I played a bit of chess at school, up in the eaves of the maths department, but hadn’t touched it in years before recently installing the Droidfish app (powered by the Stockfish chess engine) on my phone. Trial and error - and a bit of reading to refresh rusty tactical basics - soon discovered the (still embarrassingly low) percentage strength setting to result in a fair-ish game rather than a wipeout.
But I was puzzled to find that, on occasions when I ended up in a hopefully-winning position in the endgame, Droidfish started making arbitrary, pointless moves rather than trying to make life difficult. (Puzzled, but amused when it means I can effortlessly get a bunch of pawns promoted and checkmate with three queens or the like :smile: )
(Sadly, I didn’t screenshot the win with three queens, so this is the best I have…)
Apparently, if Droidfish’s analysis concludes that it’s going to lose, then any move is as good/bad as another (an understandable, if somewhat nihilistic, view) - but its analysis assumes perfect play on my part, which it should know by now isn’t guaranteed :wink:) It might do better to find a way to fight on in the hope that I trip up at some point, which is entirely possible…
(Thanks to the question “Are chess engines too briliant to play good endgame?" and others on Chess StackExchange for insights on how chess engines work.)