Most of the people involved on this thread, such as Daniel Shawul, the Koivisto authors, Andrew Grant, and dkappe are already on the Leela, Stockfish, Openbench, and TCEC discords already, and the same arguments and trolling happen over there as well especially at TCEC. dkappe in particular was banned from the Leela discord for his trolling, but he is still active in the TCEC discord. The Leela and Stockfish discords have their own collection of trolls such as usernameispassword and ARB who clog up the discords with their nonsense.connor_mcmonigle wrote: ↑Sat Jul 10, 2021 7:20 pmIt's unfortunate that the development of top engines isn't more visible on public forums such as TalkChess. However, if anyone is curious as to why this is, they need look no further than this thread. This wasn't always the case, but TalkChess as of today is proliferated with trolls and low quality posts which just kill the signal to noise ratio.mvanthoor wrote: ↑Sat Jul 10, 2021 7:03 pmIt's like a group of scientists, exchanging ideas and best practices, eventually doing awesome things which turns the world in their field upside down, and then they all die in a car crash while driving to get a hamburger... and everybody else is asking themselves: "We wonder how they did it." That's just a shame.Madeleine Birchfield wrote: ↑Sat Jul 10, 2021 6:54 pm Unfortunately have to agree with this. The cutting edge of using neural networks in search heuristics such as pruning, and in time management is in the Leela discord, while hardly anybody on Talkchess seems to be aware of neural networks as anything other than a supplement or replacement for their hce.
Koivisto 5.0
Moderators: hgm, Rebel, chrisw
-
- Posts: 512
- Joined: Tue Sep 29, 2020 4:29 pm
- Location: Dublin, Ireland
- Full name: Madeleine Birchfield
Re: Koivisto 5.0
-
- Posts: 50
- Joined: Fri Sep 04, 2020 10:30 pm
- Full name: Kim Kahre
Re: Koivisto 5.0
I've never seen any trolling in the OpenBench discord, it's a wonderfull place to discuss chess engine related ideas.Madeleine Birchfield wrote: ↑Sat Jul 10, 2021 7:38 pmMost of the people involved on this thread, such as Daniel Shawul, the Koivisto authors, Andrew Grant, and dkappe are already on the Leela, Stockfish, Openbench, and TCEC discords already, and the same arguments and trolling happen over there as well especially at TCEC. dkappe in particular was banned from the Leela discord for his trolling, but he is still active in the TCEC discord. The Leela and Stockfish discords have their own collection of trolls such as usernameispassword and ARB who clog up the discords with their nonsense.connor_mcmonigle wrote: ↑Sat Jul 10, 2021 7:20 pmIt's unfortunate that the development of top engines isn't more visible on public forums such as TalkChess. However, if anyone is curious as to why this is, they need look no further than this thread. This wasn't always the case, but TalkChess as of today is proliferated with trolls and low quality posts which just kill the signal to noise ratio.mvanthoor wrote: ↑Sat Jul 10, 2021 7:03 pmIt's like a group of scientists, exchanging ideas and best practices, eventually doing awesome things which turns the world in their field upside down, and then they all die in a car crash while driving to get a hamburger... and everybody else is asking themselves: "We wonder how they did it." That's just a shame.Madeleine Birchfield wrote: ↑Sat Jul 10, 2021 6:54 pm Unfortunately have to agree with this. The cutting edge of using neural networks in search heuristics such as pruning, and in time management is in the Leela discord, while hardly anybody on Talkchess seems to be aware of neural networks as anything other than a supplement or replacement for their hce.
-
- Posts: 1784
- Joined: Wed Jul 03, 2019 4:42 pm
- Location: Netherlands
- Full name: Marcel Vanthoor
Re: Koivisto 5.0
I found that I can avoid most of the crap by just avoiding the engine origins threads and forum. Personally, I just don't care too much about it. If I don't like an engine, I just ignore it.connor_mcmonigle wrote: ↑Sat Jul 10, 2021 7:20 pm It's unfortunate that the development of top engines isn't more visible on public forums such as TalkChess. However, if anyone is curious as to why this is, they need look no further than this thread. This wasn't always the case, but TalkChess as of today is proliferated with trolls and low quality posts which just kill the signal to noise ratio.
With regard to this thread...
The first half of the release post started with a rant like "We don't like neural networks in chess engines, because most people who use them don't understand them. We use a hand-crafted Real Men's Evaluation." Second part of the post expounds about the authors designing their own neural network, trainer, and data set and using that in their engine.
That just rubs me the wrong way.
It basically states "YOU use neural networks without understanding them. YOU are the plebs. Even though we don't like neural networks because they're used by the unknowing people, we're gonna show you how it's done. WE are awesome, because WE first created a hand-crafted evaluation, then generated our own data, then wrote a tuner and trained a network."
Like Daniël Shawul said: it's arrogant. There were people before them that did that as well. There will be people after them doing it again.
Yes, there will also be people who just copy and paste stuff. There will also be people who use other's ideas without giving credit. I actually have my suspicions here and there. If they're at some point confirmed however, I'll just not pay any attention to those people and engines anymore.
There are also people, me among them, who write everything from scratch and with some concepts, use other people's work to "bootstrap" the engine to see what's possible (such as using another engine's tapered PST's). Then they study each subject, and implement their own version from scratch. I even document my stuff and explain how it works (at least, I've made a start with that), including tests between versions. That's the reason why my engine is taking 2 years to get to the point where it has a tapered evaluation and can finally be Texel tuned, instead of 2 or 3 months.
You can't just make a broad sweep where you shove everyone into the corner stating that they don't understand what they're doing and then take the middle of the room shouting how awesome your work is. It's bound to cause irritation.
I found that out myself when I once said that "many didactic engines" weren't really "didactic" at all, because apart from a piece of code, there was no information. I obviously got panned immediately, because there ARE engines that do explain concepts with comments and accompanying websites and video's. (Obviously I knew that, but I was frustrated by the fact that I found so many unstable or incomplete engines that claimed to be "didactic" by just providing a 7 year old GitHub repository, when searching for a test partner for my own engine.)
-
- Posts: 512
- Joined: Tue Sep 29, 2020 4:29 pm
- Location: Dublin, Ireland
- Full name: Madeleine Birchfield
Re: Koivisto 5.0
It's a good place to talk about chess engine ideas until Andrew Grant, noobpwnftw, and Terje decide to turn Openbench into a World of Warcraft themed discord.
Openbench discord was bad around the 2020 US election, about a third of the content was election related and another third was World of Warcraft related. I'm not sure how much of the comments made on the election and American politics were serious or trolling, but there was a lot more of that stuff on Openbench compared to other discords. And there is no off-topic section so the chess engine stuff, election stuff, and WoW stuff all got mixed together, making it hard to follow conversations. It got better after Biden entered the White House and people stopped talking about American politics on the Openbench discord.
-
- Posts: 347
- Joined: Tue Nov 19, 2019 4:34 am
- Location: https://github.com/TerjeKir/weiss
- Full name: Terje Kirstihagen
Re: Koivisto 5.0
Oh woe is me, the small community discord I joined included discussion about non-chess topics the core members cared about.Madeleine Birchfield wrote: ↑Sat Jul 10, 2021 7:56 pmIt's a good place to talk about chess engine ideas until Andrew Grant, noobpwnftw, and Terje decide to turn Openbench into a World of Warcraft themed discord.
Openbench discord was bad around the 2020 US election, about a third of the content was election related and another third was World of Warcraft related. I'm not sure how much of the comments made on the election and American politics were serious or trolling, but there was a lot more of that stuff on Openbench compared to other discords. And there is no off-topic section so the chess engine stuff, election stuff, and WoW stuff all got mixed together, making it hard to follow conversations. It got better after Biden entered the White House and people stopped talking about American politics on the Openbench discord.
-
- Posts: 512
- Joined: Tue Sep 29, 2020 4:29 pm
- Location: Dublin, Ireland
- Full name: Madeleine Birchfield
Re: Koivisto 5.0
i just wish Openbench had an off-topic section where the non-chess topics could be discussed without interfering with computer chess discussions. I've seen multiple times that questions by members of the discord regarding patches and tests on the main Openbench server or ideas regarding possible search heuristics get ignored because everyone else was busy talking about WoW.