Re: Go program strength

[Dan Stromberg <strombrg@xxxxxxxxxxxxxxx>]

Thomas Goodsell wrote:

> I don't mean to suggest that tweaking a program is unethical (I believe
> it's perfectly acceptable), but one counter argument is that the program
> is tweaked by others making hardwired changes to the algorithms and
> exectional cases, rather than by Deep Blue (or whichever) looking at
> records of its next opponenet's past games.
>
> My $0.02
>
> Thom Goodsell
> thom.goodsell@xxxxxxxxxxxxxxxxx
> http://cc.usu.edu/~slvhh/

I think tweaking a program to do well against specific opponents is fine.

In fact, here's a possibility for such a tweak:  it seems that some of the
commercially available go programs do quite well against a normal
corner-corner-kakari kind of opening, but do much less well against a
"cosmic" opening (the one popularized by Takemiya Masaki).  Preliminary
results indicate that against one popular go program, this switch in
openings made a difference of about 60 points (!) in my game scores.