TCEC S14 Final, CCC5/-6

For the past several weeks I've been running a Monday series to track the evolution of two outstanding all-engine chess events that just keep going and going. When a tournament in either one finishes, another tournament starts almost immediately. Without a regular, periodic look at the competitions, they are difficult to follow. (For links and acronyms, see the 'TCEC/CCC Links' tab.at the top of each post on this blog.)

TCEC: Last week's post, TCEC/CCC Head Scratching, revealed a mismatch between the published records of TCEC S14 and the two participants in the S14 superfinal. How did Leela manage to qualify? I ended the discussion with:-

I checked a few other resources and couldn't find the source of the discrepancy.

As I write this, the two engines have played 37 games in the 100 game match with the score deadlocked at +6-6=25. Since the previous post, Chessdom.com has published two reports on circumstances leading up to the match:-

• 2019-02-04: Stockfish wins the Premier Division, Lc0 qualifies for the Superfinal
• 2019-02-06: Interview with Alexander Lyashuk about the recent success of Lc0

Neither report mentions how Leela qualified for the superfinal, so we have to turn to other sources. The Leela blog provides a hint:-

• 2019-01-16: Leela promotes to SuperFinal of TCEC! She will face Stockfish. (blog.lczero.org)

Here we find the crosstables for all S14 preliminary divisions where Leela played. This includes the mention:-

Premier division standings (Komodo MCTS results have been removed)

So we have the results counting seven engines instead of the eight shown in the TCEC archive. Why were the MCTS results removed? I'm sure the reason must be documented in several places, but I found the answer on Talkchess.com:-

• 2019-02-09: End of Era is there: SF is finally beaten! • 'My question is why LCZero is playing the superfinal, if Komodo was second in Premier Division? It should be Stockfish vs Komodo.' => 'Because the games against Komodo MCTS were cancelled at the end (according to TCEC rules, since MCTS crashed 3 times), and Lc0 benefited from that.'

CCC: As for the other ongoing competition, CCC5 was underway at the time of last week's 'Head Scratching' post. Stockfish emerged the winner, 2.5 points ahead of Lc0, which was 4.5 points ahead of third-placed Lc0-dev. Stockfish then played Lc0 in a 100 game match, which finished yesterday. The result hasn't been published yet, although the PGN is available.

A new event titled 'Battle of the Leelas' started with Stockfish, Komodo MCTS, and four Leela variants. Thanks to some new commands in CCC chat, I located a few key resources that I hadn't seen or had overlooked:-

• CCCC Club - Chess Club (chess.com)
• CCC PGN Archive (ditto)
• CCC #tournament-results (discordapp.com)

Another command ('!CCC6') informs,

Nightbot: CCC 6: The Winter Classic; Classical (standard) games with the eight best engines. Time control: 30/10 classical; Engines: 8; Stages: 1 main, 1 final between top two engines; Stage 1 format: 4x RR (escalation); Stage 1 games: 112; Stage 1 duration: 7 days; Finals games: 100

That's enough for this week's post. Next week we should have the score for the TCEC S14 superfinal through game 80, and find that CCC6 has started.