Dumb AI

[padillah]

Originally posted by Stickyman
I'm afraid that the point I was trying to make is lost in communication - my fault.

Would you say it is easier to become a "good" player in Civ or in Chess?

I personally believe that it is much easier to become "good" or "proficient" in Civ then it would be in chess. As result of that belief, I think it is fair to say that chess is more complex than civ for humans.

This is a perfectly fair statement and is what MightyTiny is saying 9 or so posts up - Chess is "deep thought", Civ is "wide thought".

Humans are good at one and not so much at the other. Computers are good in the other and not so much in the one.

When computers get good at both: RUN!

Tom P.

[johnmcd]

Chess is not something computers are good at. I can't believe people are saying it's about computation. It's the fact that it's emphatically not about computation that means it's taken so long to build computators that can play it to a high standard.

Playing chess on computation is like recognising your friends by measuring the distance between their eyes and the circumference of their skull. It's the least effecient way to do it when you can 'just' recognise them. That 'just' is something no computer can do.

[padillah]

Originally posted by johnmcd
Chess is not something computers are good at. I can't believe people are saying it's about computation. It's the fact that it's emphatically not about computation that means it's taken so long to build computators that can play it to a high standard.

Playing chess on computation is like recognising your friends by measuring the distance between their eyes and the circumference of their skull. It's the least effecient way to do it when you can 'just' recognise them. That 'just' is something no computer can do.

Big Blue beat Garry Kasparov 3.5 to 2.5

That's good playing by any measure. Even Mr. Kasparov was stunned when the computer made moves he thought were too "delicate" for a computer to recognize.

If you wish to dispute how well the computer that beat Garry Kasparov played you'll have to contact him. I'll take the evidence that's presented to me.

On a more gracious note: I don't think anyone here is trying to imply that computers are "better" in the sense that they "know" how to play chess. Quite the contrary.

We are presenting the fact that computers play chess in the way you mention, "brute force", as a proof of their inability to actually "think" and reason as a true AI should.

Tom P.

Oh, btw, for what it's worth the Institute for health studies has done reasearch that intimates that is exactly how we reconize people. By the features that are most recognizable: much like a caricature in our heads.

[fed1943]

My God,even the old chess board has changed?No more 8X8?
Best regards,

[padillah]

Originally posted by fed1943
My God,even the old chess board has changed?No more 8X8?
Best regards,

Aww, crap!! I meant 64 squares (8 x 8)!

Which makes the number of possible moves at any time 64 x 16 = 1024.

NOT 4096.

Thanks for the help,
Tom P.

( Doesn't lend a lot of credibility to my argument when I can't do simple multiplication, huh? )

[johnmcd]

Originally posted by padillah


Big Blue beat Garry Kasparov 3.5 to 2.5

That's good playing by any measure. Even Mr. Kasparov was stunned when the computer made moves he thought were too "delicate" for a computer to recognize.

If you wish to dispute how well the computer that beat Garry Kasparov played you'll have to contact him. I'll take the evidence that's presented to me.

On a more gracious note: I don't think anyone here is trying to imply that computers are "better" in the sense that they "know" how to play chess. Quite the contrary.

We are presenting the fact that computers play chess in the way you mention, "brute force", as a proof of their inability to actually "think" and reason as a true AI should.

Tom P.

Oh, btw, for what it's worth the Institute for health studies has done reasearch that intimates that is exactly how we reconize people. By the features that are most recognizable: much like a caricature in our heads.

I have no idea what you're point is to be honest.

My reading of the above posts is that various folk posted that chess and civ vary in the degree to which a computer can be expected to play them 'naturally' with chess quoted as a game that suits computational ability and Civ quoted as a game that does not. My point is that neither suits computational ability and nor does recognising a face. Of course we recognise people by their physical charactaristics (what else could there possibly be?), my point is that we do that in a way which is fundamentally different from the way a computer with a camera tries to. Much as we play chess or civ in a way that is fundamentally different.

[padillah]

Originally posted by johnmcd
Of course we recognise people by their physical charactaristics (what else could there possibly be?), my point is that we do that in a way which is fundamentally different from the way a computer with a camera tries to. Much as we play chess or civ in a way that is fundamentally different.

My point was that just because the manner in which something is done is different that does not make it "good" or "bad".

Chess very much does lend itself to being played through pure "brute force" computation. Since that's the way it's being done, and it's being done well enough to beat the best player in the world, I'd say that's proof enough that it can be done... and done well.

How do you explain the multitude of chess programs currently available? What are they doing if not playing chess computationally?

Tom P.

[MightyTiny]

Originally posted by johnmcd


I have no idea what you're point is to be honest.

My reading of the above posts is that various folk posted that chess and civ vary in the degree to which a computer can be expected to play them 'naturally' with chess quoted as a game that suits computational ability and Civ quoted as a game that does not. My point is that neither suits computational ability and nor does recognising a face. Of course we recognise people by their physical charactaristics (what else could there possibly be?), my point is that we do that in a way which is fundamentally different from the way a computer with a camera tries to. Much as we play chess or civ in a way that is fundamentally different.

My point was NOT that computers could be expected to be "naturally" good at chess, but rather that it is orders of magnitude easier to program a computer to play a "deep game" like chess well (or at least adequately), than it is to program a computer to play a "wide game" like Civ4 well.

And that this is because what computers do have is pure, focused, calculating power. Problems that are deep, precisely defined, with a specific, known and reasonable limit to moves that can be made at each step, like chess, can be "solved" by brute calculating power even if the algorithms aren't optimal, at least well enough to give a human a run for their money.

I am NOT saying that the algorithm doesn't matter - of course it does, and HUGELY - but what I'm saying is that with what is currently known in the algorithm front is enough: in deep, completely defined problems brute calculating power can compensate for any lackings in efficiency far enough to provide a human a challenge.

In games like Civ4 by contrast, which are "wide" problems where you can never have perfect knowledge of the possibilities or the state of the playing field, and where you have an inordinate number of things you can do at any moment, brute calculating power is next to useless. What is needed in these sorts of wide problems is the kind of massive parallelism and superior "algorithms" that human brains exemplify. In other words, in wide problems you CANNOT compensate lack of efficiency and brilliance in algorithms with more processing power, not in the least. Trying to solve such wide problems by throwing more brute force at them would be like trying to put out a city wide fire by calling all your firends to piss at it too, once you find that you can't do it that way alone.

And this is why games like Chess are so much easier to get a computer to play proficiently than games like Civ4.

[Loader2k]

I find the AI disappointing too.
Appart from obvious glitches the overall performance
doesnt seem very impressive.

I disagree with the statement that cIV is more difficult that
Chess. Its much easier to play a decent game of cIV than
of Chess. Its true that Chess is better suited for a purely
crude force approach than cIV, but so many options of cIV
can be thrown out. In Chess it makes a completely
different position whether a piece is standing at a field or
at its neigbouring field - in cIV much less so. If a city is
threatened then you -doh- better have a number of units
close by so that you can move them into the city or attack
the enemy. cIV solves this problem by putting some
garrisons into its cities and never moving them. Often it will
not even attack with all its catapults to harm my stack of
doom. It will not try to destroy your stack of doom by
massing troops. It will just stand in its cities, completely
paralysed, waiting for me to pick one after the other.

Of course Chess is a research topic for itself while cIV is a
commercial game. Thus much more effort went into
computers playing Chess than the cIV AI.

Still, since the AI is not much of a factor whether a game
sells well or not I think it is rather probable that it has
been a little neglected.

[MightyTiny]

Originally posted by Loader2k

I disagree with the statement that cIV is more difficult that
Chess. Its much easier to play a decent game of cIV than
of Chess.

I don't think there's anyone in this thread that has said anything that would contradict with that. Of course FOR A HUMAN it is much more difficult to reach a high level of play at chess than to become proficient in Civ4. That's obvious.

The whole point is that feats that are trivial for humans are often difficult to replicate with a computer AI, and the other way around: what is a difficult mental feat for a human, is much less of a challenge to make a computer do well.

I don't think this can be said any more clearly than it's already been said in this thread.

I do agree with you though that it seems obvious that the AI in Civ4 COULD be a lot better, and most probably AI development hasn't received but a fraction of the attention that's gone into other, more clearly visible and eye-catching aspects of the game.

After all, a good AI can be seen mainly through the ABSENSE of idiocy, and that's unfortunately a thing much harder to assess (and thus use as a selling point) than cool graphics.

[Loader2k]

I don't think there's anyone in this thread that has said anything that would contradict with that. Of course FOR A HUMAN it is much more difficult to reach a high level of play at chess than to become proficient in Civ4. That's obvious.

I didnt necessarily want to restrict my statement that
Chess is easier to play than cIV only for humans. I think
that cIV is for several reasons, one of them being a more
continuous game world, another one being that most
problems are local - independent of the rest of the game
world - "easier" than Chess. What is true is that cIV isnt
feasible for a purely brute force approach what many seem
to equal with AI in this thread.

I do agree with you though that it seems obvious that the AI in Civ4 COULD be a lot better, and most probably AI development hasn't received but a fraction of the attention that's gone into other, more clearly visible and eye-catching aspects of the game.

Yes, indeed. But fortunately a lot of attention has been put
into an area that also isnt so eye-catching at the first
moment and thus probably wont be too much of a factor in
sales - game balance. I think its excellent and after my
moaning about the AI i wanna acknowledge that. Great
work, great game.

[Rook]

This reminds me of something I tried a long long time ago. I had an electronic chess game (that would kick my butt readily by the way) and I also had a chess game for my Commodore 64. I thought it would be fun to have the two play each other. Whatever the C64 did, I told the electronic game and vice versa. It worked for about 15 turns and then they just kept moving the same two pieces back and forth. I could never get them to play a whole game.

[padillah]

One issue I see a problem with is that people are judging chess based on hundreds of years of the best players in the world analysing it. But, since any yahoo can beat CIV, it must be easy.

(I'm gonna get killed for this, but here goes...)

I hold that anyone that believes Civ to be an EASY game have a one on one multiplayer with Vel and tell me it's an easy game. (Sorry Vel, but you're a recognized name). Anyway I hope you see my point, the GAME of Civ is a different animal than the AI that plays it.

Our complaints are about the AI, and are simply what is being said, you CAN beat it even if it "cheats". That may be a fair assesment of the AI but it's not a fair assesment of the GAME.

If Civ were brought into the world of board games then we could equate the two GAME SYSTEMS. But there's no use comapring chess, as analysed by the entire world, and Civ as played by the, widely regarded, lame AI.

Tom P.

[LDiCesare]

It worked for about 15 turns and then they just kept moving the same two pieces back and forth.

After three repeats, this is a draw, so the game ended in a draw.

The ai in civ has to assess a global strategy and local ones. It's got many problems with its global assessments: When determining whether it is stronger than you or not, it seems to look at the demographics data, realise it has more army and think it's stronger. Except its units are scattered in its cities, while you have concentrated them near one of its cities. You ask for peace. It evaluates the global strength: Globally it's stronger than you, so it makes silly demands. You take the city. Result: It realises it is losing the war afterwards and makes some concessions, but usually not enough to prevent you from taking whatever you want.
Thus a big problem is an evaluation of global vs local effects. In Chess, it's quite easy: You can decide of a strategy and compute the tactical consequences up to 5, 10 turns for instance. In civ, the ai doesn't even look forward 1 turn the outcome of a fight, or if it does, it doesn't conclude anything intelligent. Of course, the outcome of a fight is random. Additionally, there is a fog of war which doesn't exist in chess. And the board changes with time (resources popping up, canal cities changing the topology of the seas).

[Loader2k]

I hold that anyone that believes Civ to be an EASY game have a one on one multiplayer with Vel and tell me it's an easy game. (Sorry Vel, but you're a recognized name). Anyway I hope you see my point, the GAME of Civ is a different animal than the AI that plays it.

That's true, a strategy game is only so difficult as your
opponent. But there are strategy games that are easier to
master than others. Chess and Go are hard, some game
for children with little strategy in it is easy to master, cIV in
my opinion is definitely easier than Chess or Go.
Now the game of cIV apart from the AI is still demanding
enough to be very entertaining - as i said the game
balance is very good. But without decent oppenents...
well.

If Civ were brought into the world of board games then we could equate the two GAME SYSTEMS. But there's no use comapring chess, as analysed by the entire world, and Civ as played by the, widely regarded, lame AI.

That's absolutely true. One has to wonder what level of
playing strength from the AI can be expected from a
commercial game. Certainly not one that is comparable to
modern Chess programs as these programs are sold only
because of their playing strength (well, a little eye-candy is
in) and algorithmic Chess has been studied for decades.
On the other hand one can expect more than complete
moronness from the AI imho.

Thus a big problem is an evaluation of global vs local effects. In Chess, it's quite easy: You can decide of a strategy and compute the tactical consequences up to 5, 10 turns for instance.

I disagree. In Chess it's not easy. You dont have any local
problems in Chess. Moving a piece anywhere on the board
affects the whole board. The existence of local effects
should in fact enable a better AI since local problems are
independent of each other and can be solved
independently, reducing the problem size *dramatically*.