Zenmastur vs Ovyron No it's not another GROB!

[jp]

Of course, one shouldn't ascribe too much significance to small differences in evaluations.

Yes... except Larry K. thinks you should.

[Ovyron]

25 cp is a HUGE difference in evaluation 4 moves like that in a game and you lose. Sometimes 1 is enough.

[zullil]

25 cp is a HUGE difference in evaluation 4 moves like that in a game and you lose. Sometimes 1 is enough.

I'm fairly certain that here, at move 5, a difference of 25 cp at depth 68 is "noise". Obviously, Zenmastur's 5. h4 and the move my Stockfish showed both draw.

[Zenmastur]

25 cp is a HUGE difference in evaluation 4 moves like that in a game and you lose. Sometimes 1 is enough.

I'm fairly certain that here, at move 5, a difference of 25 cp at depth 68 is "noise". Obviously, Zenmastur's 5. h4 and the move my Stockfish showed both draw.

Your evaluation at depth 68 is probably still showing a negative. Mine isn't so there is the possibility that the difference between the two is greater than 25 cp and isn't noise.

Regards,

Zenmastur

[zullil]

25 cp is a HUGE difference in evaluation 4 moves like that in a game and you lose. Sometimes 1 is enough.

I'm fairly certain that here, at move 5, a difference of 25 cp at depth 68 is "noise". Obviously, Zenmastur's 5. h4 and the move my Stockfish showed both draw.

Your evaluation at depth 68 is probably still showing a negative. Mine isn't so there is the possibility that the difference between the two is greater than 25 cp and isn't noise.

Regards,

Zenmastur

If you believe that a positive evaluation means something different than a negative one at this point, well, then we see things very differently.

Yes, Stockfish-dev at depth 68 currently shows -0.36. Meaning it's 0.00 + noise.

[Ovyron]

If you believe that a positive evaluation means something different than a negative one at this point, well, then we see things very differently.

But Zenmastur must be right, otherwise 1.f3 is 0.00 so it would be the same as 1.e4. Except 1.e4 is much better, so evaluations are critical, specially if you have to choose a positive one as white instead of a negative one, instead of thinking "it's all the same in the end."

[jp]

It's not all the same, but whether SF can tell the differences right at the start, with 31 pieces on the board, is the question.

And comparing small SF eval differences between moves analyzed at different depths could be very dubious.

[zullil]

If you believe that a positive evaluation means something different than a negative one at this point, well, then we see things very differently.

But Zenmastur must be right, otherwise 1.f3 is 0.00 so it would be the same as 1.e4. Except 1.e4 is much better, so evaluations are critical, specially if you have to choose a positive one as white instead of a negative one, instead of thinking "it's all the same in the end."

We've had this discussion too many times before. Believe what you want.

My only hope is, that for the sake of the audience, Zenmastur is optimizing to stress your limited hardware resources, rather than optimizing very noisy and imperfect Stockfish evaluations.

I should add that Stockfish-dev here is searching with Contempt = 0 and Analysis Contempt = off.

[Ovyron]

My only hope is, that for the sake of the audience, Zenmastur is optimizing to stress your limited hardware resources

Me too, so far this has been my easiest game against 1.e4 of the last 12 months, black's moves can be found in 10 minutes or less (in slow hardware!) so, plagiarizing words I've read recently, I hope Zenmastur has a fairly good sized rabbit up his sleeve.

(the rabbit refers to a position with many moves with similar scores where only one of them saves the game but my hardware doesn't allow me to go beyond their event horizon so I pick one of the bad moves and lose it. Or something like that. You'd want to use a fudged eval that increases the scores of the positions that lead to such scenarios, and this could well end with scores much more accurate than what Stockfish shows)

[Zenmastur]

My only hope is, that for the sake of the audience, Zenmastur is optimizing to stress your limited hardware resources

Me too, so far this has been my easiest game against 1.e4 of the last 12 months, black's moves can be found in 10 minutes or less (in slow hardware!) so, plagiarizing words I've read recently, I hope Zenmastur has a fairly good sized rabbit up his sleeve.

(the rabbit refers to a position with many moves with similar scores where only one of them saves the game but my hardware doesn't allow me to go beyond their event horizon so I pick one of the bad moves and lose it. Or something like that. You'd want to use a fudged eval that increases the scores of the positions that lead to such scenarios, and this could well end with scores much more accurate than what Stockfish shows)

Does that mean you have a move to play?