lichess.org
Donate

Antichess analysis

WhyDoIPlayHere

Hello, are antichess openings such as 1. g3, 1. c4, 1. Nh3, 1. b3, 1.g4 and 1. a4 gonna be counted as blunders in analysis nowadays? Back then they weren't blunders according to analysis and they aren't officially solved as a loss yet. But turns out everything except 1. e3 is a blunder? (Yes, I know that e3 wins)

Hello, are antichess openings such as 1. g3, 1. c4, 1. Nh3, 1. b3, 1.g4 and 1. a4 gonna be counted as blunders in analysis nowadays? Back then they weren't blunders according to analysis and they aren't officially solved as a loss yet. But turns out everything except 1. e3 is a blunder? (Yes, I know that e3 wins)
whatismymainbro

@WhyDoIPlayHere said ^

Hello, are antichess openings such as 1. g3, 1. c4, 1. Nh3, 1. b3, 1.g4 and 1. a4 gonna be counted as blunders in analysis nowadays? Back then they weren't blunders according to analysis and they aren't officially solved as a loss yet. But turns out everything except 1. e3 is a blunder? (Yes, I know that e3 wins)

It is just the way that it works. Never a good idea to trust a fish, after all...

@WhyDoIPlayHere said [^](/forum/redirect/post/KBAkgd6Q) > Hello, are antichess openings such as 1. g3, 1. c4, 1. Nh3, 1. b3, 1.g4 and 1. a4 gonna be counted as blunders in analysis nowadays? Back then they weren't blunders according to analysis and they aren't officially solved as a loss yet. But turns out everything except 1. e3 is a blunder? (Yes, I know that e3 wins) It is just the way that it works. Never a good idea to trust a fish, after all...
WhyDoIPlayHere

I mean yes, g3 is my favourite opening and I play c4 and Nh3 as well. Just curious why they decided to change analysis and now it's like it's forcing everybody to play only e3 heh

I mean yes, g3 is my favourite opening and I play c4 and Nh3 as well. Just curious why they decided to change analysis and now it's like it's forcing everybody to play only e3 heh
gamblej

I'm confused. Is the feeling that Fishnet antichess analysis is worse now than it was at some time in the past? If so, it would be helpful to know approximately how long ago things went south.

I'm confused. Is the feeling that Fishnet antichess analysis is worse now than it was at some time in the past? If so, it would be helpful to know approximately how long ago things went south.
WhyDoIPlayHere

@gamblej Well, I noticed it just now. I've been kinda inactive recently but a month ago when I requested analysis c4 and g4 used to be inaccuracies and other openings which I mentioned weren't marked at all. I've checked other people's games and seems like even a week ago status of the openings was the same and it changed only like 4 days ago. Check here - https://antichess.onrender.com/openings/ - these openings aren't proved as a loss, it's like they have neutral status for now so do they really have to be counted as blunders? I think blunders are openings which are already solved as losing

@gamblej Well, I noticed it just now. I've been kinda inactive recently but a month ago when I requested analysis c4 and g4 used to be inaccuracies and other openings which I mentioned weren't marked at all. I've checked other people's games and seems like even a week ago status of the openings was the same and it changed only like 4 days ago. Check here - https://antichess.onrender.com/openings/ - these openings aren't proved as a loss, it's like they have neutral status for now so do they really have to be counted as blunders? I think blunders are openings which are already solved as losing
WhyDoIPlayHere

Oh, and this tpr guy always pops up in such topics. It's not new, Watkins solved e3 years ago and everybody knows it. Yes, e3 wins with perfect play but people aren't engines and there are tons of variations to memorize. And also breaking news, we can use other openings outside e3 to make the process fun. Ok buddy, u don't even play antichess so if u want to prove something, try to win 1000 out of 1000 games against 2200-2500 rated opponents in a row (with 1. e3 as white) and then come back

Oh, and this tpr guy always pops up in such topics. It's not new, Watkins solved e3 years ago and everybody knows it. Yes, e3 wins with perfect play but people aren't engines and there are tons of variations to memorize. And also breaking news, we can use other openings outside e3 to make the process fun. Ok buddy, u don't even play antichess so if u want to prove something, try to win 1000 out of 1000 games against 2200-2500 rated opponents in a row (with 1. e3 as white) and then come back
tpr

Here is a study about it

https://lichess.org/study/7NU0Q3hs/TT7X6noA

Here is a study about it https://lichess.org/study/7NU0Q3hs/TT7X6noA
gamblej

First thing I'll say is that Fishnet's Fairy Stockfish uses HCE or Hand Crafted Evaluation. There's an antichess NNUE that allegedly adds +200 ELO (CCRL?) and this is what your browser engine uses. A recent update lets you use that browser engine to do full main line analyses of games, so first I'd try that. It should be quite a bit stronger than Fishnet.

I did some strong 30s (400 million node) evals of c4 and g4 openers. Starting position is +0.8, c4 and g4 are both -0.3. This places them right on the cusp between inaccuracy and mistake. Not blunders according to FSF Antichess NNUE at depth 24 / 400 million nodes / 30s. They can oscillate between inaccuracy and mistake as you go deeper. The hard boundary between the two is artificial and mostly meaningless. It just had to be drawn somewhere for classification sake.

First thing I'll say is that Fishnet's Fairy Stockfish uses HCE or Hand Crafted Evaluation. There's an [antichess NNUE that allegedly adds +200 ELO (CCRL?)](<https://fairy-stockfish.github.io/nnue/#current-best-nnue-networks>) and this is what your browser engine uses. A recent update lets you use that browser engine to do full main line analyses of games, so first I'd try that. It should be quite a bit stronger than Fishnet. I did some strong 30s (400 million node) evals of c4 and g4 openers. Starting position is +0.8, c4 and g4 are both -0.3. This places them right on the cusp between inaccuracy and mistake. Not blunders according to FSF Antichess NNUE at depth 24 / 400 million nodes / 30s. They can oscillate between inaccuracy and mistake as you go deeper. The hard boundary between the two is artificial and mostly meaningless. It just had to be drawn somewhere for classification sake.