Interesting is in the eye of the beholder, but I think a Hahn tournment would be interesting.
Everyone with an MC bot could easily modify it to play by setting it (carefully) not to resign, and changing it to try to maximize average score rather than probability of win. That's really just a few lines of code. (It might not hurt to look at passing too.) Then we (I'll try to join) would face the choice of whether or not to retune the engine-rather a tiresome chore. If everyone did that, I expect that we would see roughly the same stratification of the field that already exists. (But you never know). I picture the strong bots making safe moves and waiting for the weaker ones to make mistakes. What I hope would happen is that some of the people, experimenting with dynamic komi and the like, would combine it with some opponent modeling and be creative. When I take time out to concentrate on a related problem, I usually get new ideas that help me with the original problem. A couple nagging worries: 1) Taking the average score might be pretty noisy. Some luck in a tournament is good, too much is bad. Some preliminary simulations with a strongish 19x19 bot, might give some indication of whether a double round robin tournament would be adequate. 2) Also some games would be incomplete due to inadvertent resignation, bot crashes, or loss on time. Nick would need to choose a policy for scoring in those cases . The tournament could easily be decided by how an incomplete game gets scored, and that would be a bit frustrating. - Dave Hillis -----Original Message----- From: "Ingo Althöfer" <[email protected]> To: [email protected] Sent: Mon, Jan 4, 2010 1:51 pm Subject: [computer-go] Re: (no subject) wish hahn Hi all, Alain Baeckeroot wrote: As a physicist i like to experiment first, and think later, to understand what happened, which obviously was not foreseen ;-) This attitude I like very much, being an experimental mathematician. > I believe it will reveal some hidden aspect of the stronger engines, and for sure it will be *fun* to see stronger bots destroy weaker ones, (MC-RAVE end-game of strong bots is boring on 19x19) ... No need to be perfect Hahn. Changing the aim to max_points seems to be just a change in metrics, it does not looks like very complicate o with Hahn scoring looks similarily exotic like Pair-Bot-Go, nd I understand the hesitation of all who live in performance-oriented ategories. Perhaps it would help some people when such exotic tournaments Hahn scoring, Pair-Go, ...) would be very clearly in a spirit where inning and prizes does not count more than anything else. For this relaxed spirit" it might help when the bots did not enter with their raiditional (fame-burdened) names, but with related names like en-Gaga, Mogo-Gaga, ManyFaces-Gaga or whatever. or Gazenga, Gomogoga, gamanyfacesga ...) I am already dreaming of a pair-bot-tournament with continental teams: ManyFaces & Fuego for North America Zen & Aya for Asia Mogo & Valkyria for Europe - Dreamingo. -- reisknaller: GMX DSL Flatrate für nur 16,99 Euro/mtl.! ttp://portal.gmx.net/de/go/dsl02 ______________________________________________ omputer-go mailing list [email protected] ttp://www.computer-go.org/mailman/listinfo/computer-go/ =
_______________________________________________ computer-go mailing list [email protected] http://www.computer-go.org/mailman/listinfo/computer-go/
