The Babson task

To an outsider, we chess players might seem a rather uniform breed. Studious and contemplative, we spend hours absorbed in a board game to no apparent end. It is the archetypal thinker’s hobby.

But within the subculture, there are many, perhaps even a majority, who identify as pragmatists, not thinkers. Results are the driving motivation. At the board, they are drawn to ideas which are likely to wrong-foot the opponent, with no special regard for their objective merits. In their study, they disdain the more obscure, frivolous or unrealistic chess problems such as the one in the puzzle below. What, they ask, is the purpose of seeking a subtle mate in 2, when a simple move like 1 Qxb8+ wins easily? For some players, the game’s rewards are not intrinsic at all. Chess improvement is a serious matter, and their online or tournament rating is no less tangible than their bank balance.

I mean no disparagement by this caricature; the embrace of competition is fundamental to the game. But equally, I admire those who are drawn to the game’s purely cerebral aspects. For example, it is well known that king, bishop and knight can force mate against a lone king on an 8 x 8 board. Is there a technique which will work on a 1000 x 1000 board? Somebody has written a paper on that. (Spoiler: yes!) What else but intellectual curiosity could lead one to explore such a question?

In 1925, a man named Joseph Babson proposed a task which has fascinated – and frustrated – generations of composers. Can one create a chess problem in which White (who ultimately gives mate) must meet a Black pawn promotion (to a queen, rook, knight, or bishop) by promoting to precisely the same piece on the other side of the board?

Circumstances demanding underpromotions (anything other than a queen) are rare, so this is fiendishly difficult to accomplish. Nevertheless, over the past century, a handful of composers have scaled the summit. Most of these were in the form of ‘Mate in X’ type problems, but composing a ‘study’ (such as a problem in the form ‘White to play and win’) was considered harder, or even impossible. It is one thing to find a position where an underpromotion enables a brisk mate, but rarer still for that to be the unique winning move.

In recent weeks, a new composition has appeared which realises that dream, composed by Gady Costeff. White is to win, but unusually for a problem, Black moves first (see diagram).

White threatens to promote the f7 pawn, and let’s take it on faith that 1…Ke7, 1…Nd6 and 1…Nc7 all lose in the long run – the details are messy but I understand that has been checked extensively. As for the promotions: First consider 1…d1=R. Only 2 fxe8=R! will do in response. Not 2 fxe8=Q? Rxd4+ 3 Bxd4 Nc5+ 4 Nxc5 Rb4+ 5 axb4 stalemate. 2…Rxd4+ 3 Nc4! is the crucial point, freeing f7 for Black’s king. White wins easily. After 1…d1=Q Black is up a lot of material, so only a queen will do: 2 fxe8Q! Qxd4+ 3 Nc4! Nc5+ 4 Nxc5 Rb4+ 5 axb4 is not stalemate: the Qd4 can still move so White wins. After 1…d1=B, Black intends the same stalemate ruse with Nc5+ and Rb4+, so only 2 fxe8=B or N could work. Of those, only 2 fxe8=B suffices, and then 2…Ke7 3 Bxg6 hxg6 4 Bc3 wins slowly but easily. Finally, 1…d1=N threatens Nxb2#, so White must respond with check: 2 fxe8=N+ Ke7 3 Nxg6+ hxg6 4 Rxc2 Kxe8 5 Rxe2+ wins comfortably.

Costeff’s composition is a milestone achievement. Like Roger Bannister’s four-minute mile, it will surely inspire future composers, who can now be confident that the task is achievable.