I'm not sure if it is possible. I was trying to compose a position where promoting to bishop is the only way to win and if not the only way to win than the fastest way to win. Can anyone come up with a position for this?

Yes, there is. Let me see if I can find it.

http://www.chessgames.com/perl/chessgame?gid=1253800

From a real game of all things (Q or R Qf7+ =, N Qa7 and I don't see how to do anything, although I won't swear that there isn't any way to win with a8N or some plan besides a8 here)

also from a real game, the exposition is on Tim Krabbe's site:



There are also some examples in composed problems. Krabbe also asks the question whether it has ever happened in a practical game that underpromotion to a bishop or a rook was the only move to DRAW -- so far he has not found an example.

The initial question points to the Savedra position. Is there ever a situation where you promote into a rook instead of a queen? Yes, there is! The situation occurs whenever the additional power of the queen creates a stalemate. Bishop over queen is just the same.

Another situation that comes to mind would be a Babson task, where one player promotes a pawn to piece X, and the other has to promote a pawn to piece X in order to mate in time. So, if Black defends by a Bishop promotion, White has to promote a pawn to bishop to make in however many moves.

http://www.xs4all.nl/~timkr/chess/babs.html

The trivial position is Kg4 g7 vs. Kh6. g8=R is the only winning move.

interesting stuff. thanks for the replies. How many of you have had to underpromote in a game? I've played thousands of games online and the situation has never arisen. I've had the situation where I underpromote to 5 knights just bc I am annoyed that the guy isn't resigning but it has never been necessary. 1 out of 5k games sound about right where underpromotion is needed?

I've definitely underpromoted to a N to draw more than a few times in the standard position (pawn vs. rook). I've also won with this trap back in my noob days http://www.chessgames.com/perl/chessgame?gid=1250704 I'm not sure if I've ever underpromoted out of (perceived) necessity other than those.

Ditto TomCowley with promoting to a knight to draw. I had a position where if I promoted to a rook or queen I would stalemate my opponent. So I promoted to a bishop. My opponent jumped up from his chair and very loudly said, "Stalemate! It's a draw!" Everyone is now looking at us. I quietly whisper, "It's a bishop." He sat back down and kept muttering in a whiny impersonation "It's a bishop. It's a bishop." I laughed pretty hard.

A year ago I reached this position in an online blitz game.



I'm white, it's my move. I am already a rook up, and although I am a move away from being mated myself, it is easily preventable.

Still, got a mate-in-6, supposedly, and it involves an underpromotion (I played the first 2 moves before he either resigned or lost on time, I forget).

Odd I almost think you HAVE to do this. I'm trying to find another way to win without under promotion and since black's mate in 1 can't simply be blocked its hard to find!

I have underpromoted many times. Usually to a knight to get a check. One possible opening thing: 1. d4 d5 2. c4 e5 3. dxe5 d4. e3 Bb4+ 5. Bd2 dxe3!
6. Bxb4? exf2+ 7. Ke2 fxg1=N+!

The rook can move on the f-file.

I thought so too for a long time. Before posting, I checked with a computer and found that something like 1. Rf7 is very strong for white as well.

Still, maybe it'd be nice to edit this position to make the black mate unstopable except by the combo from my game. Gonna go try now...

Ok, how about this, with white to move.





I think white loses otherwise.

lol ok how about this. You're in a situation where the queening square is under attack by a minor peice or rook but it also protected by you. So if you promote to a queen it will obviously be captured/traded off. However if you underpromote to a peice of equivalent strenght to the one attacking, perhaps your opponent might leave it? And it is also a situation where you for whateve reason want more material on board. Basically, an underpromotion HOPING for a sub-optimal respsonse from your opponent. However even if your opponent makes the optimal response (taking) then the underpromotion is irrelevant and doesn't hurt you.

I have done this on the internet at least a couple of times. Once in a 'serious' game online I had 3 rooks for quite a long sequence. I still insist the underpromotion was valid, haha.

The thing is that the minute you underpromote to a piece that is strictly worse than a queen, alarm bells will start to go off in your opponent's head, and they'll be more likely to think through their response as much as they can.

Surely such a position is not even possible? The only reason to underpromote to a bishop/rook is to avoid a draw, making a draw the objective removes that reason.

For the first part. Yes the alarm bells will go off but since the response we're looking for is avoid the capture of the promoted piece, and that will happen if we promote to Q, might as well at least try it.


On the second part no not exactly. You may WANT to draw because if you promote to full strength you will still be down in material or your opponent will quickly win the promoted piece and be u;/already have forced mate etc. It is possible the underpromotion can 'self-paralyze' you into getting a stalemate in an otherwise lost position. There are compositions with this theme, but the odds of it arising in a real game are astronomical.

A weaker piece has fewer moves, meaning you can use it to stalemate yourself. They exist in compositions.

in fact, there's a very simple one I already linked in the thread.

And a really cool one here. White to play and draw. Rybka 3 doesn't find it, no matter what depth it searches to. Don't know about Rybka 4 or other engines.

The full main line is quite difficult to find, but very beautiful. kudos to anyone who can solve it (you have a massive hint!)

This is the most famous example I think.
Knigth promotions happends now and then, but bishop promotions are extremely rare.

@ RoundTower's position - I was too lazy to try and solve it so stuck it on Firebird and it assessed it as equal with a mainline involving 5.a8=B absolutely instantly.

yeah, it's an acknowledged "bug" in Rybka, rather than a problem that is intractable for computers. I say "bug" because the practical benefit of considering bishop underpromotions is likely less than the downside of devoting a few extra processing cycles to them.

it's good anecdotal evidence to suggest that Firebird is not derived from Rybka, though.

Interesting example, but what do you do if black does this: 1. Bg7+ Kxg7 2. e8N+ Kh8. You are basically back where you started, one move from checkmate. A better move may be 1. Qc3+ Qxc3 2. e8Q..