What is your chance of winning on time against an engine?

Sort:
LabMaster100Hat

It is 0.1%/100%. Engines work like that: they have some games in their database. They see your move and take that move to their database. They search for games with that move (there can be 11111 or more) and analyze the games for the best reply, and then move like the best reply. But if you get them in a problematic position, they will analyze your game that you were playing, and use the best move. There is no luck. Only events. Event no.1: You move e.g. Nf6.

Event no.2: Nf6 has been played in e.g. 475 576 games. Event no.1 and Event no.2 collide.

The engine saw your Nf6 move and is analyzing those 475 576 games for the best reply. The normal speed of an engine is 0.5-3 seconds. The engine has to analyze all of them. It can sometimes do it 1 minute, or even 5.

Martin_Stahl
LabMaster100Hat wrote:

It is 0.1%/100%. Engines work like that: they have some games in their database. They see your move and take that move to their database. They search for games with that move (there can be 11111 or more) and analyze the games for the best reply, and then move like the best reply. But if you get them in a problematic position, they will analyze your game that you were playing, and use the best move. There is no luck. Only events. Event no.1: You move e.g. Nf6.

Event no.2: Nf6 has been played in e.g. 475 576 games. Event no.1 and Event no.2 collide.

The engine saw your Nf6 move and is analyzing those 475 576 games for the best reply. The normal speed of an engine is 0.5-3 seconds. The engine has to analyze all of them. It can sometimes do it 1 minute, or even 5.

 

Most engines don't work like that. If the have an opening book and are configured to use it, they will look at that for suggested moves, but otherwise, classical engines are brute force checking moves and evaluating the position, then going one ply deeper and repeating, pruning off useless lines as they go. They look at thousands of possibilities per move.

 

Newer engines using NNUE, look at positions and look at moves available to see if they've previously seen a similar position and what that position's winning chances are. It will end up choosing to go into a position with the highest winning chances.

 

Some engines combine the two ideas. 

LabMaster100Hat
Martin_Stahl wrote:
LabMaster100Hat wrote:

It is 0.1%/100%. Engines work like that: they have some games in their database. They see your move and take that move to their database. They search for games with that move (there can be 11111 or more) and analyze the games for the best reply, and then move like the best reply. But if you get them in a problematic position, they will analyze your game that you were playing, and use the best move. There is no luck. Only events. Event no.1: You move e.g. Nf6.

Event no.2: Nf6 has been played in e.g. 475 576 games. Event no.1 and Event no.2 collide.

The engine saw your Nf6 move and is analyzing those 475 576 games for the best reply. The normal speed of an engine is 0.5-3 seconds. The engine has to analyze all of them. It can sometimes do it 1 minute, or even 5.

 

Most engines don't work like that. If the have an opening book and are configured to use it, they will look at that for suggested moves, but otherwise, classical engines are brute force checking moves and evaluating the position, then going one ply deeper and repeating, pruning off useless lines as they go. They look at thousands of possibilities per move.

 

Newer engines using NNUE, look at positions and look at moves available to see if they've previously seen a similar position and what that position's winning chances are. It will end up choosing to go into a position with the highest winning chances.

 

Some engines combine the two ideas. 

Yes.