TalkChess.com

jp wrote: ↑

Fri Apr 10, 2020 2:12 pm

It may also be partly because later Leela versions play a bit less "incorrectly", i.e. unsoundly.

Nah, giving away pieces for free because you view all options as certain losses is just a pathological side-effect. It's definitely not a way of playing less incorrectly

This will be quite interesting indeed; thanks to the Komodo team and chess.com for putting on the event! This match is getting far more advance publicity than we imagined. Hundreds of comments on Facebook, chess.com etc. Most people expect a lopsided result, they just don't agree on who will be the ...

I've put way too much time into chess and enjoy it too much to spend a lot of time in other games, but in terms (complexity of play)/(simplicity of rules), I've always been even more partial to hex than go.

It has a much smaller community than either chess or go, though.

Total playouts. Thats actually full games, so a "node" means something completely different from SF's nodes. You're comparing apples to oranges here. This is simply wrong. Leela does not perform playouts of the whole game in its search. A playout means navigating from the root via the PUCT algorith...

Another important thing to keep in mind is that humans do quite a lot of computations themselves; it's just that we're consciously aware of very few of them. With engines, we're aware of all that they have to do to play high-level chess, which can make it seem like a more drastic difference than it ...

To be honest all that has changed from the origins of chess until today is the hardware. :?: So, no software improvements at all? On equal hardware, I should expect Fritz 10 to hold its own with SF 11, or, sticking with the same engine across time, SF 5 to hold its own with SF 11? It's almost as if...

On qhapaq's site the listed ELO for dolphin1/orqha1018 is 4393. AlphaZero[1] Shogi's ELO is apparently around 4400, and DeepMind's MuZero[2] apparently has a shogi ELO of about 4700. I had to eyeball those last two since I can't see an explict ELO figure given - presumably I'm overlooking something...

If we're just wanting to test the idea, it seems it might be simpler to find a difficult TB position where the conversion is ,say, 30 moves away (or some such depth), and see if SF finds the win faster with 50-move ply counter set to 0 than with it at 39, instead of trying to make absolutely sure th...

That output is expected score in hundredths of a percent, so it thinks white's best is an expected score of just under 26%

As a minor note, it seems the move order in the game updates got changed somewhere along the way (too lazy to track down exactly where).

Game started 1.g4 d5 2.Bg2, but game updates are showing 1.g4 d5 2.c4 as the beginning.