The quality of the book seems to apply mostly to the top levels when there is little difference in playing ability otherwise.
So to me (and this is subjective) the question for any game is whether you have to spend the vast majority of your time on opening preparation if you are at the top levels. In Chess and checkers you have to spent the vast majority of your time on opening prep. In 9x9 I suspect it will be something similar to chess, but I don't really know. It's very difficult to quantify which of various things is more important because they interact in such complex ways - but to me this single factor kind of spoils the game - you don't want your skill to be primarily based on how much you can memorize. Don On Sat, Jul 17, 2010 at 7:51 AM, <[email protected]> wrote: > Quoting Martin Mueller <[email protected]>: > > I beg to disagree. I think the book is just one of four or five >> components, which are all equally important. >> >> - search >> - simulations >> - hardware >> - general Go knowledge >> - book (9x9 only) >> >> If you are lacking in one of the first four, a book won't save you. All a >> book does is help you survive the opening and achieve a non-inferior >> position from where you can start fighting. Playing without a book against >> a well-prepared opponent, you may lose without a real fight. However, see >> the performance of Many Faces which does quite well on 9x9 even without a >> book. >> > > I agree with your general comments completely. But I think the top programs > will need to be a little more diversified. Looking at CGOS games there are > many positions where for example Fuego played Manyfaces many times (and > almost no other programs) with a completely predictable result. Proabably > because Fuego follows some lines very strict and Manyfaces playing almost > deterministic in many cases. > > The point being is that a good book should be hard to exploit and lead to > postions where the other factors determine the outcome. Without a book it > might be able to exploit a single program weaknesses. Excessive exploitation > however might lead exploatation from a 3rd program which mimics the 1st > program but plays stronger moves later. > > Here is a thought of what could become more diversified. Currently all top > program only play the center point E5 as black on the first move. Therefore > E5 looks really good in the statistics > > E5 53.6 65066 > F5 47.8 25949 > F4 46.1 15535 > G5 41.6 4696 > G4 40.8 14549 > G3 38.3 3406 > > But maybe this is just a self fulfilling prophecy? Maybe F5 and F4 is also > playable when some good lines are chosen as black. There seems to be some > lines where a fair fight is possible but the number of games is very few so > it might be just random anomalies in the statistics. No doubt playing F5 and > F4 is much harder than E5. > > Best > Magnus > > > _______________________________________________ > Computer-go mailing list > [email protected] > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go >
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
