DeveloperCreated Friday, April 21st 2017 @ 11:20:42
We are currently working on many exciting new games for Riddles.IO, and one of the ideas we have is to add a Chess competition.
Chess AI engines have been around since 1950, and it is time to experiment with this on Riddles.IO as well, and to do this we'd like to hear your opinion about the following.
For bot-to-engine communication in most games we have used a coordinate-based system where 0,0 is the top left field, and 9,9 is the bottom right field on a 10 field wide board. This system is easy to parse for bots, but we have also received feedback that for traditional games people want to stick with traditional notation.
Our question is, should we use a notation like 'e2e4' or 'e7e8q' for chess like it's done in GNU Chess? Or should we use a coordinate based notation like '1,1;2,1:9' or '4,1;5,1' where the top row is row 0?
- Updated Friday, April 21st 2017 @ 12:08:40
On the question : i would prefer the human friendly notation ("e2e4" for instance), like most existing protocols (such as winboard) do.
Have you a new chess variant in mind ;-) ?
If not, I am not sure i really see the point of doing a regular chess competition with finals and prizes whereas there is a very active scene already (cf http://www.computerchess.org.uk/ccrl/4040/, talkchess.com) with established communication protocols. Moreover one of the 2 best programs in the world is open source (https://stockfishchess.org/), incredibly strong at all time controls... I would personnally be interested in new games or ones that have never been really tackled by AI programmers. My 2 cents...
- Created Friday, April 21st 2017 @ 21:22:08
will you do c++ starter bot?
- Created Saturday, April 22nd 2017 @ 13:25:28
For the same reasons as NotABug, I personally am not interested in a Chess competition. It would be much preferred to have a competition on a game that has never had much AI interest, or even better: a completely new game.
If you do decide to make a new game, it has to be deterministic unlike the riddles.io competition. Accurately evaluating bots in a non-deterministic game takes WAY more games.
Also the tournament format has to be communicated clearly from the start since this alone may decide if I want to participate or not. AlexShulga already told you all about that though ;)
DeveloperCreated Monday, April 24th 2017 @ 09:03:07
Don't worry guys, we already have a lot of new and original games in the pipeline. But we want to include some classic games as well, which will not be competitive (i.e. no finals/prizes).
The reason Joost asked the question is because we're discussing whether we want to use consistent notation across the games on the website, which might be easier to parse, or use the standard notation for chess.
- Updated Monday, April 24th 2017 @ 12:56:13
First of all, thank you guys for your work, theaigames.com and now riddles.io are really amazing!
I am totally agree with others. It seems a chess is the worst game in the world to create a new competition in it:)
1) It is too hard for beginners to write a valid chess bot from scratch (unlike 4-in-a-row, for example)
2) It is extremely not interesting to write your own bot for advanced since we all know about extremely strong open source engines as others already mentioned
It seems you guys already decided to do this and not gonna listen to the community opinion. Okay, but we are all exciting about a new cool competitions you are working on and hearing that you are spending time on chess is a little bit dissapointing :)
- Created Saturday, April 29th 2017 @ 19:47:46
I am sorry to have to say this, but I also don't think that a Chess competition is a good idea. As several others have already noted, there are already chess engines galore and there is just no chance of improving anything on them in the scope of a competition like this. I foresee lots of people copying either parts of some existing evaluation function or whole sections from the existing games, making it more a competition in the skill of hiding your copying than in games programming.
I agree wholeheartedly with the preference for lesser known or even unexplored games!
DeveloperCreated Saturday, April 29th 2017 @ 23:21:57
We decided to axe the chess game based on your feedback.
Having said that, I'd like to re-raise the question @Joostotron asked on a more fundamental level: If we were to offer a competition based on an existing game in the future and this game has an existing notation for moves, would you prefer we adhere to the existing notation or would you prefer we'd use a notation which is similar to the other games offered on theaigames.com/riddles.io?
Thanks in advance for your input!
- Created Sunday, April 30th 2017 @ 12:32:24
Nice to hear :)
I would prefer a standart theaigames.com/riddles.io notation for an abstract existing game (but it can depend on the game). And it will be really cool to see a new games like a HackMan or at least slightly modified existing games!
- Created Thursday, May 4th 2017 @ 21:18:52
Regarding the question about notation: I don't care so much. My absolute preference would be the Game Text Protocol, which is an extension of the Go Text Protocol. That would allow me to run test suites easily and other nice features.
But other than that it doesn't really matter. The notation based on figures is easy to parse and that is a good thing.