Re: [computer-go] how to use GTP in place of GMP

[Don Dailey <drd@xxxxxxx>]

And this argument works backwards too.   If I have developed a super fast
playing algorithm that specializes in playing reasonable move very very
quickly,   then I could view the slow programs that beat it as having an
"unfair" advantage.  

- Don


> Richard,
> 
> Your argument isn't  really valid, ANY computer based  time control is
> in some  sense arbitrary.  I have  a program that  ALWAYS plays better
> when given more time.  Allowing  1 hour per move unfairly penalizes it
> when it can do much better with 1 day per move.
> 
> Whatever  time  contol  is  chosen  is fair  to  all  participants  if
> announced in advance and you are given a choice to compete or not.
> 
> And  if your  program really  does  need a  lot more  time than  other
> programs, why is being fair based on your needs?
> 
> I'm not  saying this because I have  a fast program, because  I have a
> very slow program.  I test a  lot against gnugo which makes moves very
> quickly.  I can win almost half the games when I give my program about
> 10 minutes  per move (but the games  take way too long.)   Can I claim
> that any tournament  that doesn't let me think 10  minutes per move is
> unfair?
> 
> - Don

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/