Re: [gnugo-devel] Comments on lazarus.tst

[Daniel Bump]

> > loadsgf games/nngs/Lazarus-gnugo-3.1.19-200201092246.sgf 48
> > 3 gg_genmove black
> > #? [Q8]*
> 
> P8 also looks okay to me, or does it have some problem that I've
> missed?

Arend and I both feel that Q8 is better though it's difficult to
explain this convincingly. Both leave some aji.

Maybe we should allow P8 but on the other hand P8 over Q8 seems
to typify one of GNU Go's weaknesses, which is a failure to
simplify the position.

> > loadsgf games/nngs/Lazarus-gnugo-3.1.19-200201092246.sgf 70
> > 5 gg_genmove black
> > #? [R13]*
> 
> M8 to capture some cutting stones doesn't look out of the question to
> me. I would accept that as a correct answer for both test cases.

Here I agree with Arend---M8 is not the right move here.
Both M8 and R13 are better than the game move but R13 is
clearly better than M8. 

At this point in the game W has four weak groups. B should
not lose this game.

> > loadsgf games/nngs/Lazarus-gnugo-3.1.19-200201092246.sgf 257
> > 10 gg_genmove black
> > #? [P16]*
> 
> This should be move 258. P16 isn't the only move to connect. N16 and
> O17 also work. Furthermore N16 gives the extra bonus of capturing O16.

Agree, P16 isn't the only move and N16 looks better.

Arend wrote:

> > +# Killing the upper right is way bigger than saving the lower left.
> > +loadsgf games/nngs/Lazarus-gnugo-3.1.19-200201092246.sgf 254
> > +17 gg_genmove black
> > +#? [T8]*
> I think the upper right is dead as it stands. I've put variations in
> the sgf-file below.

I don't think so.

Instead of attacking the 5 nakade stones at T8, W can attack at T6. This
gives W a live eye shape.

Dan