Watch "ClutchPoints - "Basketball, I'm good at. But that's not even the best thing I'm ever good at... The greatest domino player in the world... Is myself." Jimmy Butler says basketball isn't his greatest skill 😂 (via @netflix "Starting 5")" on Streamable.
On average, there are about 30 legal moves per position and about 120 total moves per game. That means there are about 180000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 possible game branches in the average game alone. It’s simply too big to calculate.
For comparison, chess with 7 pieces left on the board is solved. The calculations to add 1 more piece increase the solution storage size from 20000 GB to 2000000 GB, would require ~40000 GB of RAM, and would take an estimated 8 months to calculate on the world’s fastest supercomputer. That’s just going from 7 to 8, chess has 32 pieces and solving 1 more piece increases the calculation needed exponentially.
It hasn’t happened yet. It’s certainly possible from a technical standpoint, infinite processing power and/or time would do the trick. Chess has so many permutations that it will take a lot of processing power/time and we just aren’t there yet. Current AIs don’t try to solve chess at all, they train to play it much like humans but with way more reps.
It’s solved for all endgames with 7 or fewer total pieces on the board out of the original 32. In theory, with a powerful enough computer, you can continue to work backwards and eventually solve chess completely, but modern computers are nowhere close to powerful enough to do that.
Solving a game similar to chess means finding route(s) where a player can always force a victory (or a draw, in the case of chess).
With this concept in mind, it’s easy to see why it’s hard to solve. The number of permutations of positions after the first few moves is gigantic.
We do have solutions if there are few pieces in the board (up to seven). But every piece added requires more computing power and more storage to memorize all possible variations.
It could be solved eventually, but we are not there yet.
There is an article on wikipedia about solving chess.
Pretty much. Engine play from the starting position is a draw 99.9% of the time. What the other guy who responded to you is talking about are computer matches where the engines play game pairs that start in uneven positions. This is much more common as it lets people see which engine is better.
For example, engine A and engine B play a game pair where white starts with a better position. When A plays with white, A wins the game. When B plays white (from the same uneven starting position), the game is a draw. This means that A won the game pair 1.5-0.5. If A played B from the starting position in 2 games, the result would almost certainly be 2 draws, but playing chosen game pairs will let the better engine have a chance at winning one and drawing the other.
why is the sky the limit in chess? why can’t it be solved. Surely a computer has figured it out.
On average, there are about 30 legal moves per position and about 120 total moves per game. That means there are about 180000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 possible game branches in the average game alone. It’s simply too big to calculate.
For comparison, chess with 7 pieces left on the board is solved. The calculations to add 1 more piece increase the solution storage size from 20000 GB to 2000000 GB, would require ~40000 GB of RAM, and would take an estimated 8 months to calculate on the world’s fastest supercomputer. That’s just going from 7 to 8, chess has 32 pieces and solving 1 more piece increases the calculation needed exponentially.
I believe the average game is some like 40 moves. 120 is far too large an estimate
It hasn’t happened yet. It’s certainly possible from a technical standpoint, infinite processing power and/or time would do the trick. Chess has so many permutations that it will take a lot of processing power/time and we just aren’t there yet. Current AIs don’t try to solve chess at all, they train to play it much like humans but with way more reps.
It’s solved for all endgames with 7 or fewer total pieces on the board out of the original 32. In theory, with a powerful enough computer, you can continue to work backwards and eventually solve chess completely, but modern computers are nowhere close to powerful enough to do that.
Solving a game similar to chess means finding route(s) where a player can always force a victory (or a draw, in the case of chess).
With this concept in mind, it’s easy to see why it’s hard to solve. The number of permutations of positions after the first few moves is gigantic.
We do have solutions if there are few pieces in the board (up to seven). But every piece added requires more computing power and more storage to memorize all possible variations.
It could be solved eventually, but we are not there yet.
There is an article on wikipedia about solving chess.
I would assume that, if perfectly simulated, two computers would end every game in a stalemate, no?
Pretty much. Engine play from the starting position is a draw 99.9% of the time. What the other guy who responded to you is talking about are computer matches where the engines play game pairs that start in uneven positions. This is much more common as it lets people see which engine is better.
For example, engine A and engine B play a game pair where white starts with a better position. When A plays with white, A wins the game. When B plays white (from the same uneven starting position), the game is a draw. This means that A won the game pair 1.5-0.5. If A played B from the starting position in 2 games, the result would almost certainly be 2 draws, but playing chosen game pairs will let the better engine have a chance at winning one and drawing the other.