Quote:
Originally Posted by Noir_Desir
My firebird 1.1 engine took less than one minute to give Qe5 as +1.8 for white, haven't checked the line but it's certainly quicker than me.
So it's an interesting position, but proves nothing with respect to engines and humans.
+1.8 is also way off. Black is dead lost here, though maybe I should rephrase my original point here since I definitely did not realize that while playing.
I don't expect a human to have some 20 move tree of variations in his head with all variations concluding in mate or a decisive edge in material. I think humans of almost any strength can see black has no chances of winning here. White has a million different perpetuals at worst. So then it comes down to draw or white win. Well white is threatening: 1. mate on the dark squares. 2. create passer on in the kingside (Qe5/Qg7+/Qh8+/Qxh7+/etc) 3. to simply mop up the loose light bishop. And Kh3 defends very well against all of black's perpetual threats for at least two tempi.
Can black defend all those threats and somehow manage to eventually pull out the perpetual? Maybe but it seems extremely unlikely. At best black has to defend a million different threats and pray to find a draw. In all reality his position will be incredibly difficult to defend. As it turns out it will be impossible to defend but I don't think that is nearly as relevant as being able to accurately assess the approximate probability of results which clearly leads to this position being +-.
So I simply find it interesting here that the computer wants to assess it as a draw for quite a while, though it is a forced win. And most strong (I don't know, FM+?) humans would almost certainly very quickly assess it as winning for white with some remote chance of black managing a draw. In other words, a position where human intuition trumps computer calculation.