It's a matter of coding an algorithm that can quickly check that, with minimal resources on every flag fall. It is trivial for a human to see that the position is deadlocked. An engine would likely show an even evaluation, but other positions, that are not deadlocked would also evaluate as even.  So something would need to be coded to check for dead positions and in other cases, impossible to mate positions.

They also don't want to do computationally heavy checks on every game with flag fall, including for "is mate possible" ones. That is why insufficient material checks only look at the material the side with time has and ignores the other side's material.

I think at the grandmaster level though, gaining rating points are something extremely difficult to achieve, but losing rating points are much easier. Thus, perhaps knowing this endgame would save grandmasters from losing unnecessary rating points (or gaining fewer rating points if the opponent is higher rated because of a draw) by obtaining a win.

eric0022: Of course.

However, from what I've read, KBN comes up only once or twice in a tournament player's career and maybe not ever. So it's question of what economists call "opportunity cost":

the loss of potential gain from other alternatives when one alternative is chosen.

No one has the time to learn all of chess. If you're studying one thing, you're not studying something else. It's up to you, or your coach if you have one, to decide which study topics will maximize return on your investment of time and effort.

This is my kick with the endgame boys here. They always leave out the ROI (return on investment) calculation in studying endgames or assume without proof that endgame study is the royal road to chess improvement.

I went through my "big" database of games (over 6 million) trying to build some common endgame positions for study. The funny thing was that B+N was more common than 2B endgames (and Silman puts that in the book ).

I have also seen at least 3 games end in B+N in tourneys I've been to. So I think it is more common than generally accepted but still pretty rare.

You are right, coding seems to be the reason why chess.com does not declare those deadlocked positions as drawn.

I expected a king and two bishops versus king endgame (where the two bishops do not sit on the same colour squares of course) to come up more frequently than a king, bishop and knight versus king endgame.

You brought up a good point on the opportunity cost. I guess it depends on whether the player will love to learn something new or rather use the time to focus more on the more important openings and endgames.

I can't really speak to the quality of the database I have but the 2 bishops was rarer out of the 6 million games I looked at as I recall.

Martin_Stahl: Silman explains his reasoning for including the "Two Bishops vs Lone King." Silman claims the ending is instructive, even enlightening, and:

It's very easy to learn and only involves a small expenditure of time to understand.

KBN, on the other hand, is:

... not at all easy to learn and mastering it would take a significant amount of time.

Makes sense to me. Again, ROI -- return on investment.

Understood. I have both the Pandolfini endgame book as well as Silman's. Pandolfini does include it.

I think learning how the pieces coordinate is probably the best benefit of trying to learn it. That said, I could probably do it from the idealized position but not from a harder position. Though I also don't have the knack for the 2B endgame either (honestly with minimal study/practice).

I think learning how the pieces coordinate is probably the best benefit of trying to learn it.

Martin_Stahl: That's the usual rationale for endgame study.

If studying the KBN mate were the only way or the best way to learn piece coordination I would be studying it tonight.

But it's not and I won't.

But it is likely the best way to study the coordination of the bishop with a knight

Can you mate with just a Knight + Bishop? The answer would appear to depend on whether or not you're the Tarrasch GUI with Rybka. I have just downloaded a fresh version of this on a new machine after my old suffered a nervous breakdown I tried it out with this endgame letting it it play the side with the pieces with 20 minutes on the clocks and got the following.

This didn't happen on the old machine. The new processor is a 1.6 Ghz pentium with 8 Gig - I wouldn't have expected that. Has anybody else had a similar problem with Tarrasch?
 

It is possible my friend 

The move 10. Ng6 looks good. But perhaps 13. Nd8 is not so good. There is a method to win the endgame from the position after move 12.

I learned it years ago and once in an otb match game I actually had to do it and I think I managed it with one move left of the 50. I don't think I could do it now. Can't remember any of the theory and no intention of relearning it.

I recall that K+N+N vs K+P is a forced win from all piece positions in some pawn positions. The pawn positions have to be learned but it's a symmetrical and easily memorable pattern. The technique is easier than the K+B+N vs K forced win.

K+N+N vs K is a draw, however. There's no forced win. But, if the defending K has a pawn then, if that pawn is blockaded, the remaining N + K can force the opposing K into what is nearly a stalemate. Then the blockading N finishes the job of checkmate whilst the pawn runs for queening.

The game itself was not meant to be of particular interest, only to illustrate that my version of Tarrasch/Rybka cannot do the mate. I have since tried this again with 30 mins. on the clocks from a different position and the same thing happens.

I must have practiced the lone king against my previous Tarrasch/Rybka at least 50 times with 20 mins. on the clocks (some of these on a slower machine) and it never failed to mate. The Rybka version number is to the best of my memory the same, so I think there has to be a software problem.

I agree with your comments. His move 13 and previous bishop move are obviously aimed at forcing my king into the "wrong corner". This is in accordance with the guidance given in e.g. the first two links of post #51 and most other commentaries but is just incorrect.

What first alerted me to the problem was his move 5. Be4 which is a major blunder and would have probably added about a dozen moves to the mate had he played accurately thereafter. This moves from a closed position (see https://en.wikipedia.org/wiki/Bishop_and_knight_checkmate#Overview) to an open position. Again the move is in accordance with the common guidance referred to above, but here that's just b*llocks. When he did eventually force me into the wrong corner the problem was fully apparent because he then went totally doolally.

I guess it varies from software to software, specifically the algorithm which is employed in each software.

Search for Delatang's Method and then learn the so called w Maneuver. Not that hard.

eric0022 wrote:

I guess it varies from software to software, specifically the algorithm which is employed in each software

True, but I would expect any  chess program to be able to do this mate these days and as I said my old version could. I'll try the Rybka support forum.

@IfPatriotGames: K+N+N vs K is NOT a forced win. Set up the position just before checkmate, and black will have two squares to choose from. One leads to checkmate and the other to stalemate or escape.