What do people call a move which maintains a winning position but is the worst among all such moves?
For example, h8=N+ is such a move in the following game:
[FEN "8/7P/6k1/8/8/8/B1K5/8 w - - 0 1"]
1. h8=N+
I will not normally play such moves in serious games, but in casual games such moves might be good practice for endgame skills.
The short answer is no.
In 25 years of chess experience I have never heard such a term. There isn't unless there is some definition somewhere in a shady club packed with sadists who love to mate with several knights!
In a winning position, the best move is the one with the shortest depth to mate. So, the “worst winning move” would be the one with maximal depth to mate. So it seems reasonable to call it the move with maximal depth to mate. However, there are two important caveats:
If you repeatedly play these kinds of moves, then your depth to mate may increase, rather than decrease. Due to the 50 move and 3-fold repetition rules, you might actually draw on accident by playing these moves. At this point, these moves are no longer still winning, so some care needs to be taken.
Practically speaking, there can be benefits to playing such a move over a move with lower depth to mate. For example, the “best move” might involve a tricky sacrifice, whereas the “worst winning move” might be a trade that leads to a simple, easily won endgame. If you can’t analyze the sacrifice fully, but are certain of the won endgame, it’s hard to call the safer play a mistake.
Take note that without the 3-fold repetition rule and the 50-move rule, choosing a worst winning move (i.e. one that leads to a losing position for your opponent with the maximum moves under best play) at each turn may lead to an infinite game rather than a win!
But anyway, the move you suggest in your question is certainly not the worst winning move, as once you promote to a knight you will have to reach checkmate within 50 moves. A worse winning move is h8=Q and then to trap the opponent to 2 squares before dallying around with your pieces and then forcing the opponent to capture your bishop just before the 50-move rule kicks in, giving you another 50 moves to play cat and mouse before the finishing bite.
There is no measure which distinguishes one winning move from another, other than perhaps number of moves (which is only important if that number could cause a draw). Winning is winning.
To define a score which differentiates winning moves, we would need an off-board metric. Consider: if someone is actively trying to practice knight and bishop end games, this may be a very good move. However, if one is trying to make the game shorter (which is a very typical metric, even if there is no draw in sight), it could be a poor choice compared to promoting to a queen.
However, if I may draw from some non-chess history, I'd recommend a term from the story of Mel. Mel was a software developer in the age of drum memory. Drum memory was a fascinating beast because all of the data was stored on a rotating drum that rotated at a very predicable rate. If you knew how long your instruction would take, you could make sure that the next instruction's memory was right underneath the read-head when it was needed.
Mel was known for doing delay loops where he would intentionally place the next instruction so that it just barely passed the head before it was needed. Then the drum would have to do a full circle before reading the memory. There were optimizers which would place the memory in the optimum place, but he sought the opposite.
His term: it was the most pessimum location.
I think that if you called this move "the most pessimum winning move," you would at least get a few smiles when you explained it!
Least best?
It depends what you mean by worst.
Is that from your perspective or from the one of the person whose decision was to make that move you consider 'worst'.
