Stockfish wins TCEC Season 23 (2024)

There were no major surprises in the Superfinal of the Top Chess Engine Championship, as Stockfish came out the comfortable winner – the final +17 score even surpassed most viewers’ expectations.

As strong as engines are, sometimes the human brain is still superior! Leela held out to win the opening and cut Stockfish’s lead to 1 point. There were murmurs among Leela’s fans. Could she do it?
Unfortunately forLeela’s fans, that bubble burst quickly. Stockfish won the eighth opening … then the ninth … then the twelfth … and then won some more. By the midway point, with Stockfish up+7, the only question was whether Stockfish would crush Leela as badly as it had crushed Komodo Dragon in the Season 22 superfinal. Stockfish had won by+19 then, so it was behind time, but the fact that some of more high-bias openings had still to be played gave Stockfish fans some hope. Stockfish quickly won another four openings, but then it misevaluated another blocked position to hand Leela her second won game pair. The chance was gone, and Leela emerged from the Superfinal battered, but still less battered than Komodo Dragon.

There was still the “bookmaker’s Superfinal”, between the two bookmakers of the Superfinal – GM Matthew Sadler & Jeroen Noomen. Jeroen took the early lead, but the 7th opening was GM Sadler’s, so he had the only Leela victory in the books. The two remained neck and neck as the superfinal progressed. The second Leela victory went to Jeroen, but several Stockfish wins went to GM Sadler to keep the score level. The score remained unclear all the way to the final game 100: if it ended decisive, Jeroen would win, while if it ended drawn, GM Sadler would win. Stockfish had the White pieces in a Czech Benoni, an opening that usually leads to blocked positions. Stockfish’s eval was soon above +2, but since Stockfish had misevaluated several such positions earlier in the Sufi, it was unclear if Stockfish could actually convert. For the longest time it looked like the game was just shuffling. Both Stockfish’s and Leela’s evaluations kept rising, but even the kibitzing GM Sadler could not tell what was happening. Finally, after 60 moves, Leela played 60…b5, opening lines and going down in defeat. If Leelawere human,we would say she cracked under pressure, except Stockfish also predicted Black would play 60…b5. Go figure – this mysterious game defies human understanding, and I for one am curious what Leela’s decisive mistake was.