The unlikeliest chess move – revisited

Is there anything left that computers can’t solve? My attempt at beating the machines a few years back resulted in the creation of a very different type of problem. I wanted to know what the least likely move in chess would be. I wasn’t interested in best moves, accurate moves, or stunning moves. Just one that makes you wonder how you got there.

I was pleased with the result – there was much discussion over the puzzle during an annual Christmas Curry evening with my chess club, and it uncorked a few bits of chess magic to appreciate and admire. However, I fell into the other trap of modern creativity – it had already been done before. Four years ago. And solved. By a much better player than me – GM David Smerden’s blog

Okay so it turns out, all was not lost by any means. On deeper inspection (after the initial heartbreak) I realised that my puzzle was crucially different by being much less logical than Mr. Smerdon’s – probably part of the reason he is a GM and I’m not! His answer to ‘The most unusual move in chess’ was still about best moves – about trying to win (or draw) a chess game. In contrast mine is purely about possible strange notations and what they say about a position. The delights and surprises are in mentally constructing ever wackier positions until you find the limit of impossibility (or a simpler way to notate the move).

smerden image

GM Smerdon’s composition is a superb idea – the best move for White is actually to promote to a bishop on a8 – utilising the bishop’s weakness to eventually force stalemate

The puzzle

If I could trace back my curiosity for ‘unusual moves’ it would be something like 15 years ago, perhaps a match for my school, where I first had to notate a move like ‘R2e4’. The nuance of having to specify that the rook moved from ‘2’ (2nd rank) to clarify which rook was moved to the e4 square could be viewed as an imperfection of ‘modern’ notation (in the old days you would write ‘e2-e4’ if you moved any piece from e2 to e4). But I do not take this view, as this puzzle and its various gems wouldn’t be able to exist without it.

Going back a step, it was clear that ‘R2e4’ would be less likely than a bog-standard ‘Re4’. Moreover, ‘R3e2’ was even more unlikely, as it implies there is one rook on e3 and one rook on e1 (by the way, if there were rooks on say d2 and e1, you would write ‘Rde2’ to move the one on d2 to e2 – as the files take precedence over the ranks.) But I went a bit further into less and less likely moves. How about Ba8 check? It’s not so clear – the bishop cannot check a king itself by moving to a8 (or any corner square).

So this search is one for the most unlikely move in chess.

Consider a move like ‘0-0-0 checkmate!’ Quite unlikely! Although it has been played before in many real games. This thread shows many of these. Strangely it starts with a miniature from a former world champion actually declining to play 0-0-0 checkmate, instead choosing Kd2 mate, perhaps rather like Ronnie O’Sullivan’s famous 146 break , he was happy enough just to show he could do it.

So to the point – ‘unlikely (but possible) moves’. The question became this:

“What’s the most unlikely move to appear in a legal chess game?”

[Note – we are not including annotations in analysis such as ‘!’, or ‘??’. Nor does it include move numbers, for example ‘7.gxh8=N++’. No – this puzzle is just about what you must physically write on the scoresheet.]

Some themes

This is reverse engineering in a sense – rather than seeing a position, finding the best move and writing it down, we see a move written down with no other context – and figure out what it tells you about the position. Back to our Ba8+ example, we find out that the check must be by discovered check, because a bishop moving into a corner attacks no more squares than it was attacking before.

Setting about this problem in a ‘human’ way, a number of ‘unlikely’ themes spring to mind:

  • Castling
  • Check
  • Checkmate
  • Captures
  • More than one of the same piece able to move to the same square

[Note: As far as I know, there is no official notation needed for either en passant or stalemate – so they don’t make it into this puzzle.]

It stands to reason that a checkmate is always more unlikely than a check – or a non-check. So our solution must be a checkmate. It also became clear to me that the theme of ‘More than one piece able to move to the same square’ had the largest scope for unlikeliness. Especially for queens – because you only start with one.

A solution?

There are too many interesting nuggets here to share in just one article. I have decided on my candidate solution for the ‘unlikeliest move’ – which I will share soon enough. But in checking over my own thoughts and explorations for this puzzle – I found something quite surprising; the unique qualities of each piece make for some very different types of thinking in finding their own unlikeliest move.

Therefore I will structure the rest of the musings on this puzzle by piece – part two will cover Kings, Queens and Rooks, and part three Bishops, Knights and Pawns. But here’s a little taster:


Black to play and mate in one? Easy! Nxh8 mate! But which one?

If a knight on the rim is dim, then a knight in the corner must be… mourned for? (Editors note – Keep smoking whatever you are smoking Mike!). That is, unless, it can deliver checkmate! Which move did you choose? ‘Nfxh8#’? or ‘Ngxh8#’? Because therein lies the unlikeliness of all this. Forgetting the position a moment, if the notated move was just ‘Nxh8#’ then the king could easily be on f7 and just be mated directly with the knight. Not so unlikely – and has undoubtedly occurred before. But introducing a second knight by stating the move ‘Nfxh8#’, suddenly there must be a knight g6 as well (the only other square with access to h8) and so the king must be somewhere else – and must be mated by discovery! Hence the concoction above; there are other variations of course, but it shows how much is implied by a single notated move: The two knights must be on f7 and g6, there must be a piece on h8, the king must be on either f6 or g7, two rooks/queens must be on the same file and rank as the king, and other pieces must block all the other squares around him. Quite a kingdom for a horse move!

If you want to give this puzzle a mulling over during the next few weeks, we’ll be back with parts 2 and 3! If you arrive at any candidate ‘unlikely moves’ – please do send them in!


Mike Harris

Mike is a regular pretender in Bristol’s top division and can also be seen propping up local tournament ladders. He writes a regular column for the Bristol Chess Times and plays a solid 20 openings a season.

4 thoughts on “The unlikeliest chess move – revisited”

Leave a Reply