There was that time long ago that a fully operational moon-sized battle station far, far away together with full fleet went against a rag-tag band of rebels. As I recall, the rebels exploited a tactical weakness to win the day on that occassion.
pan
[quote]
although Kasparov drew the rematch).
[quote]
In a way, this proves that tactics are far more efficient, which is the basic point of the whole post. Kasparov, though immensely outmatched in focus by the most powerful supercomputers, is just about a match in practice. This is basically because he knows what he is doing. The machine, on the other hand, does.
I doubt that computers can get relatively more powerful than humans, though. The difference between ten million and 1 hundred million considered moves is rather small compared to a human’s processing power.
The basic end point here is that depends.
The utility of tactics and force is contingent on technology, battlefield, morale, intelligence of the commanders (the best tactics and force in the world are useless if you don’t know how to use them) and many other factors. Its a matter of relative strength and which one is best for a given situation.
And that doesn’t even mention logistics.
Um, I’m sure you meant to say:
-
far seeing general strategic knowledge, plus competent tactical ability (such as Kasparov’s) is pretty well balanced by complete short term tactical calculation (computer). (Note also that the computer feels no emotion, especially risk of losing.)
-
‘the computer does not’ (know what it is doing)
That may be true of programs doing calculations, though every hundred-fold increase in speed of calculation of positions (not moves) means the computer sees perfectly one pair of moves further ahead.
But computer databases on endgames are already unbeatable, since they store all possible positions and just look up the next move.
So far these have reached all positions with 5 pieces. Once they reach 32 pieces, it will only be worth playing a computer for training (and it will be doubtful that chess will continue).
Interestingly, that will also solve the old conundrum about whether black and white playing perfectly results in a win for white, a draw or (most unlikely) a win for black.
I remember reading a book about game theory just prior to starting my degree, in which they postulated that this question was unanswerable. How times change!
pan
Arbela.
Now you went and made me learn something! I only knew about the latter!
Thanks!
Trinopus
However, at the point where computers know every possible move, then playing against one becomes impossible. Theorectically, one could do the same thing with a huge table on paper, its just that computers make it practical.
I don’t think Chess will die. People still love the Olympics even though no human can win against a car. Or a horse in most races (due to two legs, humans can win against horses for very short distance, and some say this includes the 100 meter dash.)
Who won the American revolution? The British army was the best in the world at that time.
Who is now in charge of Vietnam?
::Xavier sticks up his hand:: Ummmmmmm… that would be the Yankees Ma’am.
::Xavier scratches his head, then sticks up his hand:: Ummmmmmm… that would be the Polish?
Can’t be, I’m ignorant.
So at the point that a computer can actually define every possible move (which I’m presuming at the lack of my relevant computer science knowledge) that is stored in it’s memory, do tactics become rendered useless? What about if a computer plays another computer of equal ability? Who wins the match and what is the significance (if any) of tactics?
So you’re saying the best way to play a computer is to “be” like a computer i.e. brute calculation?
Is this same maxim true of all games of strategy/tactics? That at one point, when a computer has reached a certain level of computational aptitude that all human foes will be rendered “uncompetitive” against these machines?
And if so, is it really still a strategic game?
[What I’m getting at here is whether strategy can be computed or calculated as opposed to thought out.]
Well, think of a simple game like tic-tac-toe that has every move solved - if you play perfectly, both sides will tie every time. It is easy to make an unbeatable computer for tic-tac-toe. Chess is of course vastly more complicated, but once it is solved, we will know for sure whether a perfect game results in a tie, white wins or less likely, black wins. If, for example, the perfect game results in a white win, it would be ** impossible ** to win a against a properly programmed(that is, has the game solved) computer that is playing white, no matter how well you play.
An interesting example is the Russo-Finnish War in 1939. In the end, the Soviets won the war through shear numbers, but it was not a complete victory since the smaller, better-trained, and better-led Finnish troops inflicted great losses on the Soviets. The poor performance by the Soviets also influenced Hitlers decision to invade them later.
I agree with your physical examples above. However the difference with the mental calculation of chess is that every mistake can be instantly exposed.
Who wants to see all their blunders, or even find their opponent is cheating using the perfect database?
There is no relevance to tactics when the entire game can effectively be foreseen. If it turns out that White has a tactic, then Black could have avoided it on his previous move.
I assume that perfect chess will turn out to be a draw, but it’s only based on my experience - I can’t prove it.
Any computer with access to the full tables of all legal chess positions will be unbeatable (provided my opinion above is correct).
As DreadCthulhu said, think of it like tic-tac-toe. Once you have stored every possible position, you’re unbeatable.
(Although I must correct his later statement - ‘If, for example, the perfect game results in a white win, it would be impossible to win against a properly programmed(that is, has the game solved) computer that is playing white, no matter how well you play’ should read ‘it will be impossible to draw’.
Sorry if I didn’t explain very well.
If you are up against a database of all positions (e.g King + 2 bishops v King + knight), the computer will ‘play’ perfectly. It doesn’t do any calculation - it just looks up the move that leads to the best result, because it has already worked all that out.
If you are playing a computer that calculates each move, then it will have a short term ‘horizon’ (say 4 moves ahead by each side) within which you cannot surprise it with anything, tactics included.
No human can cope with that amount of information (about 810,000 positions).
So us carbon-based life-forms must rely on pattern recognition, judgement and strategy.
You need to build up an overwhelming advantage more that 4 moves ahead, so that when the computer finally recognises the danger, it is too late.
Thus you are playing the opposite to the computer, which wouldn’t recognise a strategy even if it got bitten by one!
Well chess is unusual in that all game information is available. You just need to be able to calculate!
If you play poker against a computer, will it be able to bluff? Presumably it will use a random number generator to choose its moment.
I don’t think that referring to a giant database of all possible continuations is ‘playing’, let alone using strategy.
To me, strategy means you decide what to do, based on limited information, using general principles.
Once the forces are in close contact, if your strategy was correct, the tactics should now come your way.
People can cheat using steroids and other banned tricks in sports. No difference there.
To be honest, no one ever broadcast chess matches so as to make money. Poeple play chess because its deep and rewarding and and interesting game. In truth, when dealing with a super-dedicated chess program, you aren’t making mistakes. Its simply that there are no correct moves.