« My Tiny Life Now Free | Main | MacArthur Series on Philanthropy and Virtual Worlds »

Jan 17, 2008



This doesn't affect the substance of your argument, but I don't think you are correctly identifying the deep games of poker. The aspect of poker, not spelled out in the rulebooks, that really sorts out the winners from the losers, is the question of probabilities and pot size. The winners are the people who, in general, push hard when they recognize their positive-expectation overlay and fold otherwise; whereas the losers are much less discriminating.


Yes, I should be clearer, Alan. That is in fact the most profound "deep game" of poker, and I think that actually takes even longer to learn properly. I focused on position because that's what I feel like I fully understand now (in relationship to pot odds, probabilities of cards hitting and psychology).

The classic example that a lot of people point to in hold'em is that many novice players overestimate the chances of hitting a flush if they hold two suited cards and there are two cards of the same suit on the board. But of course the player's assessment of the actual probability of hitting the flush should be heavily affected by the betting strategy of other players, which is in turn affected by position and psychology.


Sounds like a call for fieldwork to me :)


Tim -- can I ask you to flesh out a bit what you mean by "deep game"? In talking about Poker, it seems you're talking about the required path to victory. I take it when you talk about a "deep game" in VWs, you're also talking about something of that sort -- the way to get the epics, defeat the final bosses, climb every mountain, yadda.

Doesn't that presuppose that VWs are the kind of game that is played that way? In other words, aren't you taking the achiever Bartle type, so to speak, and using that viewpoint to define the essence of the game?

And even if so, while I think you're right that social organization is the deep game in that sense (see Nate's slides which suggest that winning at Eve is mostly about successfully creating the virtual firm -- complete with branding and egoboo rewards), I wonder about your second claim, which seems to me more of an ontological problem than a strategic issue.

The quasi-exploit, it seems to me, should present the person seeking VW mastery with a serious challenge of interpretation. If Boss X is intended to be defeated by Y means and I discover that using Q means actually makes the challenge presented neglible, how do I respond.

If I respond by using Q means, with an understanding that Q is an exploit (not considered by the game maker), then have I prevailed or discovered a mistake in design?

One way of approaching it, which you seem to endorse (if I'm reading you right) is that there are no mistakes, just opportunities. But I think there are some achievers who would say Q is an exploit.

That leaves you with a social question -- what game should being played? The way that particular question doesn't seem like a deep game to me, because it doesn't seem to have any guiding structure.


Your final warnings concerning monoculturism ring a bell. Could it be, in fact, that “social intelligence” consists entirely of knowing when and where to restrict free play?

It may be necessary, to answer this question, to decide on the systemic functions of play. Does play “build” social intelligence? Do we “learn” social intelligence from play (as you seem to imply)? Is that the point? If so, then one might question whether play among those with these social intelligences in place is as playful as play among those with no such thing.

To me, it seems not. (But then, of course, I seldom equate play with arbitrage.)

Once players are turned into guild-bots, their capacity for play may be diminished and, as a result, any play which explores and, in that process, potentially exploits game mechanics is likewise diminished. If this exploration and exploitation (which is really just letting the chips fall where they may) is the point, then social intelligence functions to deny rather than achieve play.

If we can agree that play is, at minimum, doing something, then social intelligence can be almost entirely described as knowing when NOT to do something. Or: The function of social intelligence is to maintain social structure; the function of play is not.


These ideas have been perfectly played out in games like Ed Federmeyer's Haunted Maze, one of the most comprehensive examples that I've seen of improving social intelligence through a, on the face of it, fairly simple game system.

The other example of this is seen in games like Pokemon where, very quickly strategies and counter strategies are guessed at from very few obvious clues. In the end though, the pokemon community regularly bans many combos and strategies as they become known and spread through the community. Marriland's exploitative employment of the second-gen black listed Skarm-Bliss combo is a recent case where a falsely modest player (Marriland) was exploiting the newbies with a tactic that veterans scoffed at. Interesting.


Actually, re-reading the OP, I think we're actually pretty close on this question.

I'd disagree, though, that the state of affairs tends toward monoculture -- if you see Julian's recent Wired article on griefers (Feb 08, p. 96), and Eve Online (nod again to Nate's recent post), he quotes a GoonFleet profile: "You may be playing EVE Online, but be warned: We are playing Something Awful."

So you actually end up, at least in the case of Eve, with factional conflicts about the nature of the game.

Anyhoo, to the extent you're arguing that it would be better if MMOGs had the strategic and gameplay depth of, say, chess (I actually find the psychological poker game kind of tiresome and not very game-like), then I'd generally agree, but if they did, they really wouldn't be the same game, would they?


I think it pulls in two different directions, Greg. The social/cultural game is as open as the world itself is, at least notionally, limited only by the plasticity of the gameworld in relationship to the social imagination of the players. (So EVE, highly plastic and full of rich sources of mimesis to the 'real' world: the sky's the limit. WoW: rigid, and generally tries to break its relationship to the sociality of the world? not so much.)

But there is also a game: the achievement, accumulation game. And there I do think the movement in Diku-style VWs is towards a persistent grammar of gameplay over time so that the more you've played, the more you know how to look over any given VW and spot the likely sources of arbitrage and advantage, the likely places where unexpected effects and hidden properties will emerge. The example of Pokemon is one good one; another is Mario Kart DS.

It strikes me that when a game's rules need to be changed in order to prevent a "deep game" from being played, or you need to create strong social conventions that outlaw or prohibit one such deep game, then there's a problem with the game itself. Ideally, you want the deep game to fulfill the game as specified by its rules, to create endless hetereogeneity within the activity of play. So yeah, I am arguing that it would be better if MMOGs had that depth--and yes, they wouldn't be the same game. They'd be better as games and maybe even as worlds.


A few points here...

1. I think (but I'm not positive) that your idea, Tim, that there are two Deep Games to poker is fallacious. I'm going to go out on a limb and guess that since you have recently learned more about the importance of physical position at the table, that concept is over-salient for you right now. My contention would be that your other Deep Game of poker is the only one.

2. MMO's like WoW/EQ/AC etc. are designed with a primary emphasis, it seems to me, on the treadmill - the "kill shit, buy shit" treadmill of advancement (as has been noted...sorta). I suspect that part of the Deep Game of MMORPGs I've seen is in the aformentioned arbitrage ,the search for the efficient way to level, etc....for some people. For others, it's really about the social aspect, even in WoW, I think. For others, it's social in the PvP sense...pwn those n00bs and such. And for still others, a few at least, it's about a financial model: how to rule the Auction House, etc. And there are surely other mindsets that I don't know.

3. As far as mono-culture, I think that things evolve in steps and then plateau for a while. I think the most popular (in the US) MMOs have hit a monoculture state. The game mags are full of stuff on it. (Why? because, for now, that's what the public is buying/playing...it'll change before long.) Attempts to open up the genres and styles have largely fallen flat. EVE seems to me the biggest exception. But I think that by the end of 2009 we'll see some very different MMOs rise. What their Deep Game will be I can't guess.


I'm not entirely sure that the monoculture as play-style is as big of a problem that we may make it out to be. While understanding of the 'deep game' in world of warcraft might be less transformative that that of Starcraft, it is still present. By starcraft, I mean the gulf between high level play and entry play on the multiplayer side. Not only will players move faster and more efficiently as they become better (largely just by degree), they will deploy units and buildings in a wholly counterintuitive manner. Approaching a starcraft game from the point of view of a novice results in some fairly plodding, straightforward play. Watching a world-class player from the viewpoint of a novice is like watching roger Federer play tennis. You know he's playing the same game, the net is still there, the ball is the same, but it will never BE the same game.

The difference between a player who understands group synergy, threat, and rotations in world of warcraft and one who does not may not be as wide as that of Starcraft (or poker), but it is there, to be sure. The difference, in my mind, is that interaction between players of different skill levels is pretty limited by design--again, an architectural feature. It is no longer important that my understanding of raid mechanics is strictly dominated by someone else in a different guild, because we face the same challenges in an instanced world. The spoils of our accomplishments are non-rival (for the most part). Strictly speaking, that can't be true in poker. We find there direct clashes and opportunities for arbitrage where we cannot in world of warcraft.

But this lack of rivalry does not make play in warcraft any less deep. It simply means that we have to search for natural experiments and other indirect methods of comparison to tease out deep play.


"Players frequently discover that some aspect of play has unintended consequences that are disproportionately advantageous."

I think it's just ridiculous how people think this is cheating, abuse, etc.

I play fighting games and CCGs for the sensation of winning, and that means that I'm playing a different game. I'm intentionally seeking out these disproportionately advantageous situations because they maximize my chances of winning.

People who think that it's wrong to do so are free to make their own arbitrary rules about those situations and to play their own game, but they need to realize that they cannot be remotely competitive with those rules in place. Their best bet is to continue having their own casual fun, which is perfectly fine (especially so in MMOs).

Players are not at fault for doing what is possible in a game. If the designers of a VW decide that an exploit is threatening the harmony of their world, then they're free to fix it. But the players are not at fault.

(Hacking, mods, etc. are not a part of the game.)


"Players are not at fault for doing what is possible in a game. If the designers of a VW decide that an exploit is threatening the harmony of their world, then they're free to fix it. But the players are not at fault.

This is not a settled subject. It is easy for me to say that users should have an idea of "I know it when I see it" for fair play and it is not the responsibility of the developers to provide a world where all unfair options are eliminated by design.

This is a little like saying that the packaging and sale of sub-prime mortgages was legitimate insofar as it was not expressly forbidden.

(Hacking, mods, etc. are not a part of the game.)"

Mods are not synonymous with hacking. In some games, mods are expressly allowed by the service agreement between the player and the game maker.


Part of the problems of 'exploits' vs edge of acceptable mechanics, really is a fine line.

I'll give a pair of examples from my MMO of choice, EVE.
Station camping, where a group of players all park outside of the undocking port of a hostile space station and attempt to kill any hostile player that undock is regulated by mechanics to try and allow the player a way to avoid death. Generally on undock the player can not be targetted for perhaps ten seconds, and this gives the player time to either plot a speedy get away, or under unenviable odds, redock and wait out the camp. One way however of killing an untargettable player is the use of "Smart bombs", an area effect weapon that damages everything within a certain radius. To mittigate this however, CCP implemented code to prevent them operating within a certain radius of the station. Easy? Sure.

However with the introduction of the titan, it was found it was possible to park the titan just outside of the radius, and the huge 'collision box' , plus the radius of the rare-spawn smartbomb explosions meant that it was possible to instantly kill anyone undocking before the timer wore off.

Now, since it wasn't *banned* to smartbomb at undocks, just generally not viable, it became a complicated question of whether it was an exploit, or just a possibly overpowered tactic. GMs seemed to lean towards 'overpowered', much to the upset of the captive players unable to participate due to an unescapable station.

The "Player Owned Station" is a large forcefield bubble near a moon with a control tower. Generally a player inside the bubble can not be targetted or killed, and can not be smartbombed. Furthermore an enemy generally can not enter it either. These POS's are generally held to be a respite where one can rest (and AFK) from the game.
A certain alliance discovered that a Titan warped into a "POS" held such momentum it would temporarily penetrate the shields and bump everything outside of it, thus subverting the shields. Generally this was however understandable from the physics of the game. Any ship can warp into the shields, they however will be auto-ejected quickly, and do not hold the mass to bump anything outside.
This was held to be an exploit, despite it not being a 'hole' in game mechanics (Although the player base DID feel it cheating)

The question is, how would said Titan pilot *know* where the line is drawn. Both are seemingly permitted by mechanics, both are large-scale "griefing" sorts of techniques, but one is allowed and the other not.

(fun fact: After 4-5 month of terrorising the enemy, and thousands of deaths later, said Titan was eventually captured and destroyed. A certain well known TN journo implied in 'Wired' that this was "griefing" the titan. No I don't get it either)


I completely agree with Alan Botkin. Deep Game, or the Meta-Game, in poker is all about Expected Value. This post is going to be less about MMOs and more about praise for Poker.

In Poker, players need to work their way through a lot of concepts, and what I like to think of as "levels of sophistication". The first things you learn are basic odds. You learn position, starting hands, race odds, and the rules of 4/2 (or if your a numbers guys just memorize the real odds). You play tight, and have an edge (or at least you did 5-10 years ago) against the average guy regardless of reads. People play crazy and you whine about "bad-beats" saying, "do you know the odds of hitting that crazy draw/race/whatever?".

Then as you get better you begin to understand game dynamics and reads. When I mention "reads", I mean you begin to stereotype players based on their behaviors and actions you've seen. Their hand selection, bluffing tendencies, betting tendencies, hands they've shown down etc... You put them into a stereotype so you can make decisions based on how you assume they will act in a given situation. At this point, physical tells and the "psychology" as you described are a quagmire and unnecessary. Physical tells, even for the most intuitive are going to require SO MUCH INFORMATION to sift through the bullshit and the edge they grant is SO SMALL its not worth worrying about.

Next, you begin to understand how and why people play loose and seem to "ignore odds". They begin to not only stereotype other but stereotype themselves. They understand how others and the table perceive them, and can make decisions based on how they think others view them. This is when players begin to understand "implied odds" and thus begins players journey through "levels of sophistication". "Implied Odds" are when as a straight up bet, you are now getting the proper odds to continue. You take the bet anyway because you believe that if you make you hand, you will extract enough out of your opponent that you will end you getting the odds you needed.

Two very basic examples: 1) NL Hold Em: You are getting 3:1 to call on the turn with your flush draw (obviously you believe all nine outs are live and will give you the best hand). You need at least 4:1 to call to receive proper odds. However, you believe that your opponent will pay off a big bet on the river if you make your draw. The extra money that you will receive on the river will give you much more the the 4:1 than you needed. 2) NL Hold Em: You are dealt a small pocket pair, let's say 55. You call a small raise from a Middle Position raiser. When it gets to the BB he puts in a sizeable re-raise. You have judged this player to be very tight, and are pretty sure he can only have QQ, KK, or AA. You also have observed that he will ride these big pocket pairs to show down. The odds of flopping a set (seeing one of the two FIVES on the flop) is 7 and 1/2 to 1. When the action gets back to you the pot odds are only laying you 2:1, not even close to the odds of you flopping the best hand. However, you believe that if you did flop a set, he would go broke with his pair. If he has enough money left in his stack (and since you believe that if you flop a set he will go broke) you can call. If the pot combined with the rest of his stack would give you 8:1, you would have the "implied odds" to call to see a flop.

As you get higher and higher in "levels of sophistication" your actions involve more and more of these considerations. The player stops caring about crazy bad beats and ridiculous hands because they know its the long term that matters. If they are making correct decisions based on Expected Value in the long run they will come out ahead. This leads to a situation where if the average observer watched you play one hand, the observer might have no idea why you would, "make such an obvious play, run such a ridiculous bluff, make such a stupid call, etc...". Because of your reads, table-image,and better more sophisticated opponents, you have to make plays which, viewed singularly would seem crazy or wrong, but when viewed in the context of the table history and your history with the opponent, you believed to be your highest "Expected Value" decision.

Better players incorporate more and more of these layers of strategy and sophistication. With Old Pros, they begin to rely more on physical tells because they believe the match is so even that the tiny edge a physical tell could grant would give them an edge. Furthermore, when your dealing with such large sums of money, those tiny percentages mean big money.

The point is I have never seen these levels of sophistication in MMOS. I have always seen MMOs to be a more static, odds-based game of strategy. Yes there are dice-rolls, and a level of "Expected-Value" in that you did the math and you lost because of an extreme variable. However, I've never seen the levels of sophistication and need for constant adaptation in strategy / game theory of poker in MMOs. I like games, I like MMOs, and I like poker. I would never claim to be a great poker player or understand all the ins and outs. I like MMO strategy and trying to execute the most efficient strats. I simply like them for DIFFERENT reasons and don't see the parallels you attempted to draw.

P.S. If I missed a key point in your argument, I apologize. It is 6AM and I'm a little out of it. I just think when it comes to "Deep" strategy no MMO has ever come close to Poker.


Just reading the comments reminds me of good old offline play, (remember that?), with games like beat em ups and FPSs. With a small group of friends you'd very rapdily get an evolutionary arms and defense race so that within our small group we would know which character to pick, or weapon to collect, to counter act the other players and ultimately win the game. This very highly evolved tactics and counter tactics system of play was then completely useless when an "outsider" would join our group and use "old tactics" that no one thought to revert to, or use mind blowingly obvious tactics which would be highly effective against us. Then that everyone would build up strategies or play styles to counter act that player until another newbie entered the fold.

Then everything went online so you just have to keep checking Youtube to keep up with the exploitable glitches and cheat tactics to get or counter act the cheap win.



I am buying very little of this. That would be very little as in zero and zip.

Problem remains with "social intelligence" in that it is always more socially intelligent to maintain the social rather than to disrupt or destroy the social and, therefore, social intelligence will always be really dumb re the latters. Problem with "expected value" is that expected expected value is then actually more important, and, then, expected expected expected value is even more important than that. And then the list goes on.

I would expect ditching the expected portion of expected values and just going with real and actual values (regardless of their expected on unexpected status) would be the way to go.

But here's the test. Develop a computer program that plays by the numbers and totally by the numbers and not by the tells nor by the expected tells nor by the expected expected tells. And then play the computer versus the expected values guys. I expect that, over the long term, the expected values guys will do most of the expecting and the computer will do most of the winning.

If so, then is the computer socially intelligent, or is social intelligence something closer to social ignorance?

The comments to this entry are closed.