Yes, it's understood that the eye rule most of us use is not perfect and
we have identified bad cases before on this list.

My analogy is that you wouldn't put expensive leather seats in a cheap
economy car.   A simple eye rule is more than sufficient for random
play-outs.  A more sophisticated rule is interesting to consider of
course,  but I'm pretty sure it would not be a measurable improvement to
a random playout based program.    In fact I think even the really good
programs like mogo use this same simple eye rule.

I thought the Gobble eye rule was the one we are using now?   Or maybe
I'm thinking of a different program?

If you are interested in corner cases I think some previous postings
here are worthy of consideration.    Did you also check Sensei's page? 

- Don


On Sat, 2008-10-11 at 15:01 +0100, Claus Reinke wrote:
> Don's draft standard reminded me of the corner cases. So here's
> an even simpler example, this time trying to show that dead invading
> stones can poison playout analysis depending on which definition of
> pseudo-eyes is used ('a' is A1, 'b' is C1).
> 
> That makes three attempted examples so far (are the examples
> valid for what they are trying to show?), trying to show:
> 
> - the "standard" Gobble pseudo-eye can both allow real eyes
>     to be filled and prevent sacrifices involving filling false eyes
>     (in all cases so far, resulting in own groups counting as more
>     likely dead, and opponent groups counting as more likely live;
>     but since the playouts are symmetric, that implies miscounting
>     both ways)
> - the local pattern approach of Gobble can be poisoned by
>     dead opponent neighbours
> 
> I understand that some engines use pseudo-liberties, or may
> not have connection information easily accessible all the time,
> but for those who do maintain proper liberty counts: are there
> any examples where Olga-style pseudo-eyes are not preferable
> to (at least as good as) Gobble-style ones?
> 
> Claus
> 
> (
> ;
> FF[4]
> GM[1]
> SZ[19]
> AP[Jago:Version 5.0]
> AB[ar][bs][bq][cq][dr][ds][dq][aq]
> AW[br][cr][ap][bp][cp][dp][eq][er][es]
> LB[as:a][cs:b]
> C[The black group is "unconditionally" alive.
> 
> Gobble: 'a' is not an eye, group is unsettled (50% chance: black 'a' dies, 
> black 'b' lives).
> 
> Olga: 'a' is an eye, black 'b' the only possible local move: group is alive.
> 
> Oleg: 'a' is not an eye, group is unsettled (50% chance: black 'a' dies, 
> black 'b' lives).]
> GN[playout-eyes3]
> )
> 
> 
> 
> 
> _______________________________________________
> computer-go mailing list
> computer-go@computer-go.org
> http://www.computer-go.org/mailman/listinfo/computer-go/

Attachment: signature.asc
Description: This is a digitally signed message part

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

Reply via email to