Hastings International Chess 29 Dec 2014 - 6 January 2015

[James Coleman]

I suppose it's a bit too much to hope for that we'll get a KBB v KN situation in the other game as well - and really go to town on this thing...

[Richard Bates]

Keith down to under 3 minutes now... (vs 34 minutes).

One other live board still going (Bob Eames-Adam C. Taylor).

EDIT: ....and there go the pawns (after 73.Bf3). Wonder if any programs can give the maximum moves to defend for Black once they are gone?

I assume they're both due another 15 minutes on flag fall.

[Christopher Kreuzer]

Keith down to under 3 minutes now... (vs 34 minutes).

One other live board still going (Bob Eames-Adam C. Taylor).

EDIT: ....and there go the pawns (after 73.Bf3). Wonder if any programs can give the maximum moves to defend for Black once they are gone?

I assume they're both due another 15 minutes on flag fall.

Oh. Possibly. Not sure. If anyone missed it, I put two diagrams in one of my posts on the previous page. Can anyone explain what is going on there? I *think* it must be something like to make it take as long as possible for the bishops to squeeze the king and knight to the edge of the board, you actually want the bishops to start from a position near one of the corners and have the defending king as close to that corner as possible. But I don't think Keith could have forced that in this case.

[Carl Hibbard]

Oh. Possibly. Not sure. If anyone missed it, I put two diagrams in one of my posts on the previous page. Can anyone explain what is going on there?

No, you sorted it now though?

[Christopher Kreuzer]

Oh. Possibly. Not sure. If anyone missed it, I put two diagrams in one of my posts on the previous page. Can anyone explain what is going on there?

No, you sorted it now though?

I meant explain how chess works, not how the software works.

[Roger de Coverly]

Can anyone explain what is going on there?

I've played down what the Shredder tablebase considers the best defence. The winning method is that it's eventually possible to win the Knight. So when the tablebase says it's 64 moves to victory, that's to mate. You capture the Knight with best play leaving a few moves of the 50 to spare. To do this, you separate the Knight from the King, not that the technique looked at all easy to apply. Essentially then the defender has to survive fifty moves without losing the Knight, or blundering into a mate.

[Richard Bates]

I think Keith's reached what used to be the old 'fortress' position (before it was cracked by computers) so odds must be on a draw now.

[Christopher Kreuzer]

Anyone in the mood to explain what is said here?

http://www.chess.com/forum/view/general ... g-position

An outline of the winning method is as follows:

1.Forced removal of K+N from KH type position. A position discovered by a couple of endgame analysts Kling and Horowitz.

2.The Bs can by reflecting off of the edges of the board can gain tempii in order to prevent the K+N from setting up a another KH type position or a pseudo KH type position on another part of the board. This GAINIING OF TEMPII BY THE BS BY REFLECTING OFF THE EDGES OF THE BOARD WAS THE BIG SURPRISE FOR HUMAN ANALYSTS AND THE KEY REVELATION FROM THE COMPUTERS THAT MADE THE FORCED WIN POSSIBLE.

3.With the K+N unable to reset the KH type position or the pseudo KH type position; it becomes possible to separate the K and N. Then it becomes possible to mate the lone K or trap the N on the edge of the Board. Once the capture of the trapped N is accomplished the position is now the well known forced mate of K+2B vs. K.

[Roger de Coverly]

[quote="Richard Bates"]so odds must be on a draw now.[/quote

It's still being quoted as a win, but the moves taken to reach the current position have pushed beyond the maximum of fifty moves allowed to catch the Knight.

[Christopher Kreuzer]

Clearly Lagarde doesn't know the winning method, or has forgotten. Which raises the question - people say you should learn to mate with BN in case that arises. Is this ending more or less likely than the 'standard' obscure ones, and should all grandmasters know how to win with BB vs N?

[Matt Fletcher]

Jonathan Bryant and I played through the endgame from pretty much this position (move 97) over the summer. We didn't get very far before giving it up as a draw.

Putting it through the tablebase, Keith's starting position was almost ideal with the shortest line to mate being 68 moves. Over the subsequent 20 moves it's come down to 55.

[Roger de Coverly]

and should all grandmasters know how to win with BB vs N?

Out of a collection of over 5 million games, there are over one thousand with KBN v K, but under a hundred with KBB v KN. The recorded historic first example is from Botvinnik and Tal in 1961. In the old days of adjournments they could mug up on it once it arose.

The broad generalisation is easy enough, you win by separating the King from the Knight. The practical method of doing this is complex, but bouncing the Bishops off the edges of the board gives a hint.

[NickFaulks]

Keith really looks as though he knows what he's doing. It will take a big mistake to lose now.

[Christopher Kreuzer]

Out of a collection of over 5 million games, there are over one thousand with KBN v K, but under a hundred with KBB v KN.

What percentage of draws and wins for the KBB v KN examples?

[Roger de Coverly]

Putting it through the tablebase, Keith's starting position was almost ideal with the shortest line to mate being 68 moves.

The start position was lost because the Knight could be won within fifty moves.

Over the subsequent 20 moves it's come down to 55.

It had to reduce by one move for every move played to stay on target, so it should now be a draw regardless of the tablebase still being able to win it.

Shredder doesn't have it, but a practical addition to ending tablebases would be a move counter, so a prediction of draw by fifty move rule could be made.