Komodo and WCCC

[Lyudmil Tsvetkov]

You still didn't grasp most of it. There can be dozens of artists from the last 100 years who are just not 'discovered' yet and still might sell their works for more than any van Gogh in future.
(if you still insist on your weird 'money = value of art') Good art is completely independent from what you think.
Anyhow my last remark, because this leads to nothing...

unfortunately, there are not.

shovelfuls of talent are not that easy to come by.

[mjlef]

I am in a lot of agreement with what you have written. A small disagreement in what to call what we do. Although I am an engineer, and proud of if, I still think the stuff Larry and I get to do each day is pretty close to pure science much of the time. We hypothesize about a new search or evaluation term. We test that hypothesis by running many thousands of games, and we use statistics to guide it in what ideas to keep or toss. There is some "engineering" like coming up with better data structure to speed up a move generator and such.

What is the code difference between K and SF?
What is the fundamental code difference between K and SF?

A complete answer would take a lot of time to put to paper. Komodo's evaluation is quite a bit larger (and slower). Both programs use standard chess programming things like nullmove, LMR, futility pruning/static nullmove pruning. They differ in a lot of details of exactly when and how things like LMR and nullmove are implemented. Although both now use a form of "Lazy SMP", Komodo adds a bunch of other things for MP searching. Endgame evaluation differs a lot. Komodo has a more extensive material imbalance system. Where we allow pruning from hash entries differs. Hash replacement scheme differs. Low depth pruning differs although has ideas around the same concepts. Stockfish seems to reduce more in nullmove than us. We have more extension rules. We reduce more in LMR. But we have a lot of conditions on these things. Of course both programs are in constant flux as both teams try to improve them.

[Rodolfo Leoni]

I'm very happy for the contents of this discussion (except for some spammings). I got a more accurate perception of the whole situation, and I concluded ICGA events must not die. Net tournaments are the best way to check engines strenght, but there's nothing like an event in real world. The fact programmers meet, know each other better, share their ideas and opinions is the best of these events. That couldn't happen on net events. Mark is aware of it, while Fish team keeps closed in its ivory tower. The lack of any comment sounds like an implicit "no comment". The "Borg collectivity" would at least say "Irrelevant!"...

Those who think they only need Stockfish don't understand Stockfish needs strong "competitors" as reference point for their progresses. The fact of having different and original engines is a way to enrich so everybody need a strong Komodo, Shredder, not only Stockfish.

BTW, I'm curious to know....

1. Does Fish team need Tord's permission in order to participate to WCCC?
2. Does a Fish fork need official Stockfish team's permission as well?

It'd be funny to see some asmFish or similar in the event...

[hgm]

Stockfish carries a GPL licence, which explicitly allows anyone to use it in any way he likes. The copy-right holder is barred from attaching any conditions to the use of his code. So yes, you do need permission from all contributers to Stockfish to use it (for entering a tournament or otherwise), but they have pro-actively granted this permission through the license agreement.

Of course this is just the legal aspect; to enter in the WCCC you also have to satisfy the conditions the ICGA demands. And this could be anything. Including that Tord should sign a letter expressing his desire for Stockfish to participate. But that is completely up to the ICGA, and I would advice against it.

[Rodolfo Leoni]

Stockfish carries a GPL licence, which explicitly allows anyone to use it in any way he likes. The copy-right holder is barred from attaching any conditions to the use of his code. So yes, you do need permission from all contributers to Stockfish to use it (for entering a tournament or otherwise), but they have pro-actively granted this permission through the license agreement.

Of course this is just the legal aspect; to enter in the WCCC you also have to satisfy the conditions the ICGA demands. And this could be anything. Including that Tord should sign a letter expressing his desire for Stockfish to participate. But that is completely up to the ICGA, and I would advice against it.

Thanks!