lichess.org

bishop & knight v. king

what % of IMs do u think can mate B & N v. King? of GMs?

On any given day, about 95%. The 5% is just to allow for weariness after a long hard game, or distraction, or extreme time trouble.
I think all of them, IMs and GMs, know how to mate with B&N. It comes up so rarely that they might make a mistake and use up all their 50 moves, or maybe blunder a stalemate.
It can be checked, of course, by doing a database search by material.

I would disagree. I expect only about half, if that, of masters can mate with bishop and knight. If former Women's World Champion Anna Ushenina didn't know how, how likely is it that many less qualified masters know the trick? Grigory Serper did an article in 2013 where he pointed out several grandmaster games in which the grandmaster couldn't mate. And that was only the few games where it turned up! Many masters have never needed to mate with B+N, and probably couldn't care less about knowing.

GM Varuzhan Akobian says: "There was many incidents. Very strong players can't - never study it."
youtu.be/7OGAiz5p_L4?t=325

They never studied it because it is extremely rare, but when happens, it happens in the worst time (by "Murphy's law"). It is definitely useful to know as there is nothing complicated. Who is able to see attacked squares, it is even simpler and faster.


I would think that almost all of them can do it, but it's either a really long time since they practised it, or they're in real time trouble. It isn't actually too hard, it's just that it's very hard to do it without knowing it, and most people just say it's so hard, rather than trying to learn it.

Lichess has a good tutorial about it, which I went through. The non-combinatorial principles or objectives of the one method shown there (there is another one), are meant to guide your combinatorial work toward best path to mate. There are easy to remember (for me at least), being geometric in nature. make barrier, move barrier, plug holes in barrier, steer K toward correct corner (decided by bishop color), need eventually to use all pieces in some combinatorial pattern, and when space becomes tight, that is when I waste moves that I have to rediscover each time by letting the king escape once or twice, before getting in the mood. Being a lowly below 2000, I naively think that it is in this last aspect that the 50 moves may get wasted...

I figure if I can learn it, at my middling level, and do it to finish a game (once, only once) then anyone with capital letters in front of their name also knows how.
Someone who cares, with access to a database, ought to simply do a search and see what turns up. How often has it occurred? What were the results?

My data base of 4.2 million games has 937 games where they reached an endgame of bishop & knight against king. 2/100ths of one percent of games.
There were 237 draws, 710 wins. That's roughly 75%, or 3 out of 4 that did win the end game.
So, considerably less than my estimate of 95%, considerably more than the one half estimate.

@sparowe14 Interesting data. However, I'd argue that a large percentage of those who have never played that endgame and thus are not in the database probably don't know the mate. I guess we'll never know, unless someone takes a poll.

Are we still talking about percentage who know the mate? Why would the proportion of those who know the endgame be different among those who played endgame versus those who have not played them? I am missing an element.

unless numbers are low (1000), and you are saying maybe a fluke. but you are not saying that "those who don't know the mate are less likely to have encountered it". More that there is a chance of a fluke. to be sure, sample people who never encountered it: "do you know the mate". 1000 looks small compared to whole database?

Could the database be characterized a bit more.
Cumulative over the years, population data, or book? (as in opening explorer). I may not be using the right terms, I am not knowledgeable about the various type of chess databases there are. But, for books (don't have to be openings only), i thought that the population information was lost. That there can't be duplicates (I guess what is a duplicate, are they rare, typos, software skipping, or actual population duplicate). Sorry I digressed a bit. just curious about existence of population data which the question is about. Feel free to correct any wrong assumption I may have used, if clear. or not. thanks

Reconnecting