous at this point
I actually understand Kramnik's point of view and I agree with his sentiment! There seems to be a lot of hate from fanboys who clearly want to misinterpret what he's trying to say. Kramnik wants to use statistical anomalies in play to categorize players so you can closely monitor them. IF someone has a crazy streak, they are more likely to cheat, so it makes sense to monitor their play. It's not accusing them of cheating, it's saying "You did exceptionally well, you played way above your usual level, so we'll secretly more closely monitor your play to make sure you're not actually cheating".
Also anti-cheating requirements for online tournaments is also a good idea, especially having to download a program. Most other online sports require PC checks for all players who are accused of cheating, so it makes sense.
Who are the fanboys fanboys of?
Hikaru Nakamura
Not sure I understand.
If Hikaru goes playing against lower rated players, I guess he's likely able to go on a very long streak, it's is legit. Hikaru cannot have a such streak in titled Tuesday against strong players like he did admit it.
@@angosalvo5734 Read OP
His approach is plain wrong and naive. Anomaly detection by itself is too naive and noisy. Real life is full of anomalies, so today ML is used in conjunction with other strategies that can more definitely prove something. For example, just try building a stock trading system using only anomaly detection, and sit back and watch all your money dissappear (which ironically is what is happening with Kramnik's reputation)
30 min in - yes even 1500 elo player are having accuracy at 90+ at some times, especially if opponent plays into in (that is make easy game for him) but that does not means he is Magnus. Is he really confused about what high accuracy game means (that is not making mistakes) or am I missing something there?
I think Kramnik has a very great perspective for online chess because of the amount of time he has spent playing offline and getting introduced online so late in his career. He clearly sees a level of play that isn't consistent with playing at live tournaments so he's aware there's a lot of cheating when he experiences level of play he hasn't throughout his career. I think online chess opens the door to a lot of talented players who otherwise cannot participate in live over the board chess. Also obviously the online format favors far faster time variants like blitz or bullet that is a different style from a classical player who spends less time playing blitz.
Kramnik ruined his reputation buy accusing Hikaru of cheating. He brought this on himself.
On a positive note, he should be awarded "Chess Content Creator of 2023".
He probably got paid by Hikaru to say all this stupid shit so that it generates drama and views.