On 23 Mar, 17:28, (Dr A. N. Walker) wrote:
In article ,
David Richerby wrote:

[...] The evidence says that
1.a3 is a bad move but we don't know for certain [...]

Whoa! *What* evidence says that? There is certainly the
empirical evidence that GMs/IMs don't play it very often, and that
1 d4/e4/etc seem "sharper". But that's a different matter. *If*
"perfect chess" is a draw, then *any* move that avoids losing is
as good, in principle, as any other.

Agreed. If one considers the praxis of Miles/Sokolsky/Benko/Norwood
etc. then that does, indeed, appear to be the case.

If, further, as most of us
expect, the drawing margin is in fact quite wide, then I would be
somewhat surprised if *any* first white move is so bad as to lose
by force.

Quite so.

This, in turn, is quite likely to mean that [eg] 1 a3 d6
2 c3 Nd7 3 Ra2 ... is as "perfect" as any other "drawn" opening.
The only reason to play something particular is the hope that an
inferior opponent is more likely to blunder.


Yes. Any given main line *may* be finally refuted, one day, but that
day is a *very* long way off, and, most probably, shall never arrive.

FWIW, when explaining alpha-beta pruning to students, I
usually use the example 1 e4 e5 2 Ba6 as an instance of a line that
*is* [I'm tolerably sure] bad.

Are you ever intolerably sure ;-)

The point of the pruning is that once
you know [tolerably certainly] that [eg] 2 ... Nxa6 wins for Black,
there is no need to waste time examining whether 2 ... bxa6 or 2 ...
Qf6 or anything else *also* wins, which speeds up the analysis by a
factor of around 30 [and again by a similar factor on every black
move]. [Sanny perhaps needs to learn this, tho' that might spoil
the fun.]


Yes. All this has been stated before in these groups, not just by you
Andy, but also, for example, by Remco Gerlich, and I have stated it
too. That's what gets me about Richerby. He appears to have the
attention span of a hamster, and is evidently unable to check the
archive. Plus he can't read, and can't follow the thread of an
argument. Plus, he's persistently idiotic.

Apart from that, he's got everything going for him, no doubt.

Somewhat on the other hand, I think the example of endgame
tablebases ought to make us very wary of claiming any knowledge of
the theoretical outcome of any level-ish position.

Define "level-ish". A position is level, or it's not. When is an edge
meaningful in an endgame?

Experience had
been that KQvKR and KBBvKN were respectively usually an "easy" win
and a draw, resp; the computer has shown that both are "difficult"
wins from general positions.

KQkr *is* easy for the side with the Queen. Even I can do it, and I'm
a patzer. It's just a question of learning the patterns. No doubt the
same is true of KBBkn.

Similarly with other 5/6 piece endings.
We are quite likely to find that some of our opinions about typical
middle-game positions have in fact been mass delusions.


Indeed, systems like Hydra may well demonstrate that certain
positions, previously considered sound, are, in fact, critical. All of
this is rather off-topic, however interesting.

Mark Houlsby

--
Andy Walker, School of MathSci., Univ. of Nott'm, UK.