Chess Engine Evaluation
Sort:
In chess engine evaluation, a minimum difference of much score would you consider decisive?
OculorumAcies wrote:
Why the differences?
I would say 2 in the opening, 4 in the midgame, 3 in the endgame.
Mar 17, 2017
0
#4
It depends on the position.
First of all there are practical considerations. If both players are confused in a messy position, then anything could happen no matter the engine evaluation.
There are also endgame draws and fortresses, which engines don't understand. They might say +2.00 in an endgame that's obviously drawn.
But if you let the engine play itself, and think a long time on each move, then an advantage close to 1.00 would win almost every time. And an advantage around 0.5 to 0.75 would win sometimes.
Mar 17, 2017
0
#5
I think +4 in the opening would itself be very devastating. People don't usually make mistakes that would lose the advantage. Midgame +2 happens time to time but I've seen some flips. It is usually +3 in the end game that makes it spike up to something silly like +60,25 or so. I am sure there are players who still made it to a draw or a win in that adversary, but I would still call it decisive.
What if the evaluation was made by say the 'God of Chess'?
Mar 17, 2017
0
#7
If everything else is equal, then losing a pawn is enough to lose the game in most positions. Players will keep playing in case their opponent makes a mistake, but a pawn down is a lot, and an engine would only show +1.
In most positions +4 is enormous, and players are already resigning right away.
Mar 17, 2017
0
#8
Act64 wrote:
What if the evaluation was made by say the 'God of Chess'?
"God of Chess" only makes 3 different evaluations:
White wins
Draw
Black wins
Mar 17, 2017
0
#9
In that case the evaluation could only be win, loss, or draw. None of this decimal stuff.
The_Chin_Of_Quinn wrote:
A pawn down may not be decisive all the time, it also depends on the position, sometimes the revelation comes only many moves later.
If everything else is equal, then losing a pawn is enough to lose the game in most positions. Players will keep playing in case their opponent makes a mistake, but a pawn down is a lot, and an engine would only show +1.
In most positions +4 is enormous, and players are already resigning right away.
Mar 17, 2017
0
#11
If everything else is equal, then a pawn down is a lot.
But in most positions everything else is not equal 
There are also positions with equal material that are totally lost for one player. I'm just giving an example of something an engine would show as +1
Mar 17, 2017
0
#12
5.12. "Whenever Rybka evaluates a position with a score of +/– 5.12 we don't need to search any further, we have our proof that in the continuation there is going to be a win or loss, and there is a forced mate somewhere deep down in the tree. We tested a random sampling of positions of varying levels of difficulty that were evaluated at above 5.12, and we never saw a solution fail. So it is safe to use this assumption generally in the search." http://en.chessbase.com/post/rajlich-busting-the-king-s-gambit-this-time-for-sure
Mar 18, 2017
0
#13
Platzerwasel wrote:
5.12. "Whenever Rybka evaluates a position with a score of +/– 5.12 we don't need to search any further, we have our proof that in the continuation there is going to be a win or loss, and there is a forced mate somewhere deep down in the tree. We tested a random sampling of positions of varying levels of difficulty that were evaluated at above 5.12, and we never saw a solution fail. So it is safe to use this assumption generally in the search." http://en.chessbase.com/post/rajlich-busting-the-king-s-gambit-this-time-for-sure
OP should note that engines other than Rybka will be different in that regard.