[systemvoltage]

Chess engines should come with another metric bar: "The twitchy-ness" of the position aka the gradient of primary eval metric as you pareto the possible moves from best to worst. The stronger this gradient, the more risky it is to play, and more changes to make a mistake.

[staunton]

This ignores the question how hard it is for a human to find the best (or a "good enough") move. It's easy to find games with 10 "only move" 's in a row where even a beginner could easily have played all if them.

[__s]

Sure, but it's a start on adding nuance to eval beyond minmax

[Sesse__]

This is not a new request; many people, including engine authors, have suggested it throughout the years. The problem is that it's seemingly very hard to reliably quantify something like this and propagate it throughout the game tree.

[rocho]

You don't need to propagate it, you just need to show the gradient of the current position alongside with the classical evaluation, to give more context to the viewers.

[JoshTriplett]

https://news.ycombinator.com/item?id=42401469

[lubesGordi]

Agreed. I always thought of it as 'how close to the cliff edge are you' metric. It'd probably be easy to do, look at all the possible moves and add up the resultant evals. If you're currently tied but you have only one good move to keep it tie while the rest of your moves give mate in 1, well, saying the board is tied is not helpful.

[mlyle]

Except a lot of the time there's an obvious threat that needs to be responded to, and a couple of obvious good responses that even terrible players spot.

[8note]

engines arent great at that. they spot the beat move, and if you dont do it, it keeps spotting that same great move until your opponent notices it.