What do you think of the Cerebellum book? It produced sharp lines that intuitively seem to threaten something. However, Schooner failed to find the best continuation too often. In Schooner's best game against WaDuuttie, the engine quickly played the inferior Ba6.
[d]2k3r1/p2r1pb1/2pBp2p/q1P1P2B/1pb4P/1Pp5/P1Q2PP1/1K1R3R b - - 0 26
The computer calculated lines by Stockfish require "Stockfish" type depth to solve. At least, that is how I think they were created. Perhaps the lines are too dangerous for engines that have a shallower search depth - especially in blitz time control. I will try Cerebellum one more time instead using the polyglot pickbest=0 flag. In that way, it might be hoped a gentler line can be selected at random. Of course, another way is to make the engine think deeper in the same amount of time.
On-line engine blitz tourney November
Moderator: Ras
-
- Posts: 570
- Joined: Mon Jul 20, 2015 5:06 pm