It really depends. Programs are so advanced now, that the expected draw may not be the actual outcome. I have engine tournaments all the time, and sometimes an engine will lose to itself. It can depend on the opening, the hashtable settings, and the hardware.
For instance, Shredder can surely lose to Shredder on the same machine! 
It really depends. Programs are so advanced now, that the expected draw may not be the actual outcome. I have engine tournaments all the time, and sometimes an engine will lose to itself. It can depend on the opening, the hashtable settings, and the hardware.
For instance, Shredder can surely lose to Shredder on the same machine!
very strange
Not so strange. Consider this gross oversimplification:
Two computers are each powerful enough to look 10 moves ahead with great clarity. Computer 1 considers position A and looks 10 moves into the future and determines what it thinks is the best move.
Then it's Computer 2's turn to look 10 moves ahead, but this time from position B (so its visible horizon is now further out than Computer 1's was when it made the move that created position B). Now Computer 2 sees a move that represents a clear & true advantage and makes it. Computer 2 will win this game - even if, after 100 games, the two computers' scores would be about the same.
Disclaimer: A computer scientist, I ain't. ;-)
I was experimenting with lines arising from 1 e4 b5, and had Rybka play it out agains itself. I'm not the most patient guy, the setting was for 5 min Blitz. Black won the majority, lol
I tried it and white won, then black, then white, then white, then black. I expect it depends.
Here's the way to look at it: chess is solvable, which means there is a "correct" move as early as move 1. If you allow each computer to consider it's move for the next billion years, the game will always look the same and the outcome will always be the same, whatever the natural outcome of chess may be.
However, we do not allow our computers to play the way a computer really should. We give it openings and claim there are certain values associated with certain pieces and come up with some formula that suggests some position is more favorable than another. So within the confines of these neccisary imperfections, computers just solve optimization problems that aren't really optimal, and inconsistant outcomes could result from this based on where and how these imperfections of the program are implimented. While that doesn't exactly answer the question originally asked, it sort of does... I think. :-P
If you allow each computer to consider it's move for the next billion years, the game will always look the same and the outcome will always be the same, whatever the natural outcome of chess may be.
I dont agree that the game will always look the same... The result will always be the same, though.
I suspect that with correct play, all games would be drawn.
I don't think that there is a first move which is a forced loss for white, so technically all 1st moves are equal. The vast majority of those probably have multiple replies which aren't forced losses for Black, so those would be equal too... No, there would still be variety
If a computer played the same computer with the same power, what do you think the outcome would be?
The generation of the nexus of the universe and the collapse of the Earth as you know it into a tiny black hole. Small disturbances would be felt through the rest of the Solar system, but only negligible results would persist to the rest of the galaxy while even the closest of our neighboring galaxies won't notice one bit. Don't try this at home.
Fair enough, rooperi. I guess I was just imagining either side consistantly making whichever move provided the most room for error for the other side. Meaning if white has two options, both should be draws, but if option A has 8 refutes that result in a draw and option B has only 1, B will be more favorable in the computer's eyes and played every time. That's getting nit-picky, though... in essence we both agree.
its like stacking bricks. cpu 1 puts one down, cpu 2 puts one over that. cpu puts one over that, etc. but in the end, the bricks will fall and white would win. i tried, it went in a pattern of wins,white, black, white, white, black.
Fair enough, rooperi. I guess I was just imagining either side consistantly making whichever move provided the most room for error for the other side. Meaning if white has two options, both should be draws, but if option A has 8 refutes that result in a draw and option B has only 1, B will be more favorable in the computer's eyes and played every time. That's getting nit-picky, though... in essence we both agree.
I was trying to make a point, (I clearly didn't do that very well) That even if Computers "solve" chess, it would not necessarily kill the game for human to human play.
There would be no such thing as ONE PERFECT GAME which human players could memorise, and having done so could consistently draw against any oppostion.
If a computer played the same computer with the same power, what do you think the outcome would be?