MCEC anyone?

[Daniel Anulliero]

We already have the Graham Banks divisions (from 1 to 8) , live games , broadcasted with the TLCV .
He also play a lot of interrestings ans funs tournaments .
The most interresting series for me .
Chesswar by olivier deville was great too , agree with HG Muller , pity , its gone ...

[mar]

We already have the Graham Banks divisions (from 1 to 8) , live games , broadcasted with the TLCV .
He also play a lot of interrestings ans funs tournaments .
The most interresting series for me .
Chesswar by olivier deville was great too , agree with HG Muller , pity , its gone ...

I agree, ChessWar was awesome (I think there was also OpenWar IIRC)

[chrisw]

I'd suggest:
- No derivatives
- No clones
- No NNUE

That list of rules seems good, but I have one issue with it. NNUE doesn't necessarily make an engine 3000+ elo, SF NNUE does that. Therefore I think you're excluding too many developers (like me) who are developing their own network implementations and training their own nets. I think an improvement could be:
- No derivatives
- No clones
- No SF NNUE evaluations. Neither the nets themselves nor the implementation.

Neural networks, or at least a hybrid between them and HCE's, are the future of evaluation functions, and I think that should at least to some extend be acknowledged.

I’m inclined to agree. Messing around with NN and if “NNUE” that does not necessarily mean an SF net or anything like it, it just means the Net is being efficiently updated by makemove() or whatever. The EU part has been known as a concept since at least 2003, it’s not unique nor new to neither SF nor Nodchip nor shogi.
Looks like punishment of genuine developers because mass of SF clones

Does this tournament format require upload of engine or can each engine connect remotely?

[amanjpro]

I'd suggest:
- No derivatives
- No clones
- No NNUE

That list of rules seems good, but I have one issue with it. NNUE doesn't necessarily make an engine 3000+ elo, SF NNUE does that. Therefore I think you're excluding too many developers (like me) who are developing their own network implementations and training their own nets. I think an improvement could be:
- No derivatives
- No clones
- No SF NNUE evaluations. Neither the nets themselves nor the implementation.

Neural networks, or at least a hybrid between them and HCE's, are the future of evaluation functions, and I think that should at least to some extend be acknowledged.

I agree with you, I reformulate the NNUE though to exclude any non-original NNEU libraries. I mean, one can take it from Igel for example, which is as bad as taking it from SF

[amanjpro]

I'd suggest:
- No derivatives
- No clones
- No NNUE

That list of rules seems good, but I have one issue with it. NNUE doesn't necessarily make an engine 3000+ elo, SF NNUE does that. Therefore I think you're excluding too many developers (like me) who are developing their own network implementations and training their own nets. I think an improvement could be:
- No derivatives
- No clones
- No SF NNUE evaluations. Neither the nets themselves nor the implementation.

Neural networks, or at least a hybrid between them and HCE's, are the future of evaluation functions, and I think that should at least to some extend be acknowledged.

I’m inclined to agree. Messing around with NN and if “NNUE” that does not necessarily mean an SF net or anything like it, it just means the Net is being efficiently updated by makemove() or whatever. The EU part has been known as a concept since at least 2003, it’s not unique nor new to neither SF nor Nodchip nor shogi.
Looks like punishment of genuine developers because mass of SF clones

Does this tournament format require upload of engine or can each engine connect remotely?

For a fair competition, I would say all the engines should run on a similar hardware, probably one game at a time, so we don't need a beefy hardware

[amanjpro]

We already have the Graham Banks divisions (from 1 to 8) , live games , broadcasted with the TLCV .
He also play a lot of interrestings ans funs tournaments .
The most interresting series for me .
Chesswar by olivier deville was great too , agree with HG Muller , pity , its gone ...

That is true, and much appreciated. But unfortunately not every engine makes the cut for Graham's tournaments (Zahak/Rustic for example are too low rated to cut it)

[mvanthoor]

That is true, and much appreciated. But unfortunately not every engine makes the cut for Graham's tournaments (Zahak/Rustic for example are too low rated to cut it)

At some point, we will both make the cut to those tournaments as well; you will probably be earlier than mine.

With regard to Nils: you're right: excluding all NNUE and Neural Networks is a bit too harsh. In the end, I'll probably start to experiment with them myself.

With regard to using an implementation, such as Daniel Shawul's NNUE library: I'm not against that. Actually implementing how to load and use a neural network doesn't seem to be very interesting. Nobody complains that the entire world uses the Fathom syzygy endgame table base implementation. Nobody requires you to implement your own EGTB module.

I think the reason is because the EGTB module has very little to do with actually playing chess. You just load the EGTB, and search it using the library. It's exactly the same for any engine.

When you load a neural network file and use it, that is not the real achievement; the real chess-playing achievement is training the neural network on the engine's own games, using your own trainer. Same as the real achievement for tuning the evaluation is using your own tuner, preferably on your own games.

[Rebel]

I have a host with infinite bandwidth, that I can use to serve the website. If one of us has a spare desktop that he can use to run the tournaments, then we can start with that. The desktop runs the games, pushes the pgn to my host, we run pgn4web on the host and viola we will have a basic tourney

Yes, that's the idea, pgn4web solves all your problems. I have it working here.

http://rebel13.nl/pgn4web-3.05/live-test.html

In case you are serious with this initiative you can use my script, what you need is a server with ftp access and Arena.

[amanjpro]

I have a host with infinite bandwidth, that I can use to serve the website. If one of us has a spare desktop that he can use to run the tournaments, then we can start with that. The desktop runs the games, pushes the pgn to my host, we run pgn4web on the host and viola we will have a basic tourney

Yes, that's the idea, pgn4web solves all your problems. I have it working here.

http://rebel13.nl/pgn4web-3.05/live-test.html

In case you are serious with this initiative you can use my script, what you need is a server with ftp access and Arena.

I'm serious, I don't have a spare desktop at the moment, I also won't be installing windows if I get one, so yeah those are my current concerns at the moment, getting a cheap desktop and having enough engines that run on Linux, unfortunately that puts C# engines at disadvantage.

And I'll definitely consult you for technical details

[lithander]

having enough engines that run on Linux, unfortunately that puts C# engines at disadvantage.

Why is that? C# engines could just as well target .Net Core instead of whatever Windows-exclusive .Net Framework they are currently using and become instantly compatible with Windows, Mac, Linux and even the Raspberry Pi. You don't even need these OS's to provide releases for them. Just use crosscompile deployment from Visual Studio on Windows with just a click.

(let me know if my Linux compiles don't work for you. Never tested it because I don't have linux^^)