There are two types of people in this world: those that return their shopping cart after they're done with it, and those directly responsible for holding humankind back from its true potential.
Also those who live in areas where they've create make work programs to have otherwise unemployable people spend their time collecting them. Or people who live in areas where you rent the carts for a dollar, and get back the dollar when you return the cart. In which case the free market has solved your moral quandary for you. You're welcome.
Tragedy of commune is not applicable if the job is staffed, obviously. Also, I'm pretty sure the original comment was not that serious in its core that it makes sense to be heavily countered with real social programs... And those are not exactly terribly common anyway, being more of a welcome edge case rather than a counterpoint.
Interesting thing about bats: They have names for each other, gossip, and the things they talk about are food and personal space. Bat children tattle-tale to enforcers about who isn't feeding them!
I remember discussing the tragedy of the commons in philosophy class, and it’s interesting to see it pop up in situations like these. It’s hard to express how captivating I found this rant about a game I’ve never played, and how far the implications of this very concept of cooperation in search of a better world stretch.
surely this can't be a benchmark for the failure of cooperation when it's a tournament game where your goal as an individual is to win over others. think tragedy of commons suffers from explaining the contradictions of its own internal game, not cooperation everywhere.
that is to say the fear of free riders presumes individual aggrandizement is the base unit of society, convenient to a society who's organizing principle is capital accumulation for a wealthy few.
Loved your rant. Main way to convince myself to do things like picking up the thrash or voting is to find mental reasons to be happy about it. Like being proud, being able to talk to people about it, etc. So I don’t try to focus on if it gives me more value purely from the situation, but I try to create a bonus reason for my mind.
As a FilthyRobot viewer. this almost perfect describes civ5 lobbies. When a player starts an "irrelevant" war, both players suffer and neither benefit enough to go on to win the rest of the game. 4 other players continue to produce buildings instead of units and not punish the player for their choice. If you remember those civ5 lobbies you'll know exactly what i'm talking about!
Small correction but tit for tat doesn’t permanently defect against someone who defects once, which you sort of implied. It’s more accurate to say tit for tat starts cooperating, then always copy your partners last move. The reason is even if you get into a defection cycle, if you allow for some “forgiveness” the tit for tat algorithm can go back to mutual cooperation. Great video though!
I agree with your assessment from this video from a theoretical point of view, however, I think the implementation of your ideas are where this strategy doesn't work and why most of the time people try and blanketly use the prisoner's dilemma and the tragedy of the commons it ends up not working. The best showing of the tragedy of the commons is this picking up trash in the park, however, each individual choice is not based on the tragedy of the commons, rather, the tragedy of the commons is a weighing factor in the choice. Someone who is walking through gives it higher weight to the commons than a mother trying to guide kids through a stroll but maybe not as high as someone trying to impress a date. Theoretically, we want to be able to condense each player's situation into one that is negligible, but it is not necessarily negligible and the weighing of all other games that the player is playing may have a larger effect on the overall decisions. In my limited interactions with tournament poker this is what I understand to be the case. Having enforces would be theoretically be beneficial to the tournament as a whole, but when many tournaments, especially online, either have high stakes or a fast seat rotation such that playing as an enforcer is not worth the personal EV in the case of high stakes or the reputation of being that enforcer is hard to develop for all but the larger names in the game in the case of the rotating seats. This discouraging of enforcer behavior is exasperated by the fact that people that are trying to enforce are given very little opportunities to do so. This is because people who enforce a high percentage of the time are more likely to leave the tournament faster and thus out of the money and looking for another job, or if you enforce a lower percentage of the time, i.e. widening your calling range on the BB, you may not even seem like you are enforcing, just playing a little loose. This is a good conversation to be had among poker groups, and depending on the volume of people playing like this (the more enforcers there are the higher EV there is for being an enforcer) the decision tree may change, and maybe even eliminate the need for one eventually, but the way tournaments are structured currently there is not much reason to be a lone enforcer. TLDR Theoretically there should be enforcers if tournament poker was optimal, but the way tournaments are structured it gives harsh penalties to whoever tries to enforce so it is not a smart long term strategy. Great video and I'm not the best writer so I can clarify if need be
i agree mostly, but think that what you have typed is very sad and depressing xD. it is concerning that the structure of the tournaments is enough to dissuade people from pursuing an optimal collaboration, especially when we consider that a lot of our other interactions with humans are in artificially-structured environments which reflect some of the sorts of things going on in tournament poker.
@@scottwhitty72 That would generally fall under collusion if people went into a tournament having prior discussion of specific tournament play. The way this problem is solved seems to be change the game in some way, not the strategy
@@corygodin4129 It doesn't have to be collusion, if everyone is of the opinion that enforcing is the best course of action for everybody (and thus themselves). It's just enforcing the enforcing. By your logic enforcing itself would already be collusion.
The "Tragedy" described in the Tragedy of the Commons is that there is less total time or energy resources for all parties involved. I believe that your analysis only holds for a closed system (the two players described) but is irrelevant when considering a broader system like a poker tournament pool or the entire poker industry. As you described, the prizes are set and there is no actual table-wide loss of money due to demolishing the other guy. If one person is guaranteed to lose, the expected value of the remaining players goes up, but the sum remains the same. In essence, poker players cooperating optimally will not increase the average expected prize value, so there is nothing to cooperate towards. That said I had never seen the Tragedy of the Commons described so accurately as multiplayer prisoners dilemma, and I loved this video.
I always love the analytical content on your channel. I've never played poker and rarely play slay the spire these days, but the concepts you bring up in your videos are almost always things that are relevant across a wide range of scenarios. There's a strong correlation between spending more time watching your videos and less time coasting through life with my brain on auto-pilot. It's an unprecedented tragedy that videos like this one aren't more popular.
When you use a strategy of reciprocation in a prisoner's dilemma, what you're essentially doing is using your future actions as a bargaining tool to make it rational for even a selfish rival to also cooperate. I could see it being frustrating to try to accomplish that with a rival who refuses to consider they're in a prisoner's dilemma at all. Apropos of nothing, there are are some really interesting variants of prisoner's dilemma, like one where there's a small random chance to "accidentally" betray the other party without being aware of it yourself. In those situations, the optimal strategy involves forgiving the occasional betrayal so you don't betray back and forth forever because of a misunderstanding.
i think that extra comment is very apropos! this sort of scenario would have that exact sort of thing going on, because when SB gets dealt a strong hand like AA it's actually reasonable for them to play aggressively - the punishment for getting involved in a big pot is far outweighed by the rewards for shoveling money into the middle with the best starting hand in no limit hold 'em. this is a like... semi-iterative prisoner's dilemma which you can only tell that you're in probabilistically, since you don't know what cards the other player has. it's very weird. i think it's complex enough to be subject matter for a phd or something.
To note, you do see a lot of cooperation in situations where ICM is really a huge factor. ICM is at it's biggest when the poker tournament is going from prize of $0 to whatever the minimum prize is, but usually a big jump and always more than players put in to join the tourney. Everyone already has big incentive to play very cautiously and not be the last person to get $0. But also, each table as a whole has incentive to be the more cautious table compared to other poker tables in the same situation. Interestingly, and in some ways hilariously and in some ways sad, the people who have the biggest incentive to betray their table, are people who have the most money. Be it the person who has the most chips on the table and so cannot be eliminated in any all in, or the person whose bank account is biggest and has least to lose in a relative sense. Often know as bullies, these few abuse the players trying to play by the societal and generally sensible rules of ICM. The rich (generally, poor players also get good hands and lucky) get richer. If talking about the tragedy of the commons, these are the people dumping into the river.
this rant brought me great joy! i rarely get to listen to someone explain Tragedy of the Commons or provide insight regarding the facets of game theory or other similar theorems and your means of talking about it makes it all a lot more digestible, cheers!
I've played a game of Settlers of Catan with family where I realized I couldn't win, but I could choose whether to devote what remained of my resources to mess with one family member who had annoyed me early in the game. Playing Kingmaker. And of course I tried this because as a kid why wouldn't it be in my perceived self interest to get back at someone who annoyed me instead of politely folding? And it became this whole thing because everybody else did innately realize that this was, in a way, collusion with the player benefiting from my decision. So it changed the way I thought about games forever. So when you explained that both poker players stood to lose expected winnings by going toe to toe with each other, I immediately understood where you were going with it. Though of course you did give it away in the title. Taking advantage of two other people devoting resources to fighting each other is a time honored strategy, so anyone who knows this wouldn't want to get caught with their pants down in, say, a strategy or grand strategy game. Playing to win the tournament, not just the hand, means thinking strategy instead of tactics, the same way that ICM does when weighing chip values for tournament play. So it's odd that people who did actually understand ICM, didn't rise to your positing of the prisoner's dilemma... You'd think they'd be the first ones to see it.
7:40 This line of thinking reminds me of macro strategy in a battle royale. An early game fight is likely to reduce both players outcome on average. Big potential loss (losing) for a small potential gain (more aggressive looting capability). Generally stealth is preferred up to a certain point.
This is why top pro players will always fold their first few hands, even if they get pocket aces. It's not worth risking a dime that early in the tournament even with a 90% chance of winning the hand.
One thing worth noting that I think a lot of people miss - tit-for-tat strategy in prisoner's dilemmas tends to perform pretty well even against predetermined algorithms. Its success is not strictly behavioral or psychological. Love the video!
Good rant. One thing about tournament poker (or anything else structured similarly) is that it has very spiky results so it's easy to cash out on trust for a huge sum. It's difficult to establish a trust in, appreciation for, and loyalty to the commons when you're only going to be playing a few games with these people who are about to be taking your money. It's still wild that they have no interest in even talking about it, and call people "emotional" when they act as an enforcer when someone starts routinely bullying the table into folding etc. But I can see how in a game where people tend to be "strategy gamers" they'd have a visceral distaste for anything that required long-term thinking and building trust for anything besides a backstab.
That was an incredible explanation of fantastically cool topic thats horrendously relevant to how I live my life, and all this from poker! All the while, I never really thought of myself as a freeloader, but this rant gave me a bit more pause now before I just choose inaction now. Thanks!
I expected the tragedy of the commons to be about card rarity. The video ended up being far more interesting than I expected, great rant and explanation.
I don't see how the tragedy of the commons applies here. The price pool always remains the same, so there is no difference in outcome between a scenario wherein everyone collaborates compared to one where no one does. Starting a collaboration comes with a risk, and it stops being beneficial as soon as all players work together. The bad outcome of investing some recourses into starting a collaboration is not just that you might get betrayed, it's also that everyone might join and you gain no advantage for your investment. This means that collaboration only works without forgiveness: the advantage of collaboration is only an advantage over players who don't join, so you have to ensure not everyone can join. If you start offering collaboration and maybe even act as an enforcer you have an initial investment to demonstrate your trustworthiness, a period when you have an advantage because some people collaborate with you, and finally the end state when everyone knows your playstyle and you stop benefitting because everyone collaborates with everyone else. I feel like with a group of professional players the middle period, the one when you actually benefit, might be really short, to short to make back the initial appeasements.
Playing the enforcer is a very hard life and anyone who does understands this distress all too well. Very obviously this has nothing to do with poker and is simply a medium to convey the nature of humans and modern society. One who has experienced life as such can confirm that it is all cost, no benefit and the only up sides are the retention of ones pride, dignity and virtue. Though these things are invaluable and the feeling of changing some ones life exquisite, it is very hard to afford a home on good deeds alone
I have a lot of thoughts on things similar to this, most especially your thoughts applying this very specific poker situation to the wider world. Sadly, I'm not particularly excellent at putting them into good words and I've been awake for almost 24 hours and am slightly delirious at the moment. The broad sense of it, though, is that I worry that evolution primes people to think in self-interested fashions; in the short and middle terms; and to emphasize certainty over ephemera. And I strongly suspect that the vast majority of people are largely acting on evolution (selfishly) and really just doll it up to seem "rational" or whatever their preferred explanation for their behavior is. In your poker example, the enforcer has to take *serious* short term hits (like being knocked out of the tournament, reducing their expected return from 3.50 to like 2.25ish) and their return is dependent not just on their behavior but also their opponents perceiving their behavior, identifying their strategy, and modifying their own strategies to suit. It's a very ephemeral return. The tit for tat strategy relies not on predicting your opponent but on reacting to them; the past is a sure thing and the future is not. A more personal note on the weight that ephemeral returns has on behavior: I've been heavily self-monitoring how much effort and thought I put into writing this comment on UA-cam because "what are the odds anyone reads it and cares?" "Will someone say nice things about it and gimme that sweet dopamine hit?" "Will I get the coveted heart-from-the-video-poster-who-is-Jorbs-a-person-that-I-think-is-very-smart?" These are all weights against me writing semi-publicly about something I think about, like, all the time. One would imagine the barrier to doing that would be pretty low but even now my fingers wanna press ctrl-A delete. Too much risk, says my brain.
I like the way you analyze information. I believe you pick these things up faster than most. You are also able to explain very well what you have found and why. I came here to learn some slay the spire but this is my favorite video you have so far. Also kinda mad at how easy you made Loop Hero look.
Here's an idea: If you think being an "enforcer" can be profitable, you should explore that situation. Use a tool like ICMizer find a single scenario following these steps: 1. SB can push really wide at equilibrium because BB has to fold tons. 2. Adjust BB's strategy to call with losing hands in order to prevent SB from jamming so much. 3. Resolve for the SB strategy given the locked BB's strategy. 4. Check that BB's overall EV has improved. This isn't perfect since, but I think if you find a situation in which the above is true, then there might be merit to what you are saying. Intuitively it almost seems plausible, where in GTO you might have a hand that is slightly losing but including it in your range greatly improves the EV of your range. It seems strange for calling an all-in because you don't care about obscuring your hand anymore, but maybe.
To me it sounds like there are misplaced incentives in tournament poker that make it deviate from cash poker in unexpected ways. I think one of the issues with tragedy of the commons problems is they are somewhat created by a blindness to that type of problem. Very insightful video.
Any tournament system adds an additional layer to the game that shifts incentives. Even outside tournaments, the fact that players play more than a single game in their life adds a layer to the game and slightly changes ideal strategy (like the iterative prisoners dilemma jorbs proposed)
Great talk. I've always had this feeling that these things were things, but I've never been able to get it to more than a feeling. You're completely correct and it's one of the great themes when people start talking about the question of "are humans fundamentally decent" or not. Luckily I've found that being a decent human being doesn't necesarily have anything to do with being a good strategic thinker, so we're probably fine if these great minds don't understand this. Most people seem to understand it on some level, even if they could not put it to words.
Interesting video, I spent many years as a (primarily tournament) poker pro and think I can share a little insight into this situation. There are two big reasons why these types of quasi-cooperative metagames aren't really thought about much. First is, as you mention, people really try to avoid even giving the appearance of colluding. The second is simply that tournaments have a large player pool, and unless you play the highest stakes, you'll rarely play against the same players. Given that showdowns are relatively rare, it is very difficult to ascertain whether or not a player is 'cooperating' in a short amount of time. So any implicit agreement to play a small pot is going to open you up to exploitation by an aggressive opponent, but you won't know for sure whether you are being exploited since, as you mention, they might just be getting dealt strong hands. And given that you may never see that same player again deep in a tournament (ICM isn't really a factor until the late game), it's very difficult for cooperative strategies to gain traction. When I played high volume, high stakes sit n go's for a while, I did actually see some of these types of meta-games becoming a factor among the regulars (which shouldn't be surprising, because in that environment it is both a much smaller player pool and you are constantly seeing the same people, and also, ICM is much more consistently an important factor).
As someone who presently follows and plays tournament poker I think I can say that the community as a whole has recognized and applied some of these ideas that you mention, and that poker theory has progressed in ways that would incorporate a lot of the suggestions you mention. The first example is related to what you mention at around 13:00 where you mention small blind limping and playing a hand very passively. My understanding of current theory is that there actually are quite a few spots depending on stack size (15-25 BB range) where mixing in limps is important because of how brutal it is to raise and then fold to an all in. This especially applies to the situation you mentioned where neither player has a strong hand. Of course this can be exploited if the SB only does this with weak hand, but the way to counterbalance is to choose some number of relatively strong hands where you limp and then shove or raise over the BB if they raise you, discouraging them from "cheating," or to use a phrase that many poker players seem to be fond of "catch them speeding." Players also tend to size down their bets and play fewer raises in general in situations with high ICM implications. They'll check more in spots where they might've tried to value bet. They might play a hand that they would've 3 bet preflop without ICM as a flat. 3bets will be sized down from 3x to 3.5x the initial raise down to 2.5x to 2.75x the initial raise Going back to the original example of limping and then checking down this can and does happen if no one hits anything postflop. Granted if they do hit a strong pair or other hand then they'll still bet and play poker, but often with a lot of the other gameplay modifications I already mentioned. All of this tying back to the idea that you want to lose the minimum and reduce some variance in these situations. Some of these checkdowns may come across as softplaying or whatnot but it's generally understood that the best option for both players is to avoid putting in more unnecessary chips and they are still acting in their own interest, just in a way that aligns with each other. The second example is related to what you mention about enforcers around 16:20. I've seen this role played at high level tables by the player with the shortest shack or close to the shortest stack depending on how flat the payouts are. This player will of course still be affected by ICM but much, much less than a lot of the other players, especially those in the middle and can play much closer to just pure chip EV in terms of their gameplay. They might even be able to make the adjustment of calling much further down than they normally would at any other point in their tournament, especially against the biggest stack because of how wide they can expect the biggest stack to be opening/shoving. Further I've seen even some medium stacks play enforcer if they think a larger stack is sufficiently getting out of line. The larger stacks, while being able to pressure the smaller stacks because of ICM are not immune from ICM effects themselves. A lot of times it can be difficult to shove into a stack that will cripple you if you don't have a sufficiently high equity hand because although you will gain a small amount of chips if you choose too many hands to shove then even some of the middling stacks will be able to play enforcer and make lighter calls and you will commit ICM suicide by losing an unnecessarily large pot. To go into second level big stacks will often simply raise their opening range as a result and the antidote to this is often to try and punish them by making light 3 bets either all in or to a normal-ish size to provide counterpressure that they could be risking a significant portion of their stack. Overall I do recognize the idea that the situations you've described are in fact dilemmas and the larger implications it has on society in terms of wider recognition, but I don't think it's maybe quite as doom and gloom as you mention. I think there is some recognition of the problems you mentioned even if I haven't quite seen it laid out so plainly before as you have here. Either way I did still enjoy listening to your thought processes here and I hope this provides some insight into how people may have recognized some of the problems you've presented.
Something I see in a lot of games is that when players encounter a strategy they don't like because they see it as "impure" or "not what the game is about", they will sometimes try to tell themselves that that strategy doesn't actually work or isn't actually effective. It seems they want to see their game's rules as more robust and in line with their preferences than they actually are.
Has anybody here read Reasons and Persons by Derek Parfit? I think y'all would really like it if you dig the moral and logical complexities in these types of situations. The first half, which is the "reasons" in the title, tries to lay out a theory for why we act the way we do. Parfit pulls on a bunch of different flavors of prisoner's dilemma to try to suggest that it is often rational to act neither selfishly (in your own best interest) or morally (cleaning the Commons). It's a philosophy book but it leans on a lot of game theory and maths. Really interesting read!
The initial scenario about ICM reminds me of Nim, where there are some positions which are unwinnable not on their own, because you've just lost, but because you are guaranteed to be moved into a state which is the one in which you've just lost. So its more important as your options narrow to hold onto those few options, than it is when you have a variety of options
Thats funny, this different angle of playing is precisely the reason I only play tournament and sit n go. Cash games feel dry and procedurally based in comparison, whereas tournament you get to find an edge by figuring out how much your opponents are willing to risk in any given hand.
This is a very good video. What you came to realize (correctly) is that there is a fundamental asymmetry between two winning strategies, namely Position (acting LAST) and Initiative (betting FIRST). (Acting last is useful because you get extra information and betting first is useful because you allow yourself an extra way to win the hand, ie by making everyone else fold). Ideally - and this is true when stacks are very deep - you want to have both. This is why for example the BB would be at a huge advantage if the SB were to simply call preflop with deep stacks. The BB can simply raise (with a lot of holdings) and put maximum pressure on the BB on all streets. On the other hand, if stacks are short (say in the 10-20 bb range) the fact that the SB acts first is now actually an advantage because they have an option to go all-in effectively nullifying the power of position (This is what is known as a push-or-fold situation). The point is that depending on the stack sizes, one of the two players can prey on the other player's Expected Value without worrying about the prisoner's dilemma. In the first case, BB can punish the SB with impunity for trying to stay in the pot. In the second case, the SB should shove (again, with impunity) knowing very well that it is unprofitable for the BB to call even if they have a very good hand. Notice that there is a stack threshold at which the push-or-fold strategy becomes unprofitable for the SB as their risk/reward ratio become exceedingly unfavorable. (There are several factors to calculate this threshold but a good heuristic is that it is typically around the 15-20 bb's mark).
As a long time fan of your slay the spire content + as a poker player, I found this video incredibly interesting. I would really love to hear your take on cash games as well in a similar video format.
Wow, this is great. I have been thinking a lot about "spite plays" in cEDH in magic the gathering lately, and this shares a lot of the same problems. Basically, if someone attacks you for lethal with a 4/4 with a good ability and two 2/2s with no abilities. And you are at 2 life with two 2/2 blockers. Should you block the 4/4 and a 2/2 to minimize damage, even though you are dead anyway. Or should you double block the 4/4 to punish the player killing you as much as possible. Also, if you can disenchant their rhystic study or something before taking lethal, should you do that, even though it does nothing to save you, only makes the player killing you less likely to win. Generally the community says "no, don't spite play" but if you do, you could get a reputation as an "enforcer" who does spite plays. That will make people think twice about attacking you for lethal, in fear of reprecussions. So long term, it is the highest EV play to always spite play as hard as you can. Ofc in cedh it is a lot easier to solve, because you don't lose any EV by spite playing, you can only gain EV, the only reason to abstain from spite plays is a social one. So from an EV perspective it should be clear cut, but there is still so much discussion about it and people calling it "un-sportsmanslike" etc. Also, you are allowed to talk to the other players and make threaths before they make decision to attack.
The solution to that cEDH matter is "you should always play for team chaos". It's not un-sportsmanlike if you alter everyone's game. Doing so, even if you lose, you still change the course of the match. After all, making an impactful play is what informs any decision during the game, isn't it?
"All Americans are temporarily embarrassed millionaires." - John Steinbeck That is to say, enforcement behavior is looked down upon in highly individualistic societies. Which is sad. You can extend your argument from poker to tax policy or even for some people whether or not to get vaccinated... But to connect back to poker, the perceived and actual costs of enforcement behavior are too high, esp. in no-limit. A 40% chance of busting out (decent, but not great hole cards), just to have a chance at teaching the other player a lesson that may just be interpreted as poor play? In your picking up trash at the park example, the analogy would have to end up being like, picking up bloody needles... But your main point here really seems to be that poker players don't talk about this at all. Which doesn't strike me as weird, and I'll say why. (That said, I never knew this was even a problem until your excellent video!),. I don't find it weird because you're thinking of the prisoner's dilemma in terms of optimal long-term outcomes, but there are just far too many players in the pool that may be savvy game theory players on a per-hand or per-tournament basis, but they don't aspire to be a grinder. They want to win big. Too many players thinking like that, and too much money getting put in to the system on the expectation of (relatively) rapid gains. It's not like financiers bankroll poker players because they think they'll get a .05% better rate than a mutual fund... Sure, there are small time pros that want to make a living, but then an effective "enforcer protocol" for poker would essentially have to be insured by a pool of cash from like-minded players to cover the losses. Which is obviously collusion / coordinated action. So to reiterate, I think it's the external pressures from the system (expectations of outsized financial gains) without any path towards effective enforcement that lead people to not even talk about this, because there is literally nothing that can be done. Like you said, it's kind of sad. Anyway, great video. I found it very fascinating.
Very interesting and I'll have to think on this more. The most apt situation I see something like this in tournament poker is the classic situation where the shortest stack is all in, and several other players have called. Traditionally, players will often "check it down" to maximize the chance that the short stack gets knocked out, and everyone else gets a pay jump. 10 years ago this was much more common in poker, but now a lot of professionals have realized there are often edges to be had by betting, especially if you're the big stack and having a short stack in actually favors you in an ICM situation. I'm curious if what you point out here could have some effect. Anyway, most ICM situations - even BvB - tend to be a lot more complicated than what you've shown here, mostly based on stack sizes. If the BB has a much larger stack, and other stacks are at risk, then they can put a ton of pressure on SB. Even more relevant here is it's often challenging to know if your opponent is actually following ICM. Even today, there are a lot of top tier professionals who don't really understand it. This has a dramatic effect on how you should play this situations, because if BB is calling too much, raising SB is typically really bad here. So yeah, I think this makes it even tougher to assume that BB is going to understand even higher order repercussions of their actions. Amusingly, bad players often act as "enforcers" in this situation without even realizing it.
The idea that the players at the table could benefit by working as a community is both interesting and hilarious to me. I'm imagining eight guys sitting down at a poker table, shaking hands, and being like, "Okay fellas, it's us against the world. All for one and one for all!" After some thought, I think a part of the problem for poker specifically is that the type of cooperation you've described doesn't "feel" like playing poker. Like you said in the video, it seems like collusion, except that in a tournament setting, it's plus EV for every cooperating player. (That last bit may not be right, I'm still not totally clear on the tournament poker rules.) Another interesting consideration is that assuming the table does indeed cooperate, the chips will be randomly distributed the longer the cooperation continues. But as players' stacks grow, they are more and more incentivized to start betraying the others. Like, a player with a big enough stack could start betraying without fear of retaliation from an enforcer, which might also force the cooperation to break. Since I'm no poker pro, I'm forced to view this problem abstracted away from all the realities of what a poker game might be like, so maybe I'm misunderstanding something. Very interesting video nonetheless. I'd be interested to see more contemplations like this in the future.
the question of when you should start defecting is definitely a very interesting one with regard to iterative prisoner's dilemmas, and at a tournament poker table there would be game theoretical solutions which change depending on the exact situation of the tournament. but of course, players don't even engage with the idea that this is going on to begin with, so any examination of this stuff ends up being entirely theoretically.
I can't quite tie this into the whole, but I cant stop thinking about early 2000 live poker memories of playing with an older person who would try to "check it down" with me in heads up pots, or blind vs blind pots. I think they saw it as - hey buddy, we're in this together, do we really want the angst of bluffing and stealing over a measly 2BB starting pot? How bout we just let our cards talk and continue going after the rest ofem'. Great video thanks for doing it.
Interesting video. I think one other thing worth considering is that tournament payouts tend to be heavily skewed towards the top, which changes the EV calculation quite a bit. It could be the case that non-cooperation results in a higher sum EV for both players if it gives each player a slightly better chance for one of the top tournament spots. I think similar considerations apply in other parts of life. We often have the choice to cooperate and settle for a nice outcome for everyone, but the rewards for being at the very top are immense.
Cool video. Game theory/Prisoner's dilemma is one of my favorite topics since once you understand it you begin seeing it everywhere. If I understand what you are saying correctly it's that that at the highest level of skill you would expect that other players know and understand Game theory and would make decisions based on what Game theory says they should do (ie cooperate for long term benefits) but instead you find many people playing based on the short term (ie defecting for short term gain). As much as I would like to comment on this, since I don't play nor have any interest in playing professional poker it's very hard for me to make a statement regarding the skills an analytical level of average poker players since in my life, when similar things have come up in games I played, I know that my opponents did not have the analytical skills capable of understanding Game theory and thus were not making their choices based on it.
I don't like the assertion beginning at 9:14 that the "picking up trash in the park" thought experiment flatly _is_ a prisoner's dilemma. Weather it is or isn't depends entirely on each individual's perception of how much utility is spent by picking up trash, how much utility is gained by having a clean park, and even more complex stuff like how many people need to pick up trash before the park is so clean that additional co-operators don't add additional utility. The thought experiment can have very different equilibria depending on how you view the costs and benefits. How a population interacts with their commons can look like a prisoner's dilemma, or it can incentivize total co-operation, or it can result in any number of equilibria with a mix of co-operators and defectors. It depends on the culture of the population.
I wonder if it's because it's more trouble than it's worth. That knowing the pure game theory on a macro level, and mastering that has enough proven results that to deviate for the sake of deviation makes you seem untrustworthy on principle. Especially in a profession where the public sport of it is to be 'hard to read'.
I’m not a professional poker player, but I’ve played fake internet poker enough for this to occur to me. When the game is casual and the money is fake, people will be hanging out and having a good time, people tend to switch on the role of enforcer and avoid destroying each other while eating up newbies or people who don’t play ball. This is the biggest thing that scares me about sitting down and playing a cash game at a casino, because I know that if the other players get along, I could get eaten alive without any deliberate collusion.
Interesting discussion. However, I don't think this is a PD because neither option dominates the other in terms of individual payoff. (e.g., if SB has a weak/losing hand, 'going all in' has a higher payoff than the alternative conditional on their opponent folding, but a lower payoff if the opponent calls.)
This is a very interesting concept, and I'd be curious to see if a strategy like the one you laid out could work. My suspicion is the current state of tournament poker is at an equilibrium. Since enforcing is self-detrimental, enforcers are rare, and there is little to punish the SB pushing their advantage. In a case where almost everyone pushes their own advantage as SB, attempting a cooperative strategy as SB would likely just result in conceding the advantage you have without impacting the gameplay of the other players. I think you got really close to the core of the issue when you were complaining that nobody would consider that a prisoner's dilemma existed. Because players in general don't consider the concept of mutually beneficial cooperation in prisoner's dilemma scenarios, they'll always play the greedy line, forcing you to do the same. The only way to make a cooperative prisoner's dilemma strategy work is if there are other players attempting the same (and this gets really close to cheating). Interestingly, I believe there are 2 different stable equilibriums both with equal initial expected value for every player. 1) the current one where each player plays a greedy line and any player attempting to cooperate gets punished by all the greedy players. 2) a cooperative strategy where a greedy player gets "blacklisted" from cooperation if they attempt the greedy line too frequently. The first case is the "losing" case of prisoner's dilemma, and the second is the "winning" case, but because all you care about in tournament poker is your value *relative* to the other players, there is no difference to an individual player's expected value between the two cases. Although I said previously that both equilibriums are stable, the first (the current state of play) is "more stable" than the other. It is pretty clear that the way a player gets punished in the current state of play is that they give up some of their current advantage with the goal of gaining more global advantage through cooperation. All the other players have to do to punish that strategy is keep playing the way they normally do. In the second "cooperative" case, this line is much less clear (as you mentioned in the video). At any time, any player can gain instantaneous advantage by playing a greedy line. The way the system keeps everything in check is through either enforcers or lack of future collaboration with greedy players. However, it can be really difficult to define a best strategy for when you should blacklist another player from attempts to cooperate (due to them breaking the "rules" of collaboration). What are the rules? How frequently is it okay to take a greedy line? Is it okay to play greedy if you get dealt AA? etc. Each player may have different rules about when to cooperate and when not to which makes it an incredibly interesting scenario from a game theory perspective but also one that could come crumbling down depending on the specific interactions between players. As a bit of a side note, I think it would be really interesting to look at how different strategies performed as poker games evolved. how does cooperation perform over the course of a full tournament when you don't know ahead of time which players will cooperate? Is there a minimum density of cooperative players that makes a cooperative strategy viable? When is the right time to stop cooperating with a player because they're doing too well or too poorly?
Jorbs spent ~20 minutes convincing me not to litter and to pick up trash. On a more serious note i live in a student complex and we have a common room. It was filled to the brim with old, broken and dirty furniture. I got irritated enough with it to start cleaning up and with the help of some of my neighbours i managed to get it cleaned and put some nice desks and couches down. I am dreading the maintenance though because i can't for hte life of me imagine any one of these people picking up a broom or mop once in a while to keep it clean... And i don't think i am diplomatic enough or have enough of a stick to get people to do it.
I've typed up and deleted like 8 comments of novella length trying to sum up my feelings about this video, but suffice to say it stresses me out, too. Biggest reason is it goes beyond games. Trying to get people to cooperate even when it's in their own best interest is extremely difficult. Don't really need to do more than vaguely gesture to the struggles of the United States on a cultural level. I think the default state of people thinking and behaving is Zero Sum, and I worry that's hard encoded in human behavior from the millenia we just pillaged other people to get their stuff.
I'd never heard additive tragedy of the commons before, it's always been reductive (original reason is the Commons was used by cattle farmer for extra grazing land under british law, which gave an advantage as they would still have their own land as well. But this led to the commons being depleted completely and no-one could enjoy its benefit) This does turn some issues in its head making it more interesting and explains why when I've played the role of what you have aptly described the enforcer why I was in a weak but long lasting position when playing with friends.
How much would the prize structure of poker favoring high performance mean that you want to sacrifice average hand value for a higher likelihood of placing high even with a increased likelihood of placing low
It's been a while since I took a game theory class, but one concept that stuck with me was the idea of resolving game theoretic dilemmas by introducing a correlated equilibrium (en.wikipedia.org/wiki/Correlated_equilibrium ) which is essentially just a public signal that can impact the player's best interests. The classic example is a traffic light. Without it, the nash equilibrium is for the current traffic flow to maintain (never have to stop so commute is faster) and for the perpendicular flow to remain stagnant (driving into oncoming traffic is generally not in your best interest). Traffic lights are a public signal that guides all drivers to achieve a higher EV (never have to wait indefinitely for traffic to clear up). I don't have experience with poker, so I don't know how this could be applied, but it may be an interesting avenue of thought at least.
Maybe all the poker players who were interested in cooperative strategies just gravitated to Bridge :P Or these days, Hanabi (which feels uncannily like cooperative poker at high levels)
For a very simple approximation of a single bet in the 2-player situation, let b be the bet, p be the probability of winning the hand given no further information, and ICM(b) and ICM(-b) be the expected net value of winning and losing the hand, respectively. In general, if ICM(b)*p > |ICM(-b)*(1-p)|, then you'd have positive expected value for betting b. For this simple model of a single bet, for every b > 0 there is some p >.5 s.t. for every p'
These are the kind of rants I subscribe for! Is it possible that backwards induction explains why there can be no "enforcement" style equilibrium?" Because unlike an iterative prisoner's dilemma that continues, potentially indefinitely, tournament poker must end with one player winning all the chips. So even if a tit-for-tat strategy could maintain some sort of equilibrium in the short run, a real tournament will eventually be reduced to a 2 player table. And once the game is heads up, any previous enforcement of the common good will have to break down. Because there is no common good between two players when it's just them and nobody else is left to profit off them "going to war with each". That's how I understand it at least. So working backwards, each player knows the common good must break down at some point and they have to decide the point at which they individually abandon their own enforcement of the common good. Add to this the possibility that enforcement is only beneficial so long as other players choose to enforcement strategy and you probably end up with the case that the optimal strategy is to never enforce at all by backwards induction. That's about as far as my understanding goes though. Curious what anyone else would add to this?
poker is often analyzed as a continuous game, even past the end of the tournament. i.e. the players can be reasonably expected to sit down and play against each other again sometime in the near future. the practical applications of this do get messy though - how much can you really rely on history you have with someone from six months ago, when they're constantly learning more about the game, for example.
@@Jorbs Interesting I would never have thought to view poker, especially tournament poker, as a continuous game. But I can see how top level players will repeatedly find themselves seated at the final table across from the same pro players throughout the year. I am definitely not smart enough, nor familiar enough with high level poker to contribute much else to the conversation but interesting video, thanks for posting it and the reply!
This kind of dynamic actually showed up in a real tournament when Gus Hansen went all in on 15 of 19 hands without even looking at his cards, eventually netting him the win. He stole many pots and twice other players tried to stop him and he got lucky on the river.
I like you're interpretation here, I think the truly mathematical approach to tournament settings is to model the event itself, which in the setting of poker transforms the game from a being about chips to a game about abstract resource management and attrition, much more similar to diplomacy than to a gambling game tied to real cash value which is how a casual game might go. In the tournament setting the 'value' of a chip is just flavorful called a 'dollar' or krown, etc, the real value is the chance to win. So maybe tournament poker would be a better game if open cooperation would be allowed. Maybe at that point though it's just actual diplo though, and the players would prefer another game. Is it possible that poker players are just too tied to the idea of cash in hand and the flavour of the game clouds their assumptions about strategic position within the tournament setting?
Back when I played online poker in the before times (before black Friday) when playing 1-table SNG this style of blind-vs-blind play was not uncommon. This was when ICM was well understood and frequently talked about but solvers and GTO wasn't a thing (at least not a thing publicly discussed on 2+2 or other similar sites).
It's good that you mention tit-for tat because it's a baseline for good prisoner's Dilemma strategy. Based on what I read a while ago, this has been tested in tournament play, and a slightly more tolerant version of tit-for-tat. But either makes sense. But your translation of what tit-for-tat would mean into actual MTT strategy is wanting a little. So when you describe cooperative play here as, SB calls and both check down till river, that is cooperative, and definitely illegal collusion, but what I want to point out it is that it isn't what tit-for-tat would look like. If I want to play a tit-for-tat style (which isn't too far from my idea) what I mean by that at a final table, is that I want to retain as much equity in that situation as I would in a cash-game, and I won't try and steal your equity by bullying you and going all-in, but nor does it mean I am going to limp AA with the idea of passively checking down to the river. What cooperative play here would look like is both playing this relatively similar to how they might do for chips, with some slight adjustments. The cooperation occurs in a strategy vs strategy sense, rather than by checking down every individual hand. Tit for tat would be something like this. I have 13BB in the BB. SB has 17BB in the small blind. 4 other players have 5 BB each. SB shoves. I have 99, I call. A traditional understanding of ICM would be that BB calls absurdly infrequently here because of the pressure, and SB gets to go wild and shove all hands because BB never calls. But if I start doing this, the outcome is I suffer a little (I lose EV compared to traditional ICM calculations, which are too tight) but when I do win I gain a lot. By comparison, if SB shoves 100% of hands and I call about 15% of the time, SB is absolutely burning money doing this. The unique thing about ICM in poker is that your opponent's errors can hurt you, which makes the EV of any action inherently extraordinarily hard to calculate, because the strategic dynamics are here truly complex and not just a matter of mathematics, and any Game Theory equilibrium is unstable in the sense that following it can be very costly if others are willing to hurt themselves, even a little.
your assertion that checking down to the river would be illegal collusion is very weird to me, especially in the context of the rest of your comment. i don't know of any definition of collusion which suggests that trying to maximize your winnings by avoiding detrimental ICM situations with the other players at the table is collusion. i think maybe you misinterpreted and thought i was saying players should always act in a way which attempts to cooperate in the prisoner's dilemma, which wasn't my point. my point was more like if i was trying to demonstrate what fold equity is and said "so you can go all-in every hand, look at how much money you'll win from fold equity". i'm not suggesting that is a good way to play, or that that is the only way fold equity ever exists, i'm just demonstrating the concept in its absolute simplest form so that someone can recognize that it exists. attempting to cooperate in a prisoner's dilemma is just one of the many reasons for players to make (self-serving!) choices in many different situations in tournament poker, and attempting to do so isn't collusion. your analysis in your third paragraph is basically what i am taking as given in this video, followed by me trying to explain why SB and BB actually both benefit from working out how to not destroy each others stacks more often. a reading of the situation without recognizing that it's a prisoner's dilemma struggles to arrive at sensible conclusions, but it's actually not that much more complex than a cash game if you recognize that it IS a prisoner's dilemma and appropriately work cooperative lines into each players' strategies, and it's fairly obvious imo that they can both be self-incentivized to attempt to do this without needing to be colluding.
Interesting idea, but I don’t find it distressing. Like you said this is a pretty complex idea and while your logic makes some sense that doesn’t mean it’s actually correct, even when played out over a long scale. If sb only ever goes to 10 when they have aces then bb isn’t an enforcer, bb is just a sucker. Even though the chips aren’t worth as much I would wager it still statistically improves your tournament winnings to go in on those aces, you don’t care if the value of the sb and bb decrease if your value increases on average. As you said poker is a meritocracy where only the best rise to the top, so if the people at the top aren’t using a strategy you’ve thought sounded good in practice it just might be that it’s a lot messier in theory. It doesn’t necessarily mean that the idea is wrong. The best players might be smart enough to realize that most people aren’t smart enough to play along, but idk. Pretty much I don’t think the best poker players are any worse because they don’t heed the theory of another player that makes sense, but quickly muddies when taking in the complexity of poker. Also poker isn’t really a great outlook on society because it is a game. people who normally would spend hours trying to track down the owner of a lost Wallet would happily take their money in poker, because by playing poker people consent to losing their money. You are not being rude to the other players because you all agreed to these rules when you entered, there was never any expectation of a greater good.
I think you're a little too zoomed in on the example. There's 2 players who would benefit from cooperating, but also 4 other players that benefit if they don't. And while those 4 can't do any game actions during a specific dilemma, there're ways they can discourage cooperation in general. One really simple thing they could do is *not* talk about situations like this where cooperation is a good strategy...
I don't disagree with your main thesis - I have never seen anyone talk about this and that is strange. But collusion through trading of defect / cooperate splits can get good results if the numbers aren't set up right. Given that small and big blinds rotate through everyone they define a really easy way to 'collude' by trading. This is a UA-cam comment so I didn't actually work out any maths but this might explain why there is so little effort to exploit this in tournaments. Either way super interesting video.
Hmm, so there's an alternate universe where jorbs becomes a world renowned poker player, and revolutionizes the game by recognizing and applying the prisoner's dilemma/TotC.
UA-cam listed the title as “The Tragedy of the Commons and Why I Hate…” I clicked expecting the rest of the title to be “Ceramic Fish and Bottled Flame.” I was not disappointed.
I think that the problem is that you need norms to have an enforcer. Suppose we have two types of players: E - "the ego player". E is perfectly logical, assumes everyone plays like E, and always makes the play that nets the best expected result in a given round. V - "the vindictive player". V sees that the ideal strategy is complete peace with his neighbors. V enforces this peace by outlawing aggression. If any aggression is made towards V, the game plan shifts to hurting the offending player. (AA hand is allowed if revealed afterwards, so games end eventually) When these players meet, it is mutually assured destruction. V enforces peace, and E enforces egotistical play, both punishing the other play style. The only winning strategy is to copy the strategy of the players around you. This results in a meta, where you lose if you don't follow the social norms because it is costly to enforce something that the other players disagree with. In the case of poker where collusion is disallowed, i feel that E is the more justified player, as vindictiveness is purely destructive for both players unless cooperation is obtained. (And i prefer the strategy game to the social game) I feel that the same conclusion applies if you replace V with anything but E, even T - the "tit for tat player"
A sense of community is really important for avoiding tragedy of the commons type scenarios. Getting people out and interacting with a large number of strangers in their community, however, is extremely difficult. If you look at the levers american politicians are trying to pull to solve the issue (when they're even looking for the levers), you basically find "increase the number of voluntary public service organizations and pray people join them" in some form or another. It would be easy just to have mandatory public service participation programs, but that's also like, pretty against the entire ethos of modern mixed-market liberal democratic society. Which is a problem, because the entire developed world uses that model, and only China, who is now facing demographic collapse and in the last few months has been shattering its highly lucrative tech sector into tiny pieces to reallocate their efforts elsewhere, has even come close. So mixed-market liberal democracy has no serious competitors and so there really aren't examples to learn from.
Would the ability for players who are still "in" to pay folded players a "bid" (set by the folded player), to know what cards they have been dealt, help fix tournament play? To me there should be some sort of minor rules change be allowed to benefit tournament play and avoid typical pratfalls and accusations of collusion. What that -needs- to be I'm not sure.
Outside of poker tournaments people aren't rational self-interest driven agents, people can be swayed by morals and ideas. Now this can lead to bad outcomes but it can also prevent us from getting completely stuck in dilemmas.
Vampire bats feed their children in a group. If one of the mothers doesnt bring food but act like she did, they dont feed her children next day. Tit for tat is simply "repeat the enemy's move next turn". If enemy ever cheats, cooperation is still possible. All of the animals use a slightly different strategy, which includes "if you've been cooperating for a long time and your opponent is cheating, it must be signal error so continue cooperating" which is game theory explanation of gaining someone's trust
There are two types of people in this world: those that return their shopping cart after they're done with it, and those directly responsible for holding humankind back from its true potential.
Also single parents who can’t leave their baby in the car :)
Also those who live in areas where they've create make work programs to have otherwise unemployable people spend their time collecting them.
Or people who live in areas where you rent the carts for a dollar, and get back the dollar when you return the cart. In which case the free market has solved your moral quandary for you. You're welcome.
you put my cart back then if it bothers you so much
Tragedy of commune is not applicable if the job is staffed, obviously. Also, I'm pretty sure the original comment was not that serious in its core that it makes sense to be heavily countered with real social programs... And those are not exactly terribly common anyway, being more of a welcome edge case rather than a counterpoint.
We seriously need more "Jorbs explains concepts" content.
Yes
Yea, i bet he'd pop off if he did it more
Interesting thing about bats: They have names for each other, gossip, and the things they talk about are food and personal space. Bat children tattle-tale to enforcers about who isn't feeding them!
I remember discussing the tragedy of the commons in philosophy class, and it’s interesting to see it pop up in situations like these. It’s hard to express how captivating I found this rant about a game I’ve never played, and how far the implications of this very concept of cooperation in search of a better world stretch.
Couldn't have said it better myself. Hard agree, this is fascinating.
It's everywhere. One of the most defining issues in society. It's the reason we've been unable to save the planet.
surely this can't be a benchmark for the failure of cooperation when it's a tournament game where your goal as an individual is to win over others. think tragedy of commons suffers from explaining the contradictions of its own internal game, not cooperation everywhere.
that is to say the fear of free riders presumes individual aggrandizement is the base unit of society, convenient to a society who's organizing principle is capital accumulation for a wealthy few.
Loved your rant. Main way to convince myself to do things like picking up the thrash or voting is to find mental reasons to be happy about it. Like being proud, being able to talk to people about it, etc. So I don’t try to focus on if it gives me more value purely from the situation, but I try to create a bonus reason for my mind.
As a FilthyRobot viewer. this almost perfect describes civ5 lobbies. When a player starts an "irrelevant" war, both players suffer and neither benefit enough to go on to win the rest of the game.
4 other players continue to produce buildings instead of units and not punish the player for their choice. If you remember those civ5 lobbies you'll know exactly what i'm talking about!
yes, irr war is another excellent example!
It's like 4 player smash bros free for all. Camper always wins because trading kills means you lose faster vs the lame crew
My tragedy of commons is when I can't get any copies of shrug it off
I've never thought about the Prisoner's Dilemma as a scaled down Tragedy of the Commons. It makes total sense!
Small correction but tit for tat doesn’t permanently defect against someone who defects once, which you sort of implied. It’s more accurate to say tit for tat starts cooperating, then always copy your partners last move. The reason is even if you get into a defection cycle, if you allow for some “forgiveness” the tit for tat algorithm can go back to mutual cooperation.
Great video though!
I agree with your assessment from this video from a theoretical point of view, however, I think the implementation of your ideas are where this strategy doesn't work and why most of the time people try and blanketly use the prisoner's dilemma and the tragedy of the commons it ends up not working.
The best showing of the tragedy of the commons is this picking up trash in the park, however, each individual choice is not based on the tragedy of the commons, rather, the tragedy of the commons is a weighing factor in the choice. Someone who is walking through gives it higher weight to the commons than a mother trying to guide kids through a stroll but maybe not as high as someone trying to impress a date. Theoretically, we want to be able to condense each player's situation into one that is negligible, but it is not necessarily negligible and the weighing of all other games that the player is playing may have a larger effect on the overall decisions.
In my limited interactions with tournament poker this is what I understand to be the case. Having enforces would be theoretically be beneficial to the tournament as a whole, but when many tournaments, especially online, either have high stakes or a fast seat rotation such that playing as an enforcer is not worth the personal EV in the case of high stakes or the reputation of being that enforcer is hard to develop for all but the larger names in the game in the case of the rotating seats.
This discouraging of enforcer behavior is exasperated by the fact that people that are trying to enforce are given very little opportunities to do so. This is because people who enforce a high percentage of the time are more likely to leave the tournament faster and thus out of the money and looking for another job, or if you enforce a lower percentage of the time, i.e. widening your calling range on the BB, you may not even seem like you are enforcing, just playing a little loose.
This is a good conversation to be had among poker groups, and depending on the volume of people playing like this (the more enforcers there are the higher EV there is for being an enforcer) the decision tree may change, and maybe even eliminate the need for one eventually, but the way tournaments are structured currently there is not much reason to be a lone enforcer.
TLDR Theoretically there should be enforcers if tournament poker was optimal, but the way tournaments are structured it gives harsh penalties to whoever tries to enforce so it is not a smart long term strategy.
Great video and I'm not the best writer so I can clarify if need be
i agree mostly, but think that what you have typed is very sad and depressing xD. it is concerning that the structure of the tournaments is enough to dissuade people from pursuing an optimal collaboration, especially when we consider that a lot of our other interactions with humans are in artificially-structured environments which reflect some of the sorts of things going on in tournament poker.
@@Jorbs prisioners dilema is depressing, if its any consolation I will try to aply it if I ever find a comunity in poker, hello from Argentina
ok ok but if we all agreed to punish people for NOT enforcing...
@@scottwhitty72 That would generally fall under collusion if people went into a tournament having prior discussion of specific tournament play. The way this problem is solved seems to be change the game in some way, not the strategy
@@corygodin4129 It doesn't have to be collusion, if everyone is of the opinion that enforcing is the best course of action for everybody (and thus themselves). It's just enforcing the enforcing. By your logic enforcing itself would already be collusion.
The "Tragedy" described in the Tragedy of the Commons is that there is less total time or energy resources for all parties involved. I believe that your analysis only holds for a closed system (the two players described) but is irrelevant when considering a broader system like a poker tournament pool or the entire poker industry. As you described, the prizes are set and there is no actual table-wide loss of money due to demolishing the other guy. If one person is guaranteed to lose, the expected value of the remaining players goes up, but the sum remains the same.
In essence, poker players cooperating optimally will not increase the average expected prize value, so there is nothing to cooperate towards.
That said I had never seen the Tragedy of the Commons described so accurately as multiplayer prisoners dilemma, and I loved this video.
I always love the analytical content on your channel. I've never played poker and rarely play slay the spire these days, but the concepts you bring up in your videos are almost always things that are relevant across a wide range of scenarios. There's a strong correlation between spending more time watching your videos and less time coasting through life with my brain on auto-pilot. It's an unprecedented tragedy that videos like this one aren't more popular.
When you use a strategy of reciprocation in a prisoner's dilemma, what you're essentially doing is using your future actions as a bargaining tool to make it rational for even a selfish rival to also cooperate. I could see it being frustrating to try to accomplish that with a rival who refuses to consider they're in a prisoner's dilemma at all.
Apropos of nothing, there are are some really interesting variants of prisoner's dilemma, like one where there's a small random chance to "accidentally" betray the other party without being aware of it yourself. In those situations, the optimal strategy involves forgiving the occasional betrayal so you don't betray back and forth forever because of a misunderstanding.
i think that extra comment is very apropos! this sort of scenario would have that exact sort of thing going on, because when SB gets dealt a strong hand like AA it's actually reasonable for them to play aggressively - the punishment for getting involved in a big pot is far outweighed by the rewards for shoveling money into the middle with the best starting hand in no limit hold 'em.
this is a like... semi-iterative prisoner's dilemma which you can only tell that you're in probabilistically, since you don't know what cards the other player has. it's very weird. i think it's complex enough to be subject matter for a phd or something.
To note, you do see a lot of cooperation in situations where ICM is really a huge factor. ICM is at it's biggest when the poker tournament is going from prize of $0 to whatever the minimum prize is, but usually a big jump and always more than players put in to join the tourney. Everyone already has big incentive to play very cautiously and not be the last person to get $0. But also, each table as a whole has incentive to be the more cautious table compared to other poker tables in the same situation.
Interestingly, and in some ways hilariously and in some ways sad, the people who have the biggest incentive to betray their table, are people who have the most money. Be it the person who has the most chips on the table and so cannot be eliminated in any all in, or the person whose bank account is biggest and has least to lose in a relative sense. Often know as bullies, these few abuse the players trying to play by the societal and generally sensible rules of ICM. The rich (generally, poor players also get good hands and lucky) get richer. If talking about the tragedy of the commons, these are the people dumping into the river.
this rant brought me great joy! i rarely get to listen to someone explain Tragedy of the Commons or provide insight regarding the facets of game theory or other similar theorems and your means of talking about it makes it all a lot more digestible, cheers!
I've played a game of Settlers of Catan with family where I realized I couldn't win, but I could choose whether to devote what remained of my resources to mess with one family member who had annoyed me early in the game. Playing Kingmaker. And of course I tried this because as a kid why wouldn't it be in my perceived self interest to get back at someone who annoyed me instead of politely folding? And it became this whole thing because everybody else did innately realize that this was, in a way, collusion with the player benefiting from my decision. So it changed the way I thought about games forever.
So when you explained that both poker players stood to lose expected winnings by going toe to toe with each other, I immediately understood where you were going with it. Though of course you did give it away in the title. Taking advantage of two other people devoting resources to fighting each other is a time honored strategy, so anyone who knows this wouldn't want to get caught with their pants down in, say, a strategy or grand strategy game.
Playing to win the tournament, not just the hand, means thinking strategy instead of tactics, the same way that ICM does when weighing chip values for tournament play. So it's odd that people who did actually understand ICM, didn't rise to your positing of the prisoner's dilemma... You'd think they'd be the first ones to see it.
7:40 This line of thinking reminds me of macro strategy in a battle royale. An early game fight is likely to reduce both players outcome on average. Big potential loss (losing) for a small potential gain (more aggressive looting capability). Generally stealth is preferred up to a certain point.
This is why top pro players will always fold their first few hands, even if they get pocket aces. It's not worth risking a dime that early in the tournament even with a 90% chance of winning the hand.
One thing worth noting that I think a lot of people miss - tit-for-tat strategy in prisoner's dilemmas tends to perform pretty well even against predetermined algorithms. Its success is not strictly behavioral or psychological. Love the video!
I would love more rants like this. Hearing you explain your ideas is always very enthralling for me
Good rant. One thing about tournament poker (or anything else structured similarly) is that it has very spiky results so it's easy to cash out on trust for a huge sum. It's difficult to establish a trust in, appreciation for, and loyalty to the commons when you're only going to be playing a few games with these people who are about to be taking your money.
It's still wild that they have no interest in even talking about it, and call people "emotional" when they act as an enforcer when someone starts routinely bullying the table into folding etc. But I can see how in a game where people tend to be "strategy gamers" they'd have a visceral distaste for anything that required long-term thinking and building trust for anything besides a backstab.
I thought this was going to be a StS rant video, where you explain the tragedy of bad common cards via poker...
Was interesting anyway :D
That was an incredible explanation of fantastically cool topic thats horrendously relevant to how I live my life, and all this from poker! All the while, I never really thought of myself as a freeloader, but this rant gave me a bit more pause now before I just choose inaction now. Thanks!
I expected the tragedy of the commons to be about card rarity. The video ended up being far more interesting than I expected, great rant and explanation.
I don't see how the tragedy of the commons applies here.
The price pool always remains the same, so there is no difference in outcome between a scenario wherein everyone collaborates compared to one where no one does.
Starting a collaboration comes with a risk, and it stops being beneficial as soon as all players work together.
The bad outcome of investing some recourses into starting a collaboration is not just that you might get betrayed, it's also that everyone might join and you gain no advantage for your investment.
This means that collaboration only works without forgiveness: the advantage of collaboration is only an advantage over players who don't join, so you have to ensure not everyone can join.
If you start offering collaboration and maybe even act as an enforcer you have an initial investment to demonstrate your trustworthiness, a period when you have an advantage because some people collaborate with you, and finally the end state when everyone knows your playstyle and you stop benefitting because everyone collaborates with everyone else.
I feel like with a group of professional players the middle period, the one when you actually benefit, might be really short, to short to make back the initial appeasements.
Playing the enforcer is a very hard life and anyone who does understands this distress all too well. Very obviously this has nothing to do with poker and is simply a medium to convey the nature of humans and modern society. One who has experienced life as such can confirm that it is all cost, no benefit and the only up sides are the retention of ones pride, dignity and virtue. Though these things are invaluable and the feeling of changing some ones life exquisite, it is very hard to afford a home on good deeds alone
Jorbs is a really good teacher
I have a lot of thoughts on things similar to this, most especially your thoughts applying this very specific poker situation to the wider world. Sadly, I'm not particularly excellent at putting them into good words and I've been awake for almost 24 hours and am slightly delirious at the moment. The broad sense of it, though, is that I worry that evolution primes people to think in self-interested fashions; in the short and middle terms; and to emphasize certainty over ephemera. And I strongly suspect that the vast majority of people are largely acting on evolution (selfishly) and really just doll it up to seem "rational" or whatever their preferred explanation for their behavior is. In your poker example, the enforcer has to take *serious* short term hits (like being knocked out of the tournament, reducing their expected return from 3.50 to like 2.25ish) and their return is dependent not just on their behavior but also their opponents perceiving their behavior, identifying their strategy, and modifying their own strategies to suit. It's a very ephemeral return. The tit for tat strategy relies not on predicting your opponent but on reacting to them; the past is a sure thing and the future is not.
A more personal note on the weight that ephemeral returns has on behavior: I've been heavily self-monitoring how much effort and thought I put into writing this comment on UA-cam because "what are the odds anyone reads it and cares?" "Will someone say nice things about it and gimme that sweet dopamine hit?" "Will I get the coveted heart-from-the-video-poster-who-is-Jorbs-a-person-that-I-think-is-very-smart?" These are all weights against me writing semi-publicly about something I think about, like, all the time. One would imagine the barrier to doing that would be pretty low but even now my fingers wanna press ctrl-A delete. Too much risk, says my brain.
I like the way you analyze information. I believe you pick these things up faster than most. You are also able to explain very well what you have found and why. I came here to learn some slay the spire but this is my favorite video you have so far. Also kinda mad at how easy you made Loop Hero look.
Here's an idea: If you think being an "enforcer" can be profitable, you should explore that situation. Use a tool like ICMizer find a single scenario following these steps:
1. SB can push really wide at equilibrium because BB has to fold tons.
2. Adjust BB's strategy to call with losing hands in order to prevent SB from jamming so much.
3. Resolve for the SB strategy given the locked BB's strategy.
4. Check that BB's overall EV has improved.
This isn't perfect since, but I think if you find a situation in which the above is true, then there might be merit to what you are saying. Intuitively it almost seems plausible, where in GTO you might have a hand that is slightly losing but including it in your range greatly improves the EV of your range. It seems strange for calling an all-in because you don't care about obscuring your hand anymore, but maybe.
To me it sounds like there are misplaced incentives in tournament poker that make it deviate from cash poker in unexpected ways.
I think one of the issues with tragedy of the commons problems is they are somewhat created by a blindness to that type of problem.
Very insightful video.
Any tournament system adds an additional layer to the game that shifts incentives. Even outside tournaments, the fact that players play more than a single game in their life adds a layer to the game and slightly changes ideal strategy (like the iterative prisoners dilemma jorbs proposed)
Great talk. I've always had this feeling that these things were things, but I've never been able to get it to more than a feeling. You're completely correct and it's one of the great themes when people start talking about the question of "are humans fundamentally decent" or not.
Luckily I've found that being a decent human being doesn't necesarily have anything to do with being a good strategic thinker, so we're probably fine if these great minds don't understand this. Most people seem to understand it on some level, even if they could not put it to words.
Loved this Jorbs!! Thank you for relaying this unfiltered.
Interesting video, I spent many years as a (primarily tournament) poker pro and think I can share a little insight into this situation. There are two big reasons why these types of quasi-cooperative metagames aren't really thought about much. First is, as you mention, people really try to avoid even giving the appearance of colluding. The second is simply that tournaments have a large player pool, and unless you play the highest stakes, you'll rarely play against the same players. Given that showdowns are relatively rare, it is very difficult to ascertain whether or not a player is 'cooperating' in a short amount of time. So any implicit agreement to play a small pot is going to open you up to exploitation by an aggressive opponent, but you won't know for sure whether you are being exploited since, as you mention, they might just be getting dealt strong hands. And given that you may never see that same player again deep in a tournament (ICM isn't really a factor until the late game), it's very difficult for cooperative strategies to gain traction.
When I played high volume, high stakes sit n go's for a while, I did actually see some of these types of meta-games becoming a factor among the regulars (which shouldn't be surprising, because in that environment it is both a much smaller player pool and you are constantly seeing the same people, and also, ICM is much more consistently an important factor).
I love this! I imagine this might take awhile to make but I would love to see more videos like this.
As someone who presently follows and plays tournament poker I think I can say that the community as a whole has recognized and applied some of these ideas that you mention, and that poker theory has progressed in ways that would incorporate a lot of the suggestions you mention.
The first example is related to what you mention at around 13:00 where you mention small blind limping and playing a hand very passively. My understanding of current theory is that there actually are quite a few spots depending on stack size (15-25 BB range) where mixing in limps is important because of how brutal it is to raise and then fold to an all in. This especially applies to the situation you mentioned where neither player has a strong hand. Of course this can be exploited if the SB only does this with weak hand, but the way to counterbalance is to choose some number of relatively strong hands where you limp and then shove or raise over the BB if they raise you, discouraging them from "cheating," or to use a phrase that many poker players seem to be fond of "catch them speeding."
Players also tend to size down their bets and play fewer raises in general in situations with high ICM implications. They'll check more in spots where they might've tried to value bet. They might play a hand that they would've 3 bet preflop without ICM as a flat. 3bets will be sized down from 3x to 3.5x the initial raise down to 2.5x to 2.75x the initial raise Going back to the original example of limping and then checking down this can and does happen if no one hits anything postflop. Granted if they do hit a strong pair or other hand then they'll still bet and play poker, but often with a lot of the other gameplay modifications I already mentioned. All of this tying back to the idea that you want to lose the minimum and reduce some variance in these situations. Some of these checkdowns may come across as softplaying or whatnot but it's generally understood that the best option for both players is to avoid putting in more unnecessary chips and they are still acting in their own interest, just in a way that aligns with each other.
The second example is related to what you mention about enforcers around 16:20. I've seen this role played at high level tables by the player with the shortest shack or close to the shortest stack depending on how flat the payouts are. This player will of course still be affected by ICM but much, much less than a lot of the other players, especially those in the middle and can play much closer to just pure chip EV in terms of their gameplay. They might even be able to make the adjustment of calling much further down than they normally would at any other point in their tournament, especially against the biggest stack because of how wide they can expect the biggest stack to be opening/shoving.
Further I've seen even some medium stacks play enforcer if they think a larger stack is sufficiently getting out of line. The larger stacks, while being able to pressure the smaller stacks because of ICM are not immune from ICM effects themselves. A lot of times it can be difficult to shove into a stack that will cripple you if you don't have a sufficiently high equity hand because although you will gain a small amount of chips if you choose too many hands to shove then even some of the middling stacks will be able to play enforcer and make lighter calls and you will commit ICM suicide by losing an unnecessarily large pot. To go into second level big stacks will often simply raise their opening range as a result and the antidote to this is often to try and punish them by making light 3 bets either all in or to a normal-ish size to provide counterpressure that they could be risking a significant portion of their stack.
Overall I do recognize the idea that the situations you've described are in fact dilemmas and the larger implications it has on society in terms of wider recognition, but I don't think it's maybe quite as doom and gloom as you mention. I think there is some recognition of the problems you mentioned even if I haven't quite seen it laid out so plainly before as you have here. Either way I did still enjoy listening to your thought processes here and I hope this provides some insight into how people may have recognized some of the problems you've presented.
Something I see in a lot of games is that when players encounter a strategy they don't like because they see it as "impure" or "not what the game is about", they will sometimes try to tell themselves that that strategy doesn't actually work or isn't actually effective. It seems they want to see their game's rules as more robust and in line with their preferences than they actually are.
Has anybody here read Reasons and Persons by Derek Parfit? I think y'all would really like it if you dig the moral and logical complexities in these types of situations.
The first half, which is the "reasons" in the title, tries to lay out a theory for why we act the way we do. Parfit pulls on a bunch of different flavors of prisoner's dilemma to try to suggest that it is often rational to act neither selfishly (in your own best interest) or morally (cleaning the Commons).
It's a philosophy book but it leans on a lot of game theory and maths. Really interesting read!
I don't play poker but now I want to read and learn more about poker theory. Great video, hope you do more like it
The initial scenario about ICM reminds me of Nim, where there are some positions which are unwinnable not on their own, because you've just lost, but because you are guaranteed to be moved into a state which is the one in which you've just lost. So its more important as your options narrow to hold onto those few options, than it is when you have a variety of options
Love this rant.
I was there when we talked about this live on stream, left with more questions than answers. Thank god
Thats funny, this different angle of playing is precisely the reason I only play tournament and sit n go.
Cash games feel dry and procedurally based in comparison, whereas tournament you get to find an edge by figuring out how much your opponents are willing to risk in any given hand.
I’ve never played poker outside of like an old ps2 game, but I just wanted to say I still found this video incredibly interesting. Thanks jorbs!
This is a very good video. What you came to realize (correctly) is that there is a fundamental asymmetry between two winning strategies, namely Position (acting LAST) and Initiative (betting FIRST). (Acting last is useful because you get extra information and betting first is useful because you allow yourself an extra way to win the hand, ie by making everyone else fold). Ideally - and this is true when stacks are very deep - you want to have both.
This is why for example the BB would be at a huge advantage if the SB were to simply call preflop with deep stacks. The BB can simply raise (with a lot of holdings) and put maximum pressure on the BB on all streets. On the other hand, if stacks are short (say in the 10-20 bb range) the fact that the SB acts first is now actually an advantage because they have an option to go all-in effectively nullifying the power of position (This is what is known as a push-or-fold situation).
The point is that depending on the stack sizes, one of the two players can prey on the other player's Expected Value without worrying about the prisoner's dilemma. In the first case, BB can punish the SB with impunity for trying to stay in the pot. In the second case, the SB should shove (again, with impunity) knowing very well that it is unprofitable for the BB to call even if they have a very good hand. Notice that there is a stack threshold at which the push-or-fold strategy becomes unprofitable for the SB as their risk/reward ratio become exceedingly unfavorable. (There are several factors to calculate this threshold but a good heuristic is that it is typically around the 15-20 bb's mark).
As a long time fan of your slay the spire content + as a poker player, I found this video incredibly interesting. I would really love to hear your take on cash games as well in a similar video format.
Wow, this is great. I have been thinking a lot about "spite plays" in cEDH in magic the gathering lately, and this shares a lot of the same problems.
Basically, if someone attacks you for lethal with a 4/4 with a good ability and two 2/2s with no abilities. And you are at 2 life with two 2/2 blockers. Should you block the 4/4 and a 2/2 to minimize damage, even though you are dead anyway. Or should you double block the 4/4 to punish the player killing you as much as possible.
Also, if you can disenchant their rhystic study or something before taking lethal, should you do that, even though it does nothing to save you, only makes the player killing you less likely to win.
Generally the community says "no, don't spite play" but if you do, you could get a reputation as an "enforcer" who does spite plays. That will make people think twice about attacking you for lethal, in fear of reprecussions. So long term, it is the highest EV play to always spite play as hard as you can.
Ofc in cedh it is a lot easier to solve, because you don't lose any EV by spite playing, you can only gain EV, the only reason to abstain from spite plays is a social one. So from an EV perspective it should be clear cut, but there is still so much discussion about it and people calling it "un-sportsmanslike" etc.
Also, you are allowed to talk to the other players and make threaths before they make decision to attack.
The solution to that cEDH matter is "you should always play for team chaos". It's not un-sportsmanlike if you alter everyone's game. Doing so, even if you lose, you still change the course of the match. After all, making an impactful play is what informs any decision during the game, isn't it?
"All Americans are temporarily embarrassed millionaires." - John Steinbeck
That is to say, enforcement behavior is looked down upon in highly individualistic societies. Which is sad. You can extend your argument from poker to tax policy or even for some people whether or not to get vaccinated... But to connect back to poker, the perceived and actual costs of enforcement behavior are too high, esp. in no-limit. A 40% chance of busting out (decent, but not great hole cards), just to have a chance at teaching the other player a lesson that may just be interpreted as poor play? In your picking up trash at the park example, the analogy would have to end up being like, picking up bloody needles...
But your main point here really seems to be that poker players don't talk about this at all. Which doesn't strike me as weird, and I'll say why. (That said, I never knew this was even a problem until your excellent video!),. I don't find it weird because you're thinking of the prisoner's dilemma in terms of optimal long-term outcomes, but there are just far too many players in the pool that may be savvy game theory players on a per-hand or per-tournament basis, but they don't aspire to be a grinder. They want to win big. Too many players thinking like that, and too much money getting put in to the system on the expectation of (relatively) rapid gains. It's not like financiers bankroll poker players because they think they'll get a .05% better rate than a mutual fund... Sure, there are small time pros that want to make a living, but then an effective "enforcer protocol" for poker would essentially have to be insured by a pool of cash from like-minded players to cover the losses. Which is obviously collusion / coordinated action. So to reiterate, I think it's the external pressures from the system (expectations of outsized financial gains) without any path towards effective enforcement that lead people to not even talk about this, because there is literally nothing that can be done. Like you said, it's kind of sad.
Anyway, great video. I found it very fascinating.
This makes so much sense, yet makes no sense. But I think I get it. Sweet vid jorbs
Loved this video. Something different. But something really fun to watch and think about the concepts in it
Very interesting and I'll have to think on this more. The most apt situation I see something like this in tournament poker is the classic situation where the shortest stack is all in, and several other players have called. Traditionally, players will often "check it down" to maximize the chance that the short stack gets knocked out, and everyone else gets a pay jump.
10 years ago this was much more common in poker, but now a lot of professionals have realized there are often edges to be had by betting, especially if you're the big stack and having a short stack in actually favors you in an ICM situation. I'm curious if what you point out here could have some effect.
Anyway, most ICM situations - even BvB - tend to be a lot more complicated than what you've shown here, mostly based on stack sizes. If the BB has a much larger stack, and other stacks are at risk, then they can put a ton of pressure on SB.
Even more relevant here is it's often challenging to know if your opponent is actually following ICM. Even today, there are a lot of top tier professionals who don't really understand it. This has a dramatic effect on how you should play this situations, because if BB is calling too much, raising SB is typically really bad here. So yeah, I think this makes it even tougher to assume that BB is going to understand even higher order repercussions of their actions. Amusingly, bad players often act as "enforcers" in this situation without even realizing it.
The idea that the players at the table could benefit by working as a community is both interesting and hilarious to me. I'm imagining eight guys sitting down at a poker table, shaking hands, and being like, "Okay fellas, it's us against the world. All for one and one for all!"
After some thought, I think a part of the problem for poker specifically is that the type of cooperation you've described doesn't "feel" like playing poker. Like you said in the video, it seems like collusion, except that in a tournament setting, it's plus EV for every cooperating player. (That last bit may not be right, I'm still not totally clear on the tournament poker rules.)
Another interesting consideration is that assuming the table does indeed cooperate, the chips will be randomly distributed the longer the cooperation continues. But as players' stacks grow, they are more and more incentivized to start betraying the others. Like, a player with a big enough stack could start betraying without fear of retaliation from an enforcer, which might also force the cooperation to break.
Since I'm no poker pro, I'm forced to view this problem abstracted away from all the realities of what a poker game might be like, so maybe I'm misunderstanding something. Very interesting video nonetheless. I'd be interested to see more contemplations like this in the future.
the question of when you should start defecting is definitely a very interesting one with regard to iterative prisoner's dilemmas, and at a tournament poker table there would be game theoretical solutions which change depending on the exact situation of the tournament. but of course, players don't even engage with the idea that this is going on to begin with, so any examination of this stuff ends up being entirely theoretically.
I can't quite tie this into the whole, but I cant stop thinking about early 2000 live poker memories of playing with an older person who would try to "check it down" with me in heads up pots, or blind vs blind pots. I think they saw it as - hey buddy, we're in this together, do we really want the angst of bluffing and stealing over a measly 2BB starting pot? How bout we just let our cards talk and continue going after the rest ofem'.
Great video thanks for doing it.
Never knew jorbs played poker professionaly, but it makes a whole lotta sense
Interesting video. I think one other thing worth considering is that tournament payouts tend to be heavily skewed towards the top, which changes the EV calculation quite a bit. It could be the case that non-cooperation results in a higher sum EV for both players if it gives each player a slightly better chance for one of the top tournament spots.
I think similar considerations apply in other parts of life. We often have the choice to cooperate and settle for a nice outcome for everyone, but the rewards for being at the very top are immense.
Cool video. Game theory/Prisoner's dilemma is one of my favorite topics since once you understand it you begin seeing it everywhere. If I understand what you are saying correctly it's that that at the highest level of skill you would expect that other players know and understand Game theory and would make decisions based on what Game theory says they should do (ie cooperate for long term benefits) but instead you find many people playing based on the short term (ie defecting for short term gain). As much as I would like to comment on this, since I don't play nor have any interest in playing professional poker it's very hard for me to make a statement regarding the skills an analytical level of average poker players since in my life, when similar things have come up in games I played, I know that my opponents did not have the analytical skills capable of understanding Game theory and thus were not making their choices based on it.
I don't like the assertion beginning at 9:14 that the "picking up trash in the park" thought experiment flatly _is_ a prisoner's dilemma. Weather it is or isn't depends entirely on each individual's perception of how much utility is spent by picking up trash, how much utility is gained by having a clean park, and even more complex stuff like how many people need to pick up trash before the park is so clean that additional co-operators don't add additional utility. The thought experiment can have very different equilibria depending on how you view the costs and benefits. How a population interacts with their commons can look like a prisoner's dilemma, or it can incentivize total co-operation, or it can result in any number of equilibria with a mix of co-operators and defectors. It depends on the culture of the population.
Very unusual but interesting video. I know this probably wont get as much traffic as the sts videos but I just want you to know that I appreciate it
I should be sleeping right now but listening to jorbs talking about problems i never thought about seems so much more interesting
I wonder if it's because it's more trouble than it's worth. That knowing the pure game theory on a macro level, and mastering that has enough proven results that to deviate for the sake of deviation makes you seem untrustworthy on principle. Especially in a profession where the public sport of it is to be 'hard to read'.
I’m not a professional poker player, but I’ve played fake internet poker enough for this to occur to me. When the game is casual and the money is fake, people will be hanging out and having a good time, people tend to switch on the role of enforcer and avoid destroying each other while eating up newbies or people who don’t play ball.
This is the biggest thing that scares me about sitting down and playing a cash game at a casino, because I know that if the other players get along, I could get eaten alive without any deliberate collusion.
Interesting discussion. However, I don't think this is a PD because neither option dominates the other in terms of individual payoff. (e.g., if SB has a weak/losing hand, 'going all in' has a higher payoff than the alternative conditional on their opponent folding, but a lower payoff if the opponent calls.)
I literally asked myself yesterday whether or not jorbs played poker professionally before. Suffice to say this video answered my question :D
This is a very interesting concept, and I'd be curious to see if a strategy like the one you laid out could work. My suspicion is the current state of tournament poker is at an equilibrium. Since enforcing is self-detrimental, enforcers are rare, and there is little to punish the SB pushing their advantage. In a case where almost everyone pushes their own advantage as SB, attempting a cooperative strategy as SB would likely just result in conceding the advantage you have without impacting the gameplay of the other players. I think you got really close to the core of the issue when you were complaining that nobody would consider that a prisoner's dilemma existed. Because players in general don't consider the concept of mutually beneficial cooperation in prisoner's dilemma scenarios, they'll always play the greedy line, forcing you to do the same. The only way to make a cooperative prisoner's dilemma strategy work is if there are other players attempting the same (and this gets really close to cheating).
Interestingly, I believe there are 2 different stable equilibriums both with equal initial expected value for every player. 1) the current one where each player plays a greedy line and any player attempting to cooperate gets punished by all the greedy players. 2) a cooperative strategy where a greedy player gets "blacklisted" from cooperation if they attempt the greedy line too frequently. The first case is the "losing" case of prisoner's dilemma, and the second is the "winning" case, but because all you care about in tournament poker is your value *relative* to the other players, there is no difference to an individual player's expected value between the two cases.
Although I said previously that both equilibriums are stable, the first (the current state of play) is "more stable" than the other. It is pretty clear that the way a player gets punished in the current state of play is that they give up some of their current advantage with the goal of gaining more global advantage through cooperation. All the other players have to do to punish that strategy is keep playing the way they normally do. In the second "cooperative" case, this line is much less clear (as you mentioned in the video). At any time, any player can gain instantaneous advantage by playing a greedy line. The way the system keeps everything in check is through either enforcers or lack of future collaboration with greedy players. However, it can be really difficult to define a best strategy for when you should blacklist another player from attempts to cooperate (due to them breaking the "rules" of collaboration). What are the rules? How frequently is it okay to take a greedy line? Is it okay to play greedy if you get dealt AA? etc. Each player may have different rules about when to cooperate and when not to which makes it an incredibly interesting scenario from a game theory perspective but also one that could come crumbling down depending on the specific interactions between players.
As a bit of a side note, I think it would be really interesting to look at how different strategies performed as poker games evolved. how does cooperation perform over the course of a full tournament when you don't know ahead of time which players will cooperate? Is there a minimum density of cooperative players that makes a cooperative strategy viable? When is the right time to stop cooperating with a player because they're doing too well or too poorly?
Jorbs spent ~20 minutes convincing me not to litter and to pick up trash.
On a more serious note i live in a student complex and we have a common room. It was filled to the brim with old, broken and dirty furniture. I got irritated enough with it to start cleaning up and with the help of some of my neighbours i managed to get it cleaned and put some nice desks and couches down. I am dreading the maintenance though because i can't for hte life of me imagine any one of these people picking up a broom or mop once in a while to keep it clean... And i don't think i am diplomatic enough or have enough of a stick to get people to do it.
This is such a common problem in our society, and it comes up across the board. It can be infuriating and so detrimental to progress and productivity.
I've typed up and deleted like 8 comments of novella length trying to sum up my feelings about this video, but suffice to say it stresses me out, too. Biggest reason is it goes beyond games. Trying to get people to cooperate even when it's in their own best interest is extremely difficult. Don't really need to do more than vaguely gesture to the struggles of the United States on a cultural level.
I think the default state of people thinking and behaving is Zero Sum, and I worry that's hard encoded in human behavior from the millenia we just pillaged other people to get their stuff.
Reminds me of the jon bois video he did about odd stories in the poker world
this was a lovely video and i'd enjoy seeing more like this a lot :)
I feel this will be interesting to discuss with other poker players not twicht chat bots... I wonder if somehow we can reach out
I'd never heard additive tragedy of the commons before, it's always been reductive (original reason is the Commons was used by cattle farmer for extra grazing land under british law, which gave an advantage as they would still have their own land as well. But this led to the commons being depleted completely and no-one could enjoy its benefit)
This does turn some issues in its head making it more interesting and explains why when I've played the role of what you have aptly described the enforcer why I was in a weak but long lasting position when playing with friends.
And here I was thinking that this would be a video about picking common cards in Slay the Spire.
How much would the prize structure of poker favoring high performance mean that you want to sacrifice average hand value for a higher likelihood of placing high even with a increased likelihood of placing low
It's been a while since I took a game theory class, but one concept that stuck with me was the idea of resolving game theoretic dilemmas by introducing a correlated equilibrium (en.wikipedia.org/wiki/Correlated_equilibrium ) which is essentially just a public signal that can impact the player's best interests. The classic example is a traffic light. Without it, the nash equilibrium is for the current traffic flow to maintain (never have to stop so commute is faster) and for the perpendicular flow to remain stagnant (driving into oncoming traffic is generally not in your best interest). Traffic lights are a public signal that guides all drivers to achieve a higher EV (never have to wait indefinitely for traffic to clear up). I don't have experience with poker, so I don't know how this could be applied, but it may be an interesting avenue of thought at least.
Maybe all the poker players who were interested in cooperative strategies just gravitated to Bridge :P Or these days, Hanabi (which feels uncannily like cooperative poker at high levels)
love the video, would love to see more like this
For a very simple approximation of a single bet in the 2-player situation, let b be the bet, p be the probability of winning the hand given no further information, and ICM(b) and ICM(-b) be the expected net value of winning and losing the hand, respectively. In general, if ICM(b)*p > |ICM(-b)*(1-p)|, then you'd have positive expected value for betting b. For this simple model of a single bet, for every b > 0 there is some p >.5 s.t. for every p'
These are the kind of rants I subscribe for!
Is it possible that backwards induction explains why there can be no "enforcement" style equilibrium?" Because unlike an iterative prisoner's dilemma that continues, potentially indefinitely, tournament poker must end with one player winning all the chips. So even if a tit-for-tat strategy could maintain some sort of equilibrium in the short run, a real tournament will eventually be reduced to a 2 player table. And once the game is heads up, any previous enforcement of the common good will have to break down. Because there is no common good between two players when it's just them and nobody else is left to profit off them "going to war with each". That's how I understand it at least.
So working backwards, each player knows the common good must break down at some point and they have to decide the point at which they individually abandon their own enforcement of the common good. Add to this the possibility that enforcement is only beneficial so long as other players choose to enforcement strategy and you probably end up with the case that the optimal strategy is to never enforce at all by backwards induction. That's about as far as my understanding goes though. Curious what anyone else would add to this?
poker is often analyzed as a continuous game, even past the end of the tournament. i.e. the players can be reasonably expected to sit down and play against each other again sometime in the near future. the practical applications of this do get messy though - how much can you really rely on history you have with someone from six months ago, when they're constantly learning more about the game, for example.
@@Jorbs Interesting I would never have thought to view poker, especially tournament poker, as a continuous game. But I can see how top level players will repeatedly find themselves seated at the final table across from the same pro players throughout the year.
I am definitely not smart enough, nor familiar enough with high level poker to contribute much else to the conversation but interesting video, thanks for posting it and the reply!
This kind of dynamic actually showed up in a real tournament when Gus Hansen went all in on 15 of 19 hands without even looking at his cards, eventually netting him the win. He stole many pots and twice other players tried to stop him and he got lucky on the river.
I like you're interpretation here, I think the truly mathematical approach to tournament settings is to model the event itself, which in the setting of poker transforms the game from a being about chips to a game about abstract resource management and attrition, much more similar to diplomacy than to a gambling game tied to real cash value which is how a casual game might go. In the tournament setting the 'value' of a chip is just flavorful called a 'dollar' or krown, etc, the real value is the chance to win. So maybe tournament poker would be a better game if open cooperation would be allowed. Maybe at that point though it's just actual diplo though, and the players would prefer another game.
Is it possible that poker players are just too tied to the idea of cash in hand and the flavour of the game clouds their assumptions about strategic position within the tournament setting?
Beautiful video, I enjoyed it immensely!
Back when I played online poker in the before times (before black Friday) when playing 1-table SNG this style of blind-vs-blind play was not uncommon. This was when ICM was well understood and frequently talked about but solvers and GTO wasn't a thing (at least not a thing publicly discussed on 2+2 or other similar sites).
Limping the SB is part of a balanced high EV strategy from the SB .
This was very nice to listen to thanks :)
It's good that you mention tit-for tat because it's a baseline for good prisoner's Dilemma strategy. Based on what I read a while ago, this has been tested in tournament play, and a slightly more tolerant version of tit-for-tat. But either makes sense.
But your translation of what tit-for-tat would mean into actual MTT strategy is wanting a little. So when you describe cooperative play here as, SB calls and both check down till river, that is cooperative, and definitely illegal collusion, but what I want to point out it is that it isn't what tit-for-tat would look like. If I want to play a tit-for-tat style (which isn't too far from my idea) what I mean by that at a final table, is that I want to retain as much equity in that situation as I would in a cash-game, and I won't try and steal your equity by bullying you and going all-in, but nor does it mean I am going to limp AA with the idea of passively checking down to the river. What cooperative play here would look like is both playing this relatively similar to how they might do for chips, with some slight adjustments. The cooperation occurs in a strategy vs strategy sense, rather than by checking down every individual hand.
Tit for tat would be something like this. I have 13BB in the BB. SB has 17BB in the small blind. 4 other players have 5 BB each. SB shoves. I have 99, I call. A traditional understanding of ICM would be that BB calls absurdly infrequently here because of the pressure, and SB gets to go wild and shove all hands because BB never calls. But if I start doing this, the outcome is I suffer a little (I lose EV compared to traditional ICM calculations, which are too tight) but when I do win I gain a lot. By comparison, if SB shoves 100% of hands and I call about 15% of the time, SB is absolutely burning money doing this. The unique thing about ICM in poker is that your opponent's errors can hurt you, which makes the EV of any action inherently extraordinarily hard to calculate, because the strategic dynamics are here truly complex and not just a matter of mathematics, and any Game Theory equilibrium is unstable in the sense that following it can be very costly if others are willing to hurt themselves, even a little.
your assertion that checking down to the river would be illegal collusion is very weird to me, especially in the context of the rest of your comment. i don't know of any definition of collusion which suggests that trying to maximize your winnings by avoiding detrimental ICM situations with the other players at the table is collusion. i think maybe you misinterpreted and thought i was saying players should always act in a way which attempts to cooperate in the prisoner's dilemma, which wasn't my point.
my point was more like if i was trying to demonstrate what fold equity is and said "so you can go all-in every hand, look at how much money you'll win from fold equity". i'm not suggesting that is a good way to play, or that that is the only way fold equity ever exists, i'm just demonstrating the concept in its absolute simplest form so that someone can recognize that it exists.
attempting to cooperate in a prisoner's dilemma is just one of the many reasons for players to make (self-serving!) choices in many different situations in tournament poker, and attempting to do so isn't collusion.
your analysis in your third paragraph is basically what i am taking as given in this video, followed by me trying to explain why SB and BB actually both benefit from working out how to not destroy each others stacks more often. a reading of the situation without recognizing that it's a prisoner's dilemma struggles to arrive at sensible conclusions, but it's actually not that much more complex than a cash game if you recognize that it IS a prisoner's dilemma and appropriately work cooperative lines into each players' strategies, and it's fairly obvious imo that they can both be self-incentivized to attempt to do this without needing to be colluding.
Interesting idea, but I don’t find it distressing. Like you said this is a pretty complex idea and while your logic makes some sense that doesn’t mean it’s actually correct, even when played out over a long scale. If sb only ever goes to 10 when they have aces then bb isn’t an enforcer, bb is just a sucker. Even though the chips aren’t worth as much I would wager it still statistically improves your tournament winnings to go in on those aces, you don’t care if the value of the sb and bb decrease if your value increases on average. As you said poker is a meritocracy where only the best rise to the top, so if the people at the top aren’t using a strategy you’ve thought sounded good in practice it just might be that it’s a lot messier in theory. It doesn’t necessarily mean that the idea is wrong. The best players might be smart enough to realize that most people aren’t smart enough to play along, but idk. Pretty much I don’t think the best poker players are any worse because they don’t heed the theory of another player that makes sense, but quickly muddies when taking in the complexity of poker. Also poker isn’t really a great outlook on society because it is a game. people who normally would spend hours trying to track down the owner of a lost Wallet would happily take their money in poker, because by playing poker people consent to losing their money. You are not being rude to the other players because you all agreed to these rules when you entered, there was never any expectation of a greater good.
this sounds suspiciously to the dynamics of cedh. I should follow up on this now.
I like how this video isn't about poker at all. 🤯
I think you're a little too zoomed in on the example. There's 2 players who would benefit from cooperating, but also 4 other players that benefit if they don't. And while those 4 can't do any game actions during a specific dilemma, there're ways they can discourage cooperation in general.
One really simple thing they could do is *not* talk about situations like this where cooperation is a good strategy...
I don't disagree with your main thesis - I have never seen anyone talk about this and that is strange. But collusion through trading of defect / cooperate splits can get good results if the numbers aren't set up right. Given that small and big blinds rotate through everyone they define a really easy way to 'collude' by trading. This is a UA-cam comment so I didn't actually work out any maths but this might explain why there is so little effort to exploit this in tournaments. Either way super interesting video.
Hmm, so there's an alternate universe where jorbs becomes a world renowned poker player, and revolutionizes the game by recognizing and applying the prisoner's dilemma/TotC.
Please stop teaching me things when I want to sleep. I need sleep, not more things to think about. Great video!
UA-cam listed the title as “The Tragedy of the Commons and Why I Hate…”
I clicked expecting the rest of the title to be “Ceramic Fish and Bottled Flame.” I was not disappointed.
I think that the problem is that you need norms to have an enforcer.
Suppose we have two types of players:
E - "the ego player". E is perfectly logical, assumes everyone plays like E, and always makes the play that nets the best expected result in a given round.
V - "the vindictive player". V sees that the ideal strategy is complete peace with his neighbors. V enforces this peace by outlawing aggression. If any aggression is made towards V, the game plan shifts to hurting the offending player. (AA hand is allowed if revealed afterwards, so games end eventually)
When these players meet, it is mutually assured destruction. V enforces peace, and E enforces egotistical play, both punishing the other play style. The only winning strategy is to copy the strategy of the players around you.
This results in a meta, where you lose if you don't follow the social norms because it is costly to enforce something that the other players disagree with.
In the case of poker where collusion is disallowed, i feel that E is the more justified player, as vindictiveness is purely destructive for both players unless cooperation is obtained.
(And i prefer the strategy game to the social game)
I feel that the same conclusion applies if you replace V with anything but E, even T - the "tit for tat player"
Amazing video thanks for making this ❤
A sense of community is really important for avoiding tragedy of the commons type scenarios. Getting people out and interacting with a large number of strangers in their community, however, is extremely difficult. If you look at the levers american politicians are trying to pull to solve the issue (when they're even looking for the levers), you basically find "increase the number of voluntary public service organizations and pray people join them" in some form or another. It would be easy just to have mandatory public service participation programs, but that's also like, pretty against the entire ethos of modern mixed-market liberal democratic society. Which is a problem, because the entire developed world uses that model, and only China, who is now facing demographic collapse and in the last few months has been shattering its highly lucrative tech sector into tiny pieces to reallocate their efforts elsewhere, has even come close. So mixed-market liberal democracy has no serious competitors and so there really aren't examples to learn from.
Would the ability for players who are still "in" to pay folded players a "bid" (set by the folded player), to know what cards they have been dealt, help fix tournament play?
To me there should be some sort of minor rules change be allowed to benefit tournament play and avoid typical pratfalls and accusations of collusion. What that -needs- to be I'm not sure.
Really interesting video, thank you! Would it be worthwhile to look at how computers play against each other?
Just imagine if bridge players started playing tournament poker... new rules would be written overnight to forbid such conspiracy.
Outside of poker tournaments people aren't rational self-interest driven agents, people can be swayed by morals and ideas. Now this can lead to bad outcomes but it can also prevent us from getting completely stuck in dilemmas.
Vampire bats feed their children in a group. If one of the mothers doesnt bring food but act like she did, they dont feed her children next day.
Tit for tat is simply "repeat the enemy's move next turn". If enemy ever cheats, cooperation is still possible. All of the animals use a slightly different strategy, which includes "if you've been cooperating for a long time and your opponent is cheating, it must be signal error so continue cooperating" which is game theory explanation of gaining someone's trust