The real truly challenge we’d all like to see at the TCEC

[Kanizsa]

After the TCEC Season 26 superfinal, which witnessed Stockfish’s absolute dominance over Leela, we are now witnessing a rather dull event: Fischer Random championship, that I believe interests very few. (IMHO it seems to me very boring, an absolute waste of time and electricity)

I propose to the TCEC website owners an intriguing bonus event—one that could truly captivate chess enthusiasts.
Let’s organize the traditional kibitzer bonus in the form of a quadrangular match involving four participants:
1 Stockfish from Superfinal TCEC Season 25
2 Stockfish from Superfinal TCEC Season 26
3 Leela BT3 from Superfinal TCEC Season 25
and
4 Leela BT4 from Superfinal TCEC Season 26.

The time control would be 30 minutes per game, using the Noomen TCEC 26 opening book. We’ll have a total of 100 games with a round-trip format.

I believe such a challenge could provide a precise insight into the development and progress of chess engines over the past year.

[jhonnold]

Boring.

[towforce]

How about...

Normal chess, but:

1. very fast time controls

2. low specification computers

That would push back the death-by-draw issue a few more years!

[Graham Banks]

Boring.

Agreed.

[chesskobra]

They could do competitions among networks. For example, people would create their own networks for the same version of Stockfish or Lc0, and enter their networks in a competition. There could be a competition of Lc0 nets, and another for SF nets. Of course, I am not suggesting this as an alternative to a competition among engines. But just an interesting idea that would motivate wider community to participate. I am currently testing different Lc0 networks to determine which network works on my laptop best, which networks play endgame best, etc.

[Kanizsa]

They could do competitions among networks. For example, people would create their own networks for the same version of Stockfish or Lc0, and enter their networks in a competition. There could be a competition of Lc0 nets, and another for SF nets. Of course, I am not suggesting this as an alternative to a competition among engines. But just an interesting idea that would motivate wider community to participate. I am currently testing different Lc0 networks to determine which network works on my laptop best, which networks play endgame best, etc.

Maybe I’m wrong, but it seems to me that authors of programs other than SF and Lc0 would prefer competitions that give their creations visibility, even if the gameplay conditions, like Fisher Random Chess, are uninteresting for spectators. I am more focused on the development and progress of chess engines over the past year, particularly those that fearlessly measure themselves against strong opponents.