Suggestions for new CCT rules

[garybelton]

Good points, is it possible to do learning with a book limit? So for instance start learning on move 7?

Or make a rule to have no learning just in CCT? I'm not sure learning is such a problem in only 16 rounds - maybe it is?

[mathmoi]

Such a rule would be impossible to enforce. You can study the public games of your opponent if its an engine by watching games on Playchess, ICC, FICS, SSDF, CCRL, CEGT etc..

I totally agree. It's not enforceable.

[Harvey Williamson]

Good points, is it possible to do learning with a book limit? So for instance start learning on move 7?

Or make a rule to have no learning just in CCT? I'm not sure learning is such a problem in only 16 rounds - maybe it is?

A book limit could be a good idea. How do you enforce a no learning rule? If you have programmed a good book learning system why shouldn't you use it?

[Togga]

interesting strategy your engine is using for opening. what is the name of your chess engine?

[garybelton]

That's not how it used to be, but I think that is good, I'm sure the engine room players wouldn't be happy if lots of people had easy access to the fruits of their labor - and utility bills

[garybelton]

Let's be clear, it's not enforceable to use what lines (Rybka generated or otherwise) go into the book, but it is possible to enforce a book limit, thus limiting "the damage", right?

[mathmoi]

Good points, is it possible to do learning with a book limit? So for instance start learning on move 7?

Or make a rule to have no learning just in CCT? I'm not sure learning is such a problem in only 16 rounds - maybe it is?

Playing CCT whith learning disabled would not be a problem, however I'd like to be able to use what I previously learned. From move one to endgame.

The real problem is some author put a lots of effort in generating their book and are not using other engines to generate them and should not be penalized. However, as I said elsewhere, I don't think we can really enforce this.

Even if we disallow the use of opening book after, say, move 10. How can you make sure I'm not effectively using a book move. I can easily chose a book move, ponder it for a minute to generate a PV/node count/evaluation to kibitz and play the book move. I won't save as much time as if I had played instantly, but I still get to play a book move.

[mathmoi]

Good points, is it possible to do learning with a book limit? So for instance start learning on move 7?

Or make a rule to have no learning just in CCT? I'm not sure learning is such a problem in only 16 rounds - maybe it is?

A book limit could be a good idea. How do you enforce a no learning rule? If you have programmed a good book learning system why shouldn't you use it?

And if you programmed a good book learning system why shouldn't you use it past a certain depth?

[garybelton]

Good to hear that you might be favorable to book limiting. I agree with learning being an ok thing, but I was asking when does the book learning kick in? I recall Bob saying something like Crafty started with the last move played from book and went backwards, but if we artificially limited the book depth would it still work? I have no problem with book learning.

[garybelton]

Even if we disallow the use of opening book after, say, move 10. How can you make sure I'm not effectively using a book move. I can easily chose a book move, ponder it for a minute to generate a PV/node count/evaluation to kibitz and play the book move. I won't save as much time as if I had played instantly, but I still get to play a book move.

Sneaky. You could rule that people should not do this. For public domain engines it should not be a problem, for private versions it could be. There are some moves (take B97 PP) where engines cannot find the moves themselves in CCT times so it would be obvious what was happening, then the author would be risking being banned, possibly forever, a risky strategy.