Going Berserk

    I gave up correspondence play a couple years back because it got boring playing 99% draws, but the other day I read a rather lengthy article by a strong correspondence player in which he recommended using four engines. 

    Actually he recommended three engine, the first one is not a real engine...it’s your brain, but your brain had[s to be coupled with the right chess software. In my case, the “brain” engine is a really weak one that probably wouldn’t contribute much. 

    As for the real engines, they have different playing styles and their programming affects the types of positions in which they will achieve the best results. His preference was to use Dragon by Komodo, Berserk and for endings he uses Stockfish. 

    Berserk may not be well known. It can be downloaded (free) HERE. On the CCLR 40/15 rating list Berserk ranks behind Stockfish 20230613 64-bit 4CPU, Torch v3 64-bit 4CPU, Dragon by Komodo 3.3 64-bit 4CP and Obsidian 14.0 64-bit 4CPU. 

    Note that Stockfish is a developmental version and they are all running on 4 CPUs. Obsidian is unfamiliar to me, but it is described as playing a strategically sound style. It prefers strategy over aggressive tactics, focuses on good piece placement and controlling key squares and is particularly adept at navigating complex positions. On the CCLR list it seems to hold its own against Stockfish and Dragon by Komodo. 

    The highly aggressive Berserk often seeking out complex tactical opportunities and prioritizing attacking play over positional consideration. This supposedly makes it the most human-like. 

    It has a penchant for finding surprising and creative moves and is not afraid to take risks. It often finding deep tactics that other engines miss. In the opening Berserk can generate original and unexpected moves. 

    Really highly rated CC players use powerful dedicated computers, but most us have off the shelf laptops and that alone puts us out of the running for a high correspondence rating! Of course, you can rent computer time online, but that’s going to the extreme. 

    I wanted to compare the Weighted Error Value that the Fritz program assigns after completing a tactical analysis, so I let it analyze (at 15 seconds per move) a 17 move online game I played. The WEV’s were only a smidgen higher for Berserk compared to Stockfish, so there’s not much difference in using the two for general analysis. 

    All this piqued my interest and so I decided to play a two game correspondence match. My opponent is an International Master rated 2300+ FIDE who is also an ICCF Correspondence Master. The goal is to test Berserk and my plan is to play openings that are highly tactical. Let’s see what happens!