Skill and Chance

This may be too obvious to blog about, but the question of games of skill vs. games of chance just comes up often, and discussions invariably involve some confusion. Here is a simple way of explaining it:

Chance means two different things.

It’s the same issue with Roger Caillois’ distinction between Alea (chance) and Agon (competition) – after all, many games of competition also contain chance elements.

Chance can mean either:

  1. That a game contains a chance element or mechanic.
  2. That the outcome of a game is determined by chance (the game is not a game of skill).

And the two meanings are constantly mixed up. That’s all.

9 thoughts on “Skill and Chance”

  1. Yes, but there are a cluster of issues around this. E.g., Chess players tend to look down on games that have any element of chance because they are “luck dependent”. But most games that do have some element of chance ARE ultimately games of skill. E.g., Poker relies on random card distributions, but you typically play many hands, and as the number of random tests rises, the bell curve sharpens, so that outcomes tend to collapse to the mean–and the strategy comes in knowing when to fold and when to hold, that is, how to respond to a particular set of random events.

    Parlett argues that, for primitive peoples, games of -pure- chance have a quasi-spiritual element, in that people view them partly as a way to test how favorably they are viewed at this instant by the gods/spirits/what have you (an approach reminiscent of the gambler in Rififi, who keeps a slot machine in his closet and uses it before he goes out to see how ‘lucky’ he is today). But if you approach the universe with from a more rationalist viewpoint, then games of pure chance become uninteresting, since outcomes really are essentially irrelevant. Unless there is money at stake, which surely explains why nobody plays Craps, say, “for fun”–it’s a gambling game, not a “game of entertainment.”

    In general, this is why I view gambling games as essentially corrupt: if a game isn’t interesting unless there’s money attached, it can’t be a terribly good game in se.

  2. Jesper, you need to end more posts with the words “More research is sorely needed”. It’s sort of your duty to the wider research community :-)

  3. Greg, the first part of your comment about many games containing chance elements being skill games was more or less the point I was trying to make.

    Perhaps money stakes are like movie licences? Some games can’t sell without.

  4. How do you treat games such as Rock, Paper, Scissors whose entire gameplay element is essentially “Player Controlled Chance”. It’s not like a strategy game where you can read an opponent and attempt to counter… it’s a quick decision that must be made in a short time.

    Could this be considered a game of “Chance” as there’s little likeliness that one person could dominate a room of people in Rock, Paper, Scissors… unlike a game of Starcraft which utilizes a similar RPS-style of using units… but on a wider scale, of course.

    Technically, RPS has NO luck or chance-dependatny elements but the nature of the game leads me to believe that RPS is actually a game of chance.

    Sure, you could put odds on the chance that your opponent is likely or unlikely to throw the same hand twice… but it’s still entirely a guess.

  5. I wrote a big piece a while back on “fairness” and how randomness fits into it.

    The poker example is a good one. In a sense, when you enter an element of chance, you force the skill “upward”, out of the core. One round of poker proves nothing, but definate averages appear over the course of many rounds. Infact, the skill becomes about reading your opponent and his/her trends – not the cards you’ve been picked.

    Same deal in something like counterstrike, or battlefield 2. The random spray on a weapon is enough to take aiming out of the picture – simple firefights become something of a crap shoot. The skill comes in being good at getting the drop on your enemies. You’ll work in a very cerebral way to outwit them so that you get the first shot in without being seen.

    Games with fewer random elements are quicker at guaging the skill of a player/a team at that skillset. Games with more random elements take longer to establish the quality of the player.

    It’s also interesting to make the distinction between true randomness, and percievable randomness. To a beginner, the physics in a video game might feel a bit random because so many rules are combinding to create incredibly diverse outcomes from even slight deviations in input. But a master player might learn the nuances of such a system, and be able to put them to use in an intentional way.

    Consider card counting in Black Jack. And craps… well… it IS a game of skill. With enough skill, it would be possible to score the rolls you wanted every time, (at which point, it becomes a choice, and an uninteresting one at that, which might be the real reason why it’s not a great game without the money). It just happens that the skill required to master the dice is an inhuman barrier to entry, and therefore, we default to calling the die a random number generator.

  6. Isn’t it a bit much to say that aiming is out of the picture in Counter-strike?

    But speaking of counter-strike, this discussion about chance reminds me of asymmetric games where the teams (or players) have different goals and means – they also tend to work only if you play the game a number of times (“take longer to establish the quality of the player”).

    On the other hand … many social games with chance do not establish the quality of the player, and that is the point, no?

    PS. Aubrey, your comment was sent to the moderation queue because it contains the word “poker” …

  7. Well, about counterstrike, that’s sort of my point about “degrees of fairness”. The skill is not completely lost, but it IS watered down when you compare it to something like Quake 3 where weapons are (on the whole) far more consistent.

    That ends up meaning that people playing counterstrike eventually reach a threshold of skill, above which, any expression of higher skill is masked by the noise introduced. For example, if player A is able to aim at player B in a split second, but player B aquires his aim just a split second slower, then the random aim offset tends to cancel out the minute advantage player A’s skill gives him. When the skills of both at aiming are that close, the outcome of their firefight will be determined by the element of luck. In that sense, mixing a random element with a skillful one blurs ones’ appreciation of the player’s skill at the pure interaction.

    Now, in Quake 3, with a lightening gun or rail gun, that split second does make a difference. As a result, the games’ appreciation of skill does not require that it be repeated as many times to appreciate the skill of aiming.

    But as I said before, it means that the focus of CounterStrike is less on aiming, and more about getting the tactical jump.

    Note that i’m not talking about the “luck” inherent in physical human interaction – i’m just talking about the randomness designed into the game itself.

    I’m not trying to say that randomness is inherently “wrong” or “bad”, although, at the core level, it does tend to fuck with ones’ sense of percievable consequence – atleast until you get used to the standard deviation of inherently random mechanics. Because of that, I definately used to think that “randomness” was simply bad, and while I think it makes for a more interesting game to design a percievably consistent system rather than default to random behaviours, I do get that randomness has justifyable uses.

    Certainly, by embracing a pseudo realistic spray pattern in shooty games, when you play you do come to understand the reasons why real SWAT teams and armies in general take the methodical strategies that they do. It seems that it’s partly to do with mitigating the risk of a lucky head shot.

    And, Dr J. , I don’t mind being moderated. If anything I need it!

  8. I just realized that my example in counterstrike was sort of wrong.

    The mean of players’ kills would still represent each ones’ relative skill if repeated enough. So the peak of the bell curve would be slightly further up on a kill/repetition graph for player A than for player B.

    My bad.

    Still, the closer the skill, and the more the randomness the longer it would take to definatively discern the better player of the two.

Leave a Reply

Your email address will not be published. Required fields are marked *