Hi parmetd,

<It finds some interesting ideas but it's so bad at analysis that you have to thoroughly check everything with another engine anyways.>

That actually sounds quite interesting. Can you give an example.

This Leela Zero is useless to me, too slow in evaluations on weaker GPU, basically complete crap compared to Stockfish. It is really not worth bothering installing.

On the same ground, stockfish beating Commodore64 chess is absurd given how much hardware stockfish uses, right?

Very interesting, thank you.

Hi K.P.

I am frustrate at such a powerful machine being wasted to prove or disprove the soundness of a sac in an OTB masterpiece when the computer cannot possibly have all the details such as clock times, tournament standings, previous history...etc..etc.

When I first got Fritz 6 I used it to help compose studies and mates in two problems.

It's solving speed was invaluable, as is a T.B. site and both often came up with a sneak shot then sends you off in a new direction, human, computer and chess having fun creating together.

If a machine like Stockfish had a 'show us the worse move' feature, show the variation that allows a mate in x moves no matter how long 'x' is or a 'every piece is a desperado' mode (it's how I play) and a sac showing it's highest evaluation even if it turns a, say 3.69 to -3.69.

You would get a lot of merry nonsense but a few 'wows' would be in there.

At the moment it being used as a glorified 'spell-checker', I want it writing poetry.

I wouldn't call it absurd, I would call it "interesting" and "inconclusive", just to be kind. Besides, you can run Comodore64 chess programs on a PC in emulation mode, several of them exists. See, for example, https://www.myabandonware.com/game/... and http://www.spacious-mind.com/html/c.... I would not have any doubt what the results of a match with Stockfish against the latter, with an Elo rating of 1650 – 1750 would be. So it can be made to happen if you want to.

You could achieve similar results by giving Stockfish a time advantage against AlphaZero in a match. Since I estimated an 80X advantage in computational performance capability of the hardware used by AlphaZero compared to the hardware used by Stockfish in their 2018 match and make the contest approximately even. That means giving Stockfish 120 mins/40 moves with 60 mins/20 moves thereafter and 1.5 mins/40 moves with 0.75 mins/20 moves for AlphaZero.

Given the results that DeepMind saw when they did something similar (https://deepmind.com/documents/260/...) when Stockfish achieved a 67% scoring percentage when AlphaZero's time disadvantage was 1:100 and a 57% scoring percentage when AlphaZero's time disadvantage was 1:30 I would also have no doubt what the results of a Stockfish vs. AlphaZero match would be when AlphaZero was given a time disadvantage of 1:80.

But that would be absurd, wouldn't it?>

<I wouldn't call it absurd, I would call it "interesting" and "inconclusive", just to be kind.>

Hi J.B.

I'll explain, I am now techno savvy.

The debate started with: 'apparently' some chess programs play better the better the hardware.

(a bit like saying Fischer played better because he wore a suit instead of jeans and a T.)

This was put forward by the Stockfish fan bots after Alpha scudded it. (Personally I think it was the sea air.)

You join the discussion at the part where AylerKupp, to prove a point, wants Stockfish to play a 100 game match v a Commodore 64 with the '64 being connected to NASA.

https://cd.tcecbeta.club/archive.ht...

Since tcec hardware less than half of alphazero hardware. It is almost equal hardware.

Leela is stronger than rybka 4 while running on tcec cpu. Upper bound not established.

Some leela cpu results here

https://cd.tcecbeta.club/archive.ht...

I'll explain, I am now techno savvy. ...>

what?????? you went to the dark side of the force?

Sorry, I was raised by my grandma and she told me "Joerg, in the end it is always the man that counts." why, in SF (science fiction not Stockfish but that is subject to another epic discussion) are the dramatic decisions always in an old fashioned one-on-one contest and the light sabers play the smallest role? John Wayne did solve problems this way a 100 years ago.

But give Achilles a light saber and, while he would still be Achilles, he will be far more destructive than if he had an ordinary saber. And give Darth Vader a pocket knife and, while he would still be Darth Vader, he will be far less destructive than if he had a light saber.

Or, to put it a different way. If you could clone Achilles (Achilles 1 with a light saber, Achilles 2 with an ordinary saber), do you have any doubts that Achilles 1 would probably defeat Achilles 2 even though they would both still be Achilles? And if you could clone Darth Vader (Darth Vader 1 with a pocket knife, Darth Vader 2 with a light saber), do you have any doubts that Darth Vader 2 would probably defeat Darth Vader 1 even though they would still both be Darth Vader?