Please explain for other games too if possible.
In: Technology
Sometimes, particularly in the opening book phase of the game. Most of the time, the chess difficulty is controlled by varying the size of the move search tree considered by the algorithm. In easy mode it might look at all its moves and all of your moves, a search depth of 2 plys. In medium mode it might look 5-6 plys deep in the best half of the moves. In expert mode it might look 8-10 plys deep at all of the moves.
A chess algorithm simplified has two parts.
a ) An opening book that lists off common moves.
b ) A search algorithm that looks X moves ahead.
In order to increase or decrease the difficulty you can either limit the number of moves the computer can take from the opening book, or make X smaller so that it doesn’t look as far ahead.
In addition to the other answers, you stop the search at a certain level. You then have to assign some value to the possible outcomes. A chess program can then, based on the selected difficulty choose to do the best move it found or a worse one.
Ok, to put this in ELI5 terms:
A chess program basically works out the best possible move each turn by working out possibilities.
When you make a move, it will look at all the possible moves it can make in return. Then it will ‘simulate’ all those moves, and then simulate all the ways you could respond to all *those* moves… then simulate all the ways it could respond to *that…*repeat.
Basically, each move the game will simulate every way the game *could* go for so many moves ahead, then pick the option that gives it the best chance of winning.
The thing is, there are an almost infinite number of ways a chess game can go, so chess programs will only look so many moves ahead and only plan ahead for the most likely moves you’ll make (the ones that will give you the best chance of winning)… so the most common way to adjust the difficult of a chess game is to limit the number of moves it can work out ahead of time.
So, if a chess game has already worked out every possible move you could make for the next ten turns, it will be really difficult to beat. It knows everything you could possible do in the next ten turns and has worked out the best way to block you… but a chess game that is only looking two or three moves ahead, and not for every possible move will be much easier.
Some AIs in chess will look at all possible moves and give them a score based on how good that move is. A high level AI will always take the highest value moves, but a easier AI might randomly take lower value moves.
They give the computer less time (less calculations) to think about its move.
On each turn there are dozens of moves the computer could make. Then dozens of moves you could make in response, and then again dozens of moves the computer could make in response to your response.
This can quickly turn into hundreds of thousands of potential moves. Reducing the amount of time or number of calculations, limits the chance of the computer picking a really great move.
Latest Answers