-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

All I need is a linux machine with a good internet connections and we
can set up this match.

We won't be proving anything because obviously Mogo will be better if
you choose the right hardware (since MFGO that David will run is fixed.)

But at least we will have a benchmark which I argue is very important -
but the hardware and time control must be stated as a matter of record.

- - Don




David Fotland wrote:
> At least for Many Faces 11, if you run it at the top level, it will play the
> same no matter what hardware you use since the search parameters are fixed.
> 
>> -----Original Message-----
>> From: [EMAIL PROTECTED] 
>> [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
>> Sent: Thursday, October 11, 2007 10:33 AM
>> To: computer-go
>> Subject: Re: [computer-go] best approach forward
>>
>>
> David Fotland wrote:
>>>> I agree. Computer go needs someone who will play large 
> tournaments are 
>>>> publish results.  I'm also curious how Many Faces would do against 
>>>> Mogo on 19x19 in a long match.  Mogo is much better at 
> endgames, and 
>>>> is much greedier, but Many Faces is much stronger tactically.  
>>>> Certainly if there were ratings it would motivate me to 
> work on Many 
>>>> Faces more.  I think games against mogo would help me find bugs in 
>>>> Many Faces.
>>>>
>>>> Are you volunteering?  Is the 19x19 cgos still going?  I could put 
>>>> Many Faces on it if there is any interest.
>>>>
>>>> David
> I believe Many Faces is probably stronger than Mogo but I don't know
> that this has been proven.   But that's my point, I don't think anyone
> really knows for sure where any of the top programs really 
> stand unless they know due to private testing on their own or 
> anecdotal claims.
> 
> But what I had in mind in some kind of ratings agency where 
> the conditions are controlled and everything is completely open.
> 
> Here is what is required:
> 
>   1. Someone with at least 2 equal DEDICATED computers plus a server.
>   2. Someone willing to do the work.
>   3. Software to manage the testing.
> 
> Anyone wishing to get on the "rating list" would have to 
> submit a binary or executable of some kind to the "testing agency."
> 
> Some kind of restriction concerning which programs get 
> tested.  With a limited resource testing agency you can't 
> have hundreds of programs and versions being tested.
> 
> Of course something like CGOS is simpler, but there is no way 
> to verify
> what is being run and what hardware it is being run on.   There is
> nothing to stop you from running Many Faces or some other 
> program and claiming it is your program or running on an 8 
> processor system and
> claiming it's 1 processor.   (although it often becomes obvious after
> some time.  I suspect that if someone got Many Faces running 
> you or someone else would be able to detect this sooner or 
> later if you were paying attention.)
> 
> Perhaps something like CGOS would be required to qualify.   Get your
> program running well on CGOS, after so many games and a good 
> rating you
> earn the right to submit a program to the agency.    Or perhaps the
> agency evaluates any submitted programs and decides which 
> ones get rated.
> 
> The computers don't have to be equivalent either.  But each 
> machine/software combination has to be considered a separate identity.
> Example:
> 
>    1. Many Faces version X running on core 2 duo e6700
>    2. Many Faces quad version X running on xyz
> 
> and so on.
> 
> Unfortunately, this does require some dedication on someones 
> part and a
> bit of trust.   All games should be public and so should the testing
> methodology, etc.
> 
> My point is that this probably won't happen in computer Go 
> but it happened long ago in computer chess.
> 
> - Don
> 
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
>>

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

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHDnBADsOllbwnSikRAiywAKDiLuimLPRnw381U7qyEoew2WYIOgCgyxRj
LrHTzagTvZgZaGftuyjM+IY=
=Ga0r
-----END PGP SIGNATURE-----
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

Reply via email to