Fischer Interesting Video

[Leo]

[jefk]

that 'Frenchman' was Philidor

PS the late mr Robert James F. imo was exaggerating a bit; since Philidor, chess still
evolved a bit until Steinitz (*). And as he points out, opening theory even evolved a bit more
after Morphy. Thereafter at highest levels is would be with slow games and adjournments
as during the Fischer years) imo mostly a matter of style, and avoiding mistakes (or
exploiting mistakes of your opponent, eg in endgames as Carlsen does. But gradually the
playing times of the games has increased, no more adjournments, and more stuff as rapid,
blitz and armageddon. So then it again isn't easy for humans ofcourse, as WC Ding Liren recently
shown (got mated in 2). So chess960 hasn't taken over yet (but with the more limited version
it might do so may quite soon); whereby the restricted TCEC chess version also is an option.

[towforce]

The first two comments on the video (if you open it in YouTube) are both good:

1. morphy said same: ''the ability to play chess is sign of a gentleman. the ability to play chess well is sign of wasted life."

2. Basically he is saying chess was fun when it was open for new discovery. Now it has been analyzed to death.

[smatovic]

[...]
1. morphy said same: ''the ability to play chess is sign of a gentleman. the ability to play chess well is sign of wasted life."
[...]

Apply onto chess programming

--
Srdja

[mclane]

Maybe computerchess is similar.

[towforce]

Apply onto chess programming

Maybe computerchess is similar.

The proverb goes, "Great minds think alike..." (link) - but today it's "German minds think alike".

In the two posts above, smatovic and mclane are saying almost exactly the same thing about a minute apart!

[mclane]

The situation in todays compterchess with the zillions of engines beeing >3000 ELO reminds me much on Fischers sayings about chess beeing not the game he knew anymore. Thats why he did fischer random chess.

Isn‘t todays computerchess in the same trap ?

[smatovic]

We had several posts on this:

Is chess programming now boring ???
viewtopic.php?t=83435

The Next Big Thing in Computer Chess?
viewtopic.php?t=81858

Fruit fly races on steroids?
viewtopic.php?t=83267

Viz said that modern chess programming is not about science but engineering. I think I agree, the science is done, the commerce too, but the engineering continues.

I use to say to my peer groups that computer chess is meanwhile a topic for students and seniors, computer science students who want to "cut their tooth on computer chess" and seniors who want to enjoy a challenging hobby, or alike.

I think people still can enjoy computer chess and chess programming, but the "golden days" seem gone, Deep Blue vs. Kasparov 1996/1997, Deep Fritz 10 vs. Kramnik 2006.

https://www.chessprogramming.org/Kaspar ... _Blue_1996
https://www.chessprogramming.org/Kaspar ... _Blue_1997

https://www.chessprogramming.org/Kramni ... Fritz_2006

--
Srdja

[towforce]

Viz said that modern chess programming is not about science but engineering. I think I agree, the science is done, the commerce too, but the engineering continues.

Hmmm. A good, thoughtful post, and I don't like being negative or argumentative, but... I'm going to argue!

* for me, the development of computing engines has been more about "having good ideas and refining them" than it has been about "science". There have been science papers (e.g. link), but for me it has been more about the papers being inspired by the developers pushing their engines than developers reading science papers and applying the findings to their engines

* if, in your words, "the science is done", then can you prove to me that it's not possible to consistently select good chess moves without generating a game tree?

[smatovic]

The science on game tree search is done, and on evaluation function as heuristic, knowledge, and we know plenty about neural networks too.

[...]
* if, in your words, "the science is done", then can you prove to me that it's not possible to consistently select good chess moves without generating a game tree?

Knowledge:search tradeoff, Shannon type A type B, you can replace search by knowledge, so, yes, you can come up with a search less engine, encode all what search does in an neural network or knowledge graph, no magic in need, but, knowledge:search tradeoff corresponds to space:time tradeoff. Imagine infinite memory and compute, and you can encode all chess knowledge in a neural network w/o search, or simpler, 32-men EGTB. Current neural networks are also a kind of heuristic for "the" perfect chess neural network, a perfect chess oracle, which knows every best move for every chess position. As mentioned, we are limited in memory and computation, and Wiener suggested in 1948 an in depth limited tree search with evaluation function as heuristic, we all still stick to that principle with our engines.

https://www.chessprogramming.org/Norbert_Wiener

--
Srdja