The good thing is that at lower level, so many players never find the "resign" button so you can get some really creative checkmates. What I usually do in this kind of situation is that I bring the enemy King to a corner, with only one square free next to it, then I promote all my pawns into Knights, then I put all of my Knights in one line, the last Knight move being a checkmate.
It has to be a move that has occurred once because these moves don't happen in a vacuum. The moves before the move have to occur in the right order and that may legit never happen short of players cooperating to create the outcome as a result of this question (players cooperating to create the outcome independent of the analysis would count). I'd love to see an analysis of the rarest move accounting for the rarity of the position but that would add an exponential amount of candidates. Hell, I'd love to see an analysis of the number of candidate moves that would
Without watching, I would say the rarest move ever is pawn promoted to another pawn, just cause the king found out that pawn was sleeping with the queen and decided to get rid of him. Top that one!
If you think English has too many syllables, think about the fact that some people have to use *megszentségteleníthetetlenségeskedéseitekért* on a regular basis.
But what if you ignore disambiguation? That might be more interesting than these technicalities. Maybe also ignore checks and checkmates if we only care about the move itself, not the effect of the move
Interesting. What about promotion ? Could we do a promoted double disambiguated capture mate with bishop ? Promoting would still allow a discovery mate. It looks like we can add another layer of rarity. Like at 9:00 bishop could come from underpromotion if we shift the diagram.
What an amazing video! Great visualizations and content! I was personally thinking of capture to underpromote to checkmate, like exf8=N#. I encountered this move when I was building a rather bad chess engine that would take all moves played from my 10.000 games on lichess and just sync the result of the game to the moves played in the game. Then, it would check for all the legal moves on the board, how well those moves did in that database, and then play the move that performed best. While building this, I also encountered the huge variety in possible chess moves, which surprised me a lot. Great video to explain these concepts and analyse this mystery, thanks for this!
Pedantry: "rfotn" isn't a possible english word because it doesn't follow english phonotactics- it has nothing to do with sonority sequencing. I mean, I'm sure phonotactics generally evolve towards falling in line with typical sonority sequencing, so they'd be related concepts, but there are plenty of english words which don't make that triangle shape, like strats and strength
But the whole disambiguating thing is sort of a quirk of notation. It’s not like that fact really means much in context. The other piece could often be swapped out and the capture or checkmate would still happen
I don’t get when you say 30 % of possible moves have never been played. So those are the rarest ones ? Not just the 4 you selected. I’m a misssing something ?
One could argue "the rarest move" is definitely one that happened only once, and not one that never happened. You would also never say "the rarest stamp in my collection is the one I don't have".
when we get a step back to the rarest move (without catogorie) i'm pretty sure it is something like: Ba1xb2# (or one other of the 4 edges) the reason is: the other bishops have to be on a3 and c1 there is no other possibility and it can't be a discovered check. As we saw in the video a discovered mate is more likly then a mate with a bishop in these situations. And: under promotion to bishop is the rarest, and since there have to be 2 black bishops, even rarer.