** **

1. Why don't the computers agree about the proper recapture for White?

2. Will AN want to capture a piece that has moved 1 time with a piece that's moved 3 times?

3. If AN does not play Bxd3 this move, should we consider 13. Rd1?

The obvious answer in most cases of engine divergence would say it's because there's not much difference anyway. They calculate as deep as they can, the resulting positions have much the same evaluation, so choice comes down to tiny differences in their selection algorithms.

But those tiny differences may be the more important answer. I'm no computer expert, but I know that the raw brute-force calculate-everything routine has been superseded in the new generation of software. Now they have more efficient pruning devices and ways of focussing on key lines - the artificial intelligence version of positional understanding.

But it's still not 'real' positional understanding, is it? (Whatever that is.) They don't and can't think the way we do. Human brains might be dodgy rapid calculators, but they're great at spotting patterns, analogies, links, connections...

So my answer is this. Computers differ because their programmers have given them different ways of mimicking human thought. They're still great at doing tactics, but a fine choice like this is about more than tactics.

click for larger view

Very different from our situation, true. But the point is that Botvinnik 'saw' where the rooks would be needed long-term. Most players, even GMs, would think in terms of playing the Ra1 to c1 or d1, and the Rf1 to e1. Standard centralizing play. Botvinnik saw further. He understood that however the central tension gets resolved, the rooks need to be on the e- and f-files, whether to support a pawn push, simply protect squares, or threaten tactics against the enemy queen.

I won't go into it, but it's a textbook example of world-level intuition. I don't think a computer would play this move - if I'm wrong and somebody's engine does play it, it's probably doing so by accident. ;)

Relevance to World vs GMAN? Slight. But I'm trusting computers less and less since this game began (while continuing to use them, play them, analyse with them, make a living by hitting their keyboards, etc etc).

** **

I think most the ideas are still in place, so I will offer this for consideration:

** <Rd1> Analysis ** I will try to summarize the points being made on my forum.

Points in favor:

1. Rd1 would be a novelty, or semi-novelty, so the World may be able to pool our resources and out work AN. And such a move might be a lot of fun.

2. Rd1 allows the World to temporize the Queen side minor pieces and wait one more move to decide where we might want them.

3. Rd1 offers us the chance to recapture on d3 with the Rook, and leave the c file closed, and the Queen where she stands.

Points against:
1. Black's response is unknowable. Different versions of Rybka don't even agree on Black's response. 0-0 and Nxd3 are both possible. It would be very hard to work on lines until Black's move is played. <Rookfile> has also suggested Black might play the novelty 12. Rd1 Qc7.

2. There is some general sense that the World must generate development quickly, and a Rook move now might be too slow, as our Queen side minor pieces are still at home.

3. There are generally good reasons why certain moves aren't played, so we may be taking on too much to try something the best GM's haven't played.

Some other thoughts that don't seem either positive/negative.

1. If Black plays 12...Nxd3, the computers want 13. Qxd3. I think that is simply wrong. That might be one area we may improve upon computer analysis. But it's just an unknown, since Black can have a good game without Nxd3.

I have a question about a proposed line suggested by <Rookfile> I am not a strong player, and have no engines for analysis. I would request someone respond on my forum (click on Elvis) to this post. I realize that the proposed move 12. Rd1 has little chance of being voted in, but that is of no concern to me. So here goes:

<Rookfile> proposed this line: <C) 12.... Nxd3 13. Rxd3 e5! (Let's transpose this into a Najdorf like setup) 14. Bg5 Be6 15. Na3 Rc8! (ruling out white ideas of Nc5) 16. Rad1 Rc6 17. c4 !? 0-0 18. c5 - a slight advantage for white.>

1.Non-commital or not revealing the idea as to where our is going.

2.If black plays Nd3 then Rd3. on d3 has multiple ideas. We can double or triple on the d-file. We can also use it for attack if the need arises by playing Rh3.

13.Rd1 Bd7 14.Na3 Nd3 15.Rd3 Rc8 16.Rad1

and black has some serious problems..

Ra4->h4->g4..

*** ***