Modern programming is about testing +-0.1 Elo change 500k games w/ 10000 concurrent cores. It's all about small evolutionary steps.
Good old days was about a talented programmer writing good code in the first case. That didn't need a house full of cores. It was about creating something from nothing.
@Talkchess guys: What is pre-historic chess programming about?
Moderator: Ras
-
- Posts: 529
- Joined: Sat Mar 02, 2013 11:31 pm
-
- Posts: 4594
- Joined: Tue Apr 03, 2012 4:28 pm
- Location: Midi-Pyrénées
- Full name: Christopher Whittington
Re: @Talkchess guys: What is pre-historic chess programming about?
Low cost of entry. All wheels already done. Easy student project. Gain title “top engine programmer” quite easily. Status. Why the surprise?AndrewGrant wrote: ↑Mon Apr 29, 2024 2:04 amI also agree with this, but to a lesser extent. I complained very loudly when NNUE was introduced, because it would gut hand-crafted eval, which was the single most unique part of engines typically, as search has been homogenizing for a long time.
Now authors use the same trainers. Slightly different if not the same architectures. Sometimes even the same near copy-paste implementation. And often times, they don't generate their own data.
I was filling out this chart based on the CCC engines the other day...
![]()
Do you have a title for your spreadsheet? I propose Grey Scale of Eduardness.
-
- Posts: 56
- Joined: Fri Sep 04, 2020 10:30 pm
- Full name: Kim Kahre
Re: @Talkchess guys: What is pre-historic chess programming about?
All strong NNUE engines have strong similarities, but I don't think that's a bad thing at all. They all contribute to the development of computer chess, it really is a team effort. StockFish is the MVP but others have a lot to contribute as well. Maybe it's not as important who's 6th or 7th on some elo list as it used to be.
-
- Posts: 4594
- Joined: Tue Apr 03, 2012 4:28 pm
- Location: Midi-Pyrénées
- Full name: Christopher Whittington
Re: @Talkchess guys: What is pre-historic chess programming about?
It’s not necessarily a bad thing, it’s just a thing. My guess is that non-programmers would be quite shocked at just how similar they actually are. Curious detail is that although they’re more than close enough to be derivatives, it’s not at all clear what is the common ancestor, if any. More possible is that each starts as a common generic simple engine, stuff gets added, and gradually they converge to the same evolved thing.Koivisto wrote: ↑Mon Apr 29, 2024 3:41 pm All strong NNUE engines have strong similarities, but I don't think that's a bad thing at all. They all contribute to the development of computer chess, it really is a team effort. StockFish is the MVP but others have a lot to contribute as well. Maybe it's not as important who's 6th or 7th on some elo list as it used to be.
-
- Posts: 462
- Joined: Mon Jun 07, 2010 3:13 am
- Location: Holland, MI
- Full name: Martin W
Re: @Talkchess guys: What is pre-historic chess programming about?
Not a bad thing at all. Who doesn't want fierce competition among strong engine? It's what we were looking for when Rybka reigned supreme for so long, in a class all of its own. What was becoming true then is almost a certainty now, the analyses of these engines are virtually indistinguishable except among maybe 10 people in the world today outside of academic purposes, but maybe that is wishful thinking. All fine, since engines have been converging since the Elo race started and expected. Novelty has become a rarity but who knows how to bring it back while still making progress, unless there is not much more progress to be made.Koivisto wrote: ↑Mon Apr 29, 2024 3:41 pm All strong NNUE engines have strong similarities, but I don't think that's a bad thing at all. They all contribute to the development of computer chess, it really is a team effort. StockFish is the MVP but others have a lot to contribute as well. Maybe it's not as important who's 6th or 7th on some elo list as it used to be.
-
- Posts: 12171
- Joined: Thu Mar 09, 2006 12:57 am
- Location: Birmingham UK
- Full name: Graham Laight
Re: @Talkchess guys: What is pre-historic chess programming about?
gaard wrote: ↑Mon Apr 29, 2024 4:32 pmNot a bad thing at all. Who doesn't want fierce competition among strong engine? It's what we were looking for when Rybka reigned supreme for so long, in a class all of its own. What was becoming true then is almost a certainty now, the analyses of these engines are virtually indistinguishable except among maybe 10 people in the world today outside of academic purposes, but maybe that is wishful thinking. All fine, since engines have been converging since the Elo race started and expected. Novelty has become a rarity but who knows how to bring it back while still making progress, unless there is not much more progress to be made.
Chess engines aren't yet unbeatable at normal time controls, but we seem to be closing in on that.
This next remark really belongs in the "fruit flies on steroids" thread (which, at heart, is about what we really want from chess computers), but for me, the online engines at chess.com, which are nowhere near the elo standard of the best engines, meet all my requirements more than well enough: there's a variety of different "bots" to play against (and, of course, a huge number of human opponents!), and it will do a handy post-match analysis of your game. It does other things as well, like automatically keeping a record of all your games - which is obviously handy.
I also want to find a pattern that simplifies and explains chess - but that's out of scope for this thread.
Want to attract exceptional people? Be exceptional.