Tobold's Blog
Thursday, December 19, 2013
Is chess a game of skill?

The answer to the headline question appears to be rather obvious: Of course chess is a game of skill, where your success against an opponent is very much determined by your respective skills in the game. But the reason I am asking the question is that in the context of video games, it is a lot less obvious what people mean when they talk about skill. Basically there are two parts to an action in a video game: Making the right move, and executing the move right. And more and more people apply the term "skill" only to the second part of this.

The reason for that is that the execution part is often the only demanding part of a video game. It is not an intellectual challenge to know where to jump in a platformer game, to know where to shoot the enemy in a shooter game, or to know not to stand in the fire in a MMORPG raid encounter. The challenge in all of these games is to execute it right, which is mostly a question of reaction time, sometimes targeting, and frequently of remembering some button combination or sequence. Even in strategy games the focus of the game has moved away from being a better strategist to being able to click faster and switch faster between different parts of the battlefield. If you watch any so-called "e-sports" event, you will see a lot of hectic clicking and button-pressing, and very few people in a quiet thinker's pose like in a chess tournament.

As a consequence, when video gamers speak of their "leet skillz", they are almost exclusively talking about their skills of execution. In that view, chess is not a game of skill, because the execution of a move takes no skill at all. In fact, if you did it wrong and placed your chess piece badly, your opponent or a judge would correct you.

This attitude to skill was also very much in evidence in the discussion thread of my previous post, where several people mentioned that artificial intelligence is deliberately dumbed down, because it otherwise would just crush every player. Which is correct as long as you only talk about the ability of the computer to execute video game moves right. Of course a computer can "click" much faster than a human, target with 100% accuracy, and have perfect timing in his moves.

But if you play a game that doesn't completely rely on skills of execution, but where the game is to figure out the best move, it turns out that artificial intelligence isn't actually all that great. If you play a game like Total War: Rome II, you will see the computer opponent attack heavily fortified cities with tiny armies that have visibly no chance to achieve anything; or you'll see units get lost in battle, standing around not knowing where to go. While a computer is relatively good at chess, that is because chess is a very structured game, with a limited number of pieces, each of with having only a limited number of possible moves. Even that takes already a lot of number crunching. But a less structured strategy game, with far more units and far more possibilities, is often more than an AI can handle.

Total War and such aren't more demanding for the AI, it's just that some of the brightest people on earh have spent years figuring out how to code a good AI for chess.

It's rather obvious as a good chess player and a good total war player that the later is far less complex despite the number of units involved.
I'd divide it one further, so that skill is a combination of 'knowing what to do next', 'being able to effect that action', and 'doing this quickly'. Insight, execution, twitch.

This would help distinguish between non-insight based games.

Such as a platformer where it's relatively obvious what to do and you have lots of time to do it, you just need to get the right sequence of button presses to get the character to do what you want. Heavy execution based gameplay.

Compared to a first person shooter, where on seeing a target it's relatively obvious what to do, and that is to line up your crosshair/cursor with the target, and what counts is how quickly you can do that. Heavy twitch based gameplay.

I used to get mad and roll my eyes at people talking about frames in action games and such, thinking that people with twitch skills were somehow lesser or that games focusing on them were weaker than more intellectual games, but it's really just a matter of taste. People prefer different game styles and that's fine.

An instant counterargument to someone saying a game takes no skill is that if that were true, how come some people play so badly? The simple presence of poor play proves that there is a measure for skilled play.
It's a combination of things. Chess seems like a strategy game to us because we use heuristics for long term planning (a rook and two pawns are typically as good as two minor pieces in the middlegame, and often a bit better going into endgame, that kind of thing). Computers don't really do heuristics but they specialise in deep tactical searching. Since chess has a limited number of moves, AND since sequences of moves are tightly coupled so that the precise moves matter, computers are able to deal with and even defeat our powerful heuristic abilities.

A Civ-type could probably have a good AI developed if people tried hard enough, but it would work a different way. You have lots of possible moves and fog of war if it has that (so no standard minimax tree-searching), so heuristics would need to be used. They might use some kind of Monte-Carlo approach to match strategies against typical player strategies, in a version of minimax. I don't think it's ever been done, though. The games we think have good AIs have simply less egregious AI flaws, or they cheat in a way that is cleverly designed not to be too annoying.

Maybe there should be a clarification between sports and games, and video games which require timing would fall into the e-sports category.
I think in a broader sense, "skill" means various things in various contexts. You compare video game "skill" to Chess "skill", but the same observation can be make with other things, such as sports. Does playing basketball require skill? Does hockey? What kind of skill?

Take a spectrum -- at one end, put Chess or Go. At the other, put target shooting. Golf probably sits somewhere in the middle. Different video games fit on different points on the spectrum depending on design. Pac-Man sits on one side, Zork on the other.
Interesting post.

A lot of it comes down to how we want to use the word "skill". If you can run faster than me, is "skill" the right word to describe the difference between us?

I used to play tennis, but nowadays if I played I would suck very badly because of knee problems. But is it appropriate to say that I am now less skilled than I was?

In online chess against random opponents, there is a non-trivial proportion that I win just because they are having to figure out what to do themselves on the spot whereas I remember an analysis of that position. Maybe I already know the non-obvious refutation of their plausible and safe looking move because I read it in a book once. Is "skill" the appropriate word to describe my advantage there?

There might be more of a continuum between these kinds of abilities than first appears. For example when I first started playing tennis there were serves that I wouldn't have been able to return because they were just too fast for me. Later I was able to return serves of that speed, which came from learning things about how to watch the ball properly, learning stances that meant I could launch my body that fraction of a second faster etc.

Now I'm not really into FPS or games that are about moving a mouse quicker of hitting keys faster, those are boring to me. But we could maybe stretch the word "skill" to cover some aspects of those games, just like we could with hitting a tennis serve or playing the piano.
Paradox's Clausewitz Engine is widely praised for its AI. Sadly I have never managed to get beyond the tutorial in any of their games. Perhaps I will try again over the Christmas break because I really would like to experience it for myself.
Agreed that creating a good Total War AI probably isn't inherently a more difficult problem than creating a good a chess AI; it's just that fewer resources have been spent on it.

Also, my perception of modern strategy games (OK, I'm mostly thinking Starcraft) is that you actually have to be extremely good at both aspects. It's true that good strategy won't get you too far if you can't react and click quickly, but the reverse is also true. But I may be wrong; I don't really play these games. Certainly you do need some ability to think and understand what you're doing; reaction times alone will not make you a good Starcraft player. It's possible that among good players, reaction times and speed matter more than strategy; I don't know.

This sort of thing (what your article is talking about, I mean) used to annoy me when I spent a short period playing Worms games online. Worms as a game has three aspects of skill as I see it. There's the twitch gameplay ("skill" if we like). There's the positioning, strategy, choice of weapons and so on. And there's the negotiation aspect of persuading other players that you're not the biggest threat. But virtually the whole online community seemed dedicated, through the use of arbitrary rules to eliminating the latter two aspects of the game and reducing it to a purely twitch game.
Dictionary to the rescue! From Merriam-Webster:

skill noun
: the ability to do something that comes from training, experience, or practice

obsolete : cause, reason
a : the ability to use one's knowledge effectively and readily in execution or performance
b : dexterity or coordination especially in the execution of learned physical tasks
: a learned power of doing something competently : a developed aptitude or ability

Of course you can redefine the word to mean whatever you want, but this is the meaning people seem to agree with, and it covers everything, including execution.
The common point is that it's something which is learned through practice, so all the examples given are skill: chess, running, not standing in the fire, etc.

Then of course we have the internet definition of skill, which (as can be easily learned on WoT) is:
"something I have and all other players don't"
at times generalized to:
"something the game I like to play requires, while games I don't like doesn't".

Honestly, I'm starting to feel that all the discussion is like beating a dead horse, so dead it's turning fossil…..
Then why would you want to participate in that discussion?

I don't think that there is anything like a "dead horse" discussion in absolute terms. That would require every single person in the world having heard all the arguments and stated his point.

There is an endless number of discussions on the internet, most of which you probably don't want to participate in, or don't want to participate in any more. Just pick the ones you *do* have something to contribute to. Don't tell others they can't discuss a subject any more, just because you got bored of it.
I'm guessing they dumb down the AI in total war as well.

Or not so much dumb down, but deliberately don't assign many work hours to building it. Which is the same thing.
I wouldn't call that the same thing at all. If they dumbed it down, they could leave a button to switch on the max AI for people who wanted to challenge it.

If that button isn't there, I doubt the super AI is.
The question is though, if Victory in a game (and sports are pretty much games) is ot based on skill, then what is it based on.

I think there are essentially three options.

1) Skill. Someone is simply better at it, through being innately better combined with better training and experience. Tennis comes to mind. So does chess.

2) Luck. These are games where the outcome is largely based on getting lucky. Snakes and Ladders for instance is purely a luck game, as you make no decisions whatsoever, you just roll the dice.

3) Money. Pay2Win anyone? But really even in Sports there are Pay2Win games (or at least partially Pay2Win games). Car racing is one, where budget matters a LOT.

4) Time. Some games heavily stack the deck in favour of more time invested players. I could be terrible at game X, but if I spent 3000 hours and you spent 500, chances are my time has earned me X levels/equipment etc. that simply doesn't give you a chance no matter how good you are.

The question is, how do we find out which is which?

Now Chess is very clear, it's skill. There is simply no luck involved. There is no money factor. And the Time factor does not give an inherent bonus, it just gives experience which is actually Skill. And Lo and Behold, how do we know this? Because people who are known to be good at chess, keep winning. Tennis? The same. Nadal anyone?

Luck? Poker. There is obviously some skill involved, but never has the World Champion been the same twice in a row. Hence it can't be purely skill otherwise otherwise someone would have gotten that by now. Poker is largely luck based.

Pay2Win: A lot of free2play games.

Time: Wow. If I'm level 84 and you're level 42 guess what happens?

But beyond that it gets complex. How about Formula 1? The richest (or most spending team) won years in a row now. So clearly that is Money. BUT.. Within that team, there are two drivers who both profit from that money. And the same one won years in a row. So that must be skill.

So it is really rather complex.

Just one small thing: You can win a game of Chess because you're lucky. For example, there's always a chance your opponent doesn't understand your plan before it's too late.

Naturally, luck averages out if you play many games. But so it does for any game.
ls, that still isn't Luck. It's still smarts. You made a plan your opponent didn't see. How could it be luck? Where is the dice roll? Where is the lucky card draw that allowed you to crush your opponent?

Oh, there is none. Because what happened is, you concocted a plan your opponent didn't see through. One might call that lucky but really it isn't. It is a gamble betting on doing something smart that the adversary isn't skillful enough to see through.

Don't tell others they can't discuss a subject any more, just because you got bored of it.

Where EXACTLY did I say anything of the kind?
To be honest I don't even understand how your answer relates to my comment.
You said: "Honestly, I'm starting to feel that all the discussion is like beating a dead horse, so dead it's turning fossil….."

My question is why you would make such a comment. Either you don't like the discussion, and thus stay away, or you are interested and participate. Why participate only to say that you aren't interested?
I agree with Helistar's idea that Internet posters define skill in terms of their skill/tastes.

I also am uncomfortable with the disparaging of AI. Rather the disappointingly meager game AI is a result of the lack of investment in it. Partly because it has been cheaper just to use people in the games. Storybricks is EQN is hopefully a precursor of things to come. Some of the new space games talk about having AI ships around when there are not humans available. If games were to ever have an economy, you could have NPCs buying and selling.

On one hand, AI is getting more sophisticated. OTOH, mindless twitch games seem to be winning out. So I have hope; there is much potential. But "success" is not guaranteed.


I *REALLY* disagree with Baktru on poker. Luck is a component in poker or backgammon. But skill is certainly very important. California regulates poker as a game of skill not chance.


The AI in Paradox's most recent Clausewitz-engine games -- Europa Universalis IV and Crusader Kings II, is actually pretty terrible at playing either game.

Only very recently has the AI become competent at managing its armies. And in both games, army management involves two variables: (1) Comparing army size to the supply limit of provinces in a path, so that the army doesn't starve to death, and (2) Comparing the strength of one army to the strength of another army plus reiforcements. The AI is barely able to do either. Even after the latest upgrades, the Clausewitz AI will regularly exceed local supply limits and suffer massive starvation casualties, and it still dogpiles into battles without considering the odds of victory.

And the AI still can't make sane strategic marriages, organize alliances, or think more than a few game years ahead. Both games are pretty much built around the AI's ineptitude.
@Tobold: I make such a comment because the "skill" thing gets brought up often, in forums (and here), and the starting point is always the same: redefining the meaning of "skill" to mean something which is not what it really means, followed by the conclusion of whatever the poster wants to "demonstrate" (and which always works, of course, since you redefined skill to mean whatever you need to "demonstrate" your point).
So yes, I'm starting to feel these discussions like MMO grind, but I still think I can contribute by reminding everyone that the word "skill" has a well-defined meaning, no need to redefine it to something different every single time.....

I believe it is a very important skill
Chess is only a game of skill at the very highest levels. Below that, it's a game of knowledge. Virtually every chess game up to a certain number of moves has been documented, alongside it's outcome, in "the book" (this is a real thing). If you, like a computer, could memorize "the book," you could likely beat 99% of the chess players out there: all but the very best, who know more than you and, when you get to an "out of the book" situation, could beat you based on their skill. This is also why computers can now beat chessmasters; they know enough not to have to rely on skill.

That's what games with more complexity try to overcome: a solution. Chess isn't solved and won't likely be any time soon, but there are games out there with greater complexity (which is not the same as complication. We can talk more about that another time if you want) that have more pieces as well as less pieces. Take Go, for instance, which is likely never to be solved since its board positions and possible outcomes number in the same area as stars.

Additionally, when we talk about AI now, we don't talk about it in the way we used to. AI used to mean replicating a human mind in a digital form, as in Godel, Escher, Bach (and its author, though I forget his name right now). AI's moved just to doing really well at one thing based on feeding it a ton of information about that thing. That's why you don't find "thinking" machines that can get creative; they don't have information outside of their sphere.

Crazy stuff.
Chess is a game of skill relative to the skill of the other person. If you happen to have remembered all those moves, then sure, it's just memorisation (though that's a skill).

But if you don't, and the other person doesn't know it, then it's skill.

If ones memory was deep enough then no chess play would involve skill.
I think people severely underestimate how much good decision making matters in games that appear to be all action. There are lots of people who can train up their reflexes, but that doesn't mean they will win world championships.

When you watch eSports there is a lot of frantic clicking, but when you watch improvisational jazz there is a lot of frantically moving hands too. That doesn't mean they aren't inventing the music as they go along. It means that you have to be good enough at the technical part if you want to be able to even try the thinking part.

In Starcraft 2 the hardest difficulty setting has the computer actively cheating to try to beat you and it isn't even a remote challenge for the best players in the world.

I noticed the same phenomenon a lot at the end of my WoW days. People would go on forums and talk about how the game had no skill, all twitch reflexes, but having a world leading guild (a legitimate if problematic claim), I can tell you that my twitch reflexes are extremely poor by pro-gamer standards.

Reflexes are a barrier to entry. If you can't keep up with the pace of the game then you can't play at all. Once your reflexes exceed the threshold needed to play at the top level - which is quite a bit lower than most people think it is for most games - strategy and decision making starts being the main factor.
Ha! This is so right! I had someone once told me that using V.A.T.S. in Fallout meant you just had no skill. By that logic, any of the games prior to 3 (and a whole hell of a lot of games) require 0 skill.
@sheepthediamond: Learning the chess "book" only takes you so far as the book goes, which is only as far as the opening. The game of chess goes much farther than book preparation, and a person who only knows the book but does not know the strategies behind the book or even basic tactics will get crushed by an average player.

Sure, there are lines 20 moves deep, but players can always play offbeat lines that only aim for equality into the middlegame and get out of preparation early. No one is memorizing such outlandish lines except for those at the top, and only as direct preparation against a known opponent.
Post a Comment

<< Home
Newer›  ‹Older

  Powered by Blogger   Free Page Rank Tool