I agree completely, and I only mentioned that to demonstrate the kinds of walls being put up by the (I believe) previous head of the AGA rating system. His proposal would require a full "many games against several rated players" at each code change, and I am sure that everybody on this list would find that anywhere from silly to bizarre.
But if there is in fact a new person in charge of the ratings system, we might not be up against that. Cheers, David On 9, Aug 2011, at 5:01 PM, Jason House wrote: > The "exactly the same code" requirement is kind of silly. People get better > and worse over time, as do bots with changing code. The only real issue I see > is if a bot enters with a serious bug that then gets fixed. On 9, Aug 2011, at 5:01 PM, Dave Dyer wrote: > >> "exactly the same code that was ranked" > > This is not a reasonable requirement. > > Ratings are a crude tool; it's not worthwhile to try to nail them > down beyond a rough range. Certainly there will be occasions where > programs true ratings slip due to "improvements" in the software, > or improvements in their opponents' understanding of their weaknesses. > > Humans get "code" upgrades and downgrades all the time, it's called learning. > You just take it on faith that the human is still more or less the same from > one day to the next. The same should apply to programs. > On 9, Aug 2011, at 5:07 PM, steve uurtamo wrote: > i think that if it's "exactly the same code" during the entire > tournament, then it's a reasonable restriction. but i realize that > that's a minor issue in the big picture. "exactly the same code" > forever would be a very weird restriction indeed. _______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
