Playing Bright 0.5c I have detected -or it is just a mirage?- that the engine evaluate somewhat optimistically the positions, in any case more optimistically -same position- than any other engine in my knowledge. I wonder if this slight inclination of the judgement towards the positive side does not have some influence for bettering the already incredible strong game Bright gives.
It is clear that a bias like that would be very harming to a mediocre engine. IF it is just average, a positive inclination will tend to put it in even deeper trouble. BUT if it is very strong, could not happens that the positive bias make the engine prone to search even deeper lines that otherwise could be replaced by sub optimal ones?
Just a question for my programmers pals here.
Fern
Unbalanced Scores in Evaluation
Moderator: Ras
