Fairly difficult EG puzzle

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
jp
Posts: 1331
Joined: Mon Apr 23, 2018 5:54 am

Re: Fairly difficult EG puzzle

Post by jp » Fri Apr 03, 2020 11:19 pm

acase wrote:
Fri Apr 03, 2020 10:58 pm
jp wrote:
Fri Apr 03, 2020 9:40 pm
It remains true that Sting claiming +1.29 while giving a drawing line at depth 28 is wrong, though.
and it remains true that it still found the key move (Nd5) at the 7 second mark whether the follow up was absolutely correct or not.
I don't know why anyone would place significance on an engine stumbling on to a correct first move at time t but showing it had no clue about the correct solution at that time.

As far as we can tell from Sting's output, we don't even know if it worked out the correct ... cxd5 line at depth 30 or more. It could have switched simply because it misevaluated its drawing line to be + 1.74 at depth 30, so it chose the +1.73 line for Black instead. (I'm not saying that happened, but if it did, the output would look no different.)

acase
Posts: 728
Joined: Wed Mar 15, 2006 12:14 am
Location: Columbus, Ohio USA
Full name: Andrew R Case

Re: Fairly difficult EG puzzle

Post by acase » Fri Apr 03, 2020 11:30 pm

Well, I don't think it "stumbled" upon the correct move, it clearly knew something must have been there to pick that move instead of it's original 1.Bb1?, and if you had played the moves out against Sting in the 1.Nd5 cxd5 2.exd5 Be8 line it certainly would have played a different move other than 3.g4? namely 3.Bg6!.

Post Reply