[computer-go] CGOS Monte Carlo Simulations

[Don Dailey]

On Mon, 2006-09-11 at 15:11 +0200, sylvain.gelly at m4x.org wrote:
> > I'm working on a new server and I am considering how to deal with this.
> > (...)
> > Ideas?   The first consideration is what extra features do we want to
> > provide, then it becomes "how to provide it." 
> 
> Hello Don,
> 
> first thank you again for providing cgos which is a great tool. I am looking 
> forward to seeing the new server :-).
> 
> In my opinion, the versioning could be too complicated and not so useful 
> because we can do versioning "by hand".

Yes, I am torn about this.   I do want to encourage program authors to
version their programs - it's "bad manners" for a given entity to
suddenly improve without warning - sucking rating points away from the
pool unfairly.    That's not a big deal but you can't meaningfully
compare your own program against a moving target either.    

Nevertheless, you can't force an author to do versioning no matter what
system is set up,  so perhaps what we are doing now is best.

>  The web page to provide the ability to remove entries from the display and 
> other options would be great, but it can be a lot of work for you. Perhaps, a 
> simpler way is to have an option to cgosGtp.tcl to remove the entries, like :
> 
> "./cgosGtp.tcl login passwd removeEntry"

Yes, that's a reasonable way to handle it.

> Having the possibility to start the rating from an "expected rating" would be 
> good. So that, the rating would converge more quickly. I don't know if we 
> must think of a way to avoid cheating like "I expect a rating of 10000 :-)". 
> Limiting the expected rating to a realistic range, and starting with a high 
> K-factor would be sufficient.

The way I rate unknown players is sub-optimal.  There are more efficient
ways to get a quicker convergence for a new player and I will probably
implement something much better for this.  

- Don



> Those points seem to me the most important, yet simple enough. Good luck with 
> this work.
> 
> Sylvain
> 
>