Re pondering

[kranium]

i've never tested ponder on/ponder off, but conceptually it seems clear:

hypothetical:

lets say you match an engine against itself at a long TC game (perhaps 40/120), ....but one side has ponder on, and the other ponder off (all other variables are the same).

it seems clear to me that the side with ponder on gets 100% more time (and 100% more CPU usage) to analyze the game than his opponent...

if the engine with ponder on then achieves a reasonable ponder hit (again hypothetically, for ease lets say 50%), then realistically this means it will spend 50% more time (and CPU) than its opponent, -> analyzing relevant positions. (in the 40/120 game mentioned above, this would result in 60 extra minutes every 40 moves).

wouldn't this (shouldn't this) be a great advantage...?

[Matthias Gemuh]

....but one side has ponder on, and the other ponder off (all other variables are the same).

Pondering is of course a clear advantage if the opponent is not allowed to ponder. Nobody disputes that. An elo gain of about 30 elo is then possible.
I noticed that in WBEC when my engine learned to ponder.

Matthias.

[Guenther]

I am the original poster of this topic and am sorry i have started a feud! I only wanted to know how to obtain the best quality chess possible. ponder on or off simple.

I guess the question is not related to 1 minute bullet chess, because in
this case it would be absurd to care about 'best quality chess' anyhow?

Guenther

[hgm]

Why do you say that? I always do my piece-value measurements at 40/1, to get enough statistics in the available time, but I still like to get the best possible chess that the available time allows If I could do with a lower quality, I would shorten the TC, to get more games, rather than switch ponder on, to get the same number of games in the same time, with that lower quality.

Everything is relative, and todays bullet is last decade's correspondence Chess for computers.

[Guenther]

Why do you say that? I always do my piece-value measurements at 40/1, to get enough statistics in the available time, but I still like to get the best possible chess that the available time allows If I could do with a lower quality, I would shorten the TC, to get more games, rather than switch ponder on, to get the same number of games in the same time, with that lower quality.

Everything is relative, and todays bullet is last decade's correspondence Chess for computers.

Best quality chess is best quality chess I don't know how you interpret the
original message? Best possible chess for certain circumstances is something different ...
It is far more important first to care for a proper setup before one cares
for the difference between ponder on/off, especially if a person asks who has a single cpu machine.

Guenther

[hgm]

Then 'best-quality Chess' is the same as 'the largest integer'. It does not exist...

[ernest]

I would shorten the TC...

TC is acronym for what? Thanks!

[hgm]

Time Control