What counts as optimal play? Couldn't the winner could be either like in regular chess?

Optimal play is still defined in regular chess, it’s just computationally intractable.

Is it basically asking whether there is a possible strategy that is impossible to defeat?

Yes

I think that this market has the same basic idea as another market but oddly enough they seem to be at very different percentages

https://manifold.markets/IsaacKing/is-the-wolf-and-sheep-chess-variant

@Qoiuoiuoiu Yup; they seem like identical markets. I'm pretty convinced that white wins.

@gabrieldwu yeah same... I messed around on https://fairyground.vercel.app/advanced.html but I didn't really check all that well. I think it's possible to set up a proof if you let it run long enough

My intuition is that this is an easy win for Black. For a pawn to promote, it needs 5 moves in a row - and it needs to be protected some of the time, or the queen can almost always take it within 2 moves. Even if protected, that only changes the queen moves needed to 4 (or rarely 5) per 2 pawns, unless the protecting pawn is itself protected - which is foreseeable and easily preventable. Meanwhile, a queen on rank 2 can gobble up one pawn per move while also forking if possible, or move elsewhere for a guaranteed fork. White's only threat that takes away the queen's freedom of movement is an already-protected pawn on rank 6 or higher.

Very tentative evidence in favour of white: I (not a good chess player) just played a game against myself and white narrowly won.

@AngolaMaldives I think I remember searching this up and there was a stackoverflow post about how Black wins (which is why I bet the market down). If someone finds that and posts it in the comments I'll resolve the market.