Forum

I think that people will like a match between Stockfish and Rybka (Fritz 16) These engines are two strong engines, one is the actual champion and the other is an old one.

Comment with your ideas here, and we'll submit them to chess.com! 1. Stockfish vs The Rest - A match, not a tournament. Stockfish plays one side (or Lc0 if it ever catches up and becomes stronger). The other side is played by a collection of other strong engines; whichever move is selected by the largest number of engines will be played; in a tie between two or more engines, the average rating of the engines selecting the moves will be the tiebreak. 2. A series of matches. Start with the two lowest rated engines in the tournament; they play a match, the loser is eliminated, and the winner advances. The winner then plays the third lowest rated engine in the tournament, and the loser is eliminated, and the winner advances to play the next engine. Repeat until the top engine has also played. This could be used as a way to make a tournament with a large number of engines and slow time controls finish in a reasonable amount of time. 3. Alternative to #2. Instead of matches of two engines, do mini round-robin tournaments with four engines. The bottom two are eliminated, the top two advance, and the next two are added for the next round. Any more ideas?

It will be interesting to see Gaviota in CCCC3.

Komodo dismantled Ethereal.

Me and a few members of the twitch chat decided to create a survey of which opening you like/don't like or don't know about for stage 3 now the opening book has around 100 openings then 50 were selected from there so all the openings in the survey will not include all the openings in the survey and keep in mind that each engine gets to play black and white in each selected opening. Survey is found below; results will be posted later. https://docs.google.com/forms/d/1faIuvDoO0EiTz6uCgZlAKlfWRXgjphRQaWtiEC0xbv8/viewform?edit_requested=true Your friendly CCC volunteer moderator ENcyndaquil-pikachu/systiem

Crafty is a TB engine, but tablebases are off for Crafty. This might be a mistake. Crafty's description describes Cray "Blit", not "Blitz." Finally, the description for Equinox is, while sort of funny at first glance, really messed up given that the creator of Equinox died of cancer. "An equinox is when the earth's equator passes through the center of the sun. Other than that, we have no idea why this was named Equinox. Maybe it's named after the fancy gyms in America..." Surely this could be corrected?

I submitted a bug report for this but when clicking on Xiphos the wrong flag shows up in the profile and on stream because it shows as an American flag when it was made in Serbia. I would have a picture of it but that's not working for some reason.

I'm assuming this is planned to be a regular event on an annual/semiannual/something basis? I don't know about the rest of you, but I don't see a whole lot of point to watching Stockfish and Komodo beat up on Crafty and Nirvana. There are a lot of games that are pretty one-sided. But those lower-rated engines are still inhuman monsters and can play some very competitive games against each other. I don't see any reason to shut them out entirely. There are also a lot of mid-table battles that might make interesting games as standalone games, but aren't really relevant to the tournament as a whole, because there's no meaningful difference between finishing 11th or 12th. It might be interesting to split the tournament in two and have an upper tier and a lower tier. 12 engines in each division. After the tournaments are complete, the top two in the lower division are promoted for next time round, while the bottom two in the top division get relegated. To add new engines in later tournaments, you could either increase the size of the lower division, or relegate the bottom two engines and have a qualification tournament. Or, if you want to add a large number of engines, add a third division. And you could still have a second and third playoff round following the current format (or something similar) to determine the overall winner. This would accomplish several things: 1. Every game would be move competitive. We would no longer have the big three beating up on weaker engines. 2. The games that are currently meaningless mid-table battles would instead be part of a competitive promotion/relegation battle. This would create an extra incentive for engine developers to attempt to strengthen their engines a little bit to win promotion or avoid relegation. 3. Instead of a double round robin, you could have a quadruple round robin in each division, while having slightly fewer total games. Currently there are 23*24=552 games (you don't have to divide that by two because it's a double round robin). If there are 12 engines in each division, a quadruple round robin would have 11*12*2=264 games in each division, for a total of 528 games - 24 fewer than the current format. So in terms of CPU time, this would be slightly cheaper to run while having each engine play twice as many meaningful games against competitors closer to its own level. There is one potential drawback. New engines that show up on the scene (like Leela) will have to work their way up through the lower division before they can compete against the top engines. But if these tournaments are run somewhat regularly, that's not a big deal. Most new engines that show up aren't top 10 engines when they're first released, so having to start in the lower division and win promotion isn't a bad thing.

I was watching the stream during a Alfilx64 game and some people are wondering if it is function correctly or not like if it bugged/configured properly or only running on one core when (i believe) it can support up to 4 cores. -ENcyndaquil-pikachu/systiem a volunteer chat moderator

I noticed some of the engines are having drastic eval changes and wanted to know why it's happing. I've seen it the most severe with crafty which goes to a drawn position to a actually eval every other move sometimes. I've been asked about this as well in the chat. -ENcyndaquil-pikachu/systiem a volunteer moderator

I have a suggestion that we can play the engines in the chess.com chess championship.

The frequent nature of CCCCs and rapid game play could give us computer chess fans a chance to watch high quality games from the state of the art in computer chess research. With the emergence of Deus X, Komodo MCTS, Scorpio NN, and Scorpio MCTS, there seem to be a lot of interesting approaches to chess engines coming out. I'm not entirely sure how ready any of these engines are, except for Deus X, which we have seen play in TCEC. If they are stable, it could lead to some very interesting chess, which nobody has seen before. And, because they are still under lots of development, it means that we could see them grow over time. As an example, as we know, K MCTS doesn't support more than twelve cores. However, having it in CCC means that as scaling improves, we get to see the gain in strength quickly, too, which is really exciting.

I calculated the performance ratings of the engines after the first round robin using bayeselo. These are based solely on their tournament performance, not any engine's original rating, so I used an average elo of 3204 (not including Lc0's estimated 3300 rating.) Name Perf CCRL Diff-------------------------Houdini 3423 3400 +23Stockfish 3409 3439 -30Komodo 3406 3404 +2Shredder 3320 3287 +33Fire 3314 3326 -12Lc0 3303 3300 +3Booot 3280 3272 +8Ethereal 3272 3282 -10Andscacs 3268 3244 +24Fritz 3215 3200 +15Xiphos 3214 3179 +35Vajolet 3198 3101 +97Pedone 3194 3090 +104Texel 3184 3144 +40Gull 3181 3184 -3Arasan 3158 3123 +35Fizbo 3157 3283 -126Laser 3145 3226 -81Ivanhoe 3126 3063 +63Wasp 3122 3041 +81Nemorino 3119 3099 +20Senpai 3002 3112 -110Crafty 2957 3013 -56Nirvana 2929 3186 -257 As expected there is quite a lot of variation in the ratings. It looks like Lc0's 3300 estimate was pretty spot on. Also Nirvana is having a really bad tournament (unfortunate for me, I had it placing 17th!)