@dboing it's the self-play elo, new versions of stockfish aren't outpreforming older ones. It is the biggest plateau stockfish has ever had. It might not be perfect chess but near perfect with current hardware and the current architecture.
This is like the scene in American Gangster where Denzel confronted one of his suppliers about selling watered down Blue Smoke.
>To prevent future violations, ChessBase will create the position of a Free Software Compliance Officer
So this is the undercover agent going for some real compensation? Jk
Imo SF team got all they wanted and the reputation damage to CB, who went money grabbing with rip-offs, is huge.
I don’t know enough about software tbh, but the one year timeout seems like a big deal when you clearly don’t have the skills assembled to come up with original stuff on your own.
In a fast paced landscape like software engineering and marketing that might even be enough time for a competitor to take over, especially when the trust in your brand is completely destroyed by exposing your lies.
Who’s gonna buy the next edition of Fat Fritz now?
So this is the undercover agent going for some real compensation? Jk
Imo SF team got all they wanted and the reputation damage to CB, who went money grabbing with rip-offs, is huge.
I don’t know enough about software tbh, but the one year timeout seems like a big deal when you clearly don’t have the skills assembled to come up with original stuff on your own.
In a fast paced landscape like software engineering and marketing that might even be enough time for a competitor to take over, especially when the trust in your brand is completely destroyed by exposing your lies.
Who’s gonna buy the next edition of Fat Fritz now?
I mean, Chessbase did profited from Fat Fritz 2 and Houdini 6. As what @Sarg0n said, "Two winners or two losers?". Probably two winners.
@justaz said in #41:
> @dboing it's the self-play elo, new versions of stockfish aren't outpreforming older ones. It is the biggest plateau stockfish has ever had. It might not be perfect chess but near perfect with current hardware and the current architecture.
Yes, I was understanding that ELO too.. Thanks i did not have the full perspective of the trajectory across versions.
But there is no way to measure how close to perfect such measure bring successive versions of SF.. It is still self reffering.
Only by having new "blood" looking at blind spots/spaces of SF in tournament competitions like A0-Lc0 did between generation SF8 and SF11 (12 patching the holes, and up to now optimizing more of the same from then), would we know that we were not that close.. but it always depend on changing the pool of competitors...
With LC0 waiting for its costly design to become better implemented in new hardware, or tournament specification to become more useful for human chess needs, not just mimicking human tournaments, which have way more diverse "hardware", by way of human individual personal trajectories being multiple, and not cloned... This is not going to happen again. Unless somebody wakes up to the question of what else could we let engine compete about than speed of computation... ELO could be about more controlled things that would allows improvements specific to accuracy independent of speed. separate the speed component from the accuracy in the ELO.. I think the block is not even wanting to ask the question of why do we still have engine competition at all..
What were the ELO kinetics before NNue?
Perhaps time may not be a good measure for ELO kinetics.. versions? code changed quantity.. Dev hours spent... Or perhaps some characterization of tournament competition (including SF dev pool constituents, as you describe) over time..
In any case, thanks for placing your post in that light, that i might be the worse ELO climb slowing in some recent history.
That gives some referential. We lack those in engines accuracy analyses (not done, I think, assuming ELO tells the whole chess story).
> @dboing it's the self-play elo, new versions of stockfish aren't outpreforming older ones. It is the biggest plateau stockfish has ever had. It might not be perfect chess but near perfect with current hardware and the current architecture.
Yes, I was understanding that ELO too.. Thanks i did not have the full perspective of the trajectory across versions.
But there is no way to measure how close to perfect such measure bring successive versions of SF.. It is still self reffering.
Only by having new "blood" looking at blind spots/spaces of SF in tournament competitions like A0-Lc0 did between generation SF8 and SF11 (12 patching the holes, and up to now optimizing more of the same from then), would we know that we were not that close.. but it always depend on changing the pool of competitors...
With LC0 waiting for its costly design to become better implemented in new hardware, or tournament specification to become more useful for human chess needs, not just mimicking human tournaments, which have way more diverse "hardware", by way of human individual personal trajectories being multiple, and not cloned... This is not going to happen again. Unless somebody wakes up to the question of what else could we let engine compete about than speed of computation... ELO could be about more controlled things that would allows improvements specific to accuracy independent of speed. separate the speed component from the accuracy in the ELO.. I think the block is not even wanting to ask the question of why do we still have engine competition at all..
What were the ELO kinetics before NNue?
Perhaps time may not be a good measure for ELO kinetics.. versions? code changed quantity.. Dev hours spent... Or perhaps some characterization of tournament competition (including SF dev pool constituents, as you describe) over time..
In any case, thanks for placing your post in that light, that i might be the worse ELO climb slowing in some recent history.
That gives some referential. We lack those in engines accuracy analyses (not done, I think, assuming ELO tells the whole chess story).
@sangisangy said in #8:
> should be 3 years at least, 5 years seems fine
@cherouvim said in #9:
> Anyone knows why is the FSFE that will get some money instead of the Stockfish team?
@A_0123456 said in #17:
> What is SCID
> should be 3 years at least, 5 years seems fine
@cherouvim said in #9:
> Anyone knows why is the FSFE that will get some money instead of the Stockfish team?
@A_0123456 said in #17:
> What is SCID
What is SCID in plain English?
Ok...
Thank.s alot every one
shane.s chess information database
This topic has been archived and can no longer be replied to.