In my experience as someone that matches my engine against pretty much the same engine in much faster hardware online, this is a losing strategy. Sometimes my opponent has a fail high and my engine takes another 6 moves before having a fail low. By then there's no saving move because the blunder was played ages ago, so taking time trying to find a saving move is suicide.chrisw wrote: ↑Sun Apr 05, 2020 11:18 am Not sure if it still does, but when I used to take a close interest in what other engines did, Hiarcs was noticeable for sometimes taking a humongous amount of time (I’m talking 90% of total time left for the entire game) on one move. I think I saw a quote from Uniacke to the effect that if a score fell, the engine would then do everything it could to try (eg extend the search) and find a better root move, even at the cost of playing the rest of the game with way less time. Makes sense, if a move is really bad, you lose whatever, so best to avoid if possible.
I've been able to save -2.00 disadvantage positions against people out-searching me by 10ply, and that would have been impossible without enough time in the clock. If the opponent is superior (only superior opponents cause you a fail low) and you haven't made a decisive blunder yet (if you have what you do makes no difference) you better save enough time on the clock to play the rest of the game, because without it you're going to fail to draw a drawn position because of the rate needed to play your moves (0.00 positions that are lost by the clock.)