|
Spiggy posted:I'm willing to donate some elo to anyone that wants a few points. I'm still learning basic concepts so having a low ranking doesn't hurt my feelings. Same
|
#
?
Jan 14, 2021 08:07
|
|
|
|
| # ? Jun 11, 2024 21:39 |
|
|
For both of you who read this thread but not the discord, here's an interesting article exploring how to create a computer chess opponent that not only plays at a certain skill level but also plays like a human at that skill level would (instead of, say, playing perfectly for 8 moves and then playing a terrible move on move 9). They have bots on Lichess at different ratings if you want to give it a shot, too. quote:What does it mean for a chess engine to match human play? For our purposes, we settled on a simple metric: given a position that occurred in an actual human game, what is the probability that the engine plays the move that the human played in the game? https://www.microsoft.com/en-us/research/blog/the-human-side-of-ai-for-chess/
|
|
#
?
Jan 15, 2021 01:05
|
|
|
ulmont posted:For both of you who read this thread but not the discord, here's an interesting article exploring how to create a computer chess opponent that not only plays at a certain skill level but also plays like a human at that skill level would (instead of, say, playing perfectly for 8 moves and then playing a terrible move on move 9). cool thanks im one of those two people
|
|
#
?
Jan 15, 2021 01:09
|
|
|
but how does that work given you’re usually i to new game territory by move 10 or so
|
|
#
?
Jan 15, 2021 01:09
|
|
|
ok i clicked the link and read the article
|
|
#
?
Jan 15, 2021 01:16
|
|
|
Wait, so that's just based on the fact that people are premoving their openings and moving so fast? (that Bh6 thing)
|
|
#
?
Jan 15, 2021 01:20
|
|
|
Zwabu posted:Wait, so that's just based on the fact that people are premoving their openings and moving so fast? (that Bh6 thing) Yup.
|
|
#
?
Jan 15, 2021 01:32
|
|
|
i was messing around with too many pawn moves today and was surprised to find i wasn’t the first person to reach this position: but i was the first to reach this one: 
|
|
#
?
Jan 15, 2021 12:25
|
|
|
ulmont posted:For both of you who read this thread but not the discord, here's an interesting article exploring how to create a computer chess opponent that not only plays at a certain skill level but also plays like a human at that skill level would (instead of, say, playing perfectly for 8 moves and then playing a terrible move on move 9). Interesting article, but it implies in the beginning that the recent surge in interest is thanks to AI improvement. Chess engines hurt the popularity of chess and the fact it's been recovering a bit has nothing to do with them getting stronger. AI-powered chess engines have consistently bested human players since 2005, and the chess world has undergone further shifts since then, such as the introduction of the heuristics-based Stockfish engine in 2008 and the deep reinforcement learning-based AlphaZero engine in 2017. The impact of this evolution has been monumental: chess is now seeing record numbers of people playing the game even as AI itself continues to get better at playing.
|
|
#
?
Jan 15, 2021 13:53
|
|
|
Doctor Malaver posted:Interesting article, but it implies in the beginning that the recent surge in interest is thanks to AI improvement. Chess engines hurt the popularity of chess and the fact it's been recovering a bit has nothing to do with them getting stronger. this seems like a situation where correlation is not causation
|
|
#
?
Jan 15, 2021 16:52
|
|
|
Bilirubin posted:this seems like a situation where correlation is not causation this seems like a situation where AI researchers want to pump up the impact of AI
|
|
#
?
Jan 15, 2021 17:52
|
|
|
Wijk an Zee started this morning. There was one last minute change, with Donchenko in for Dubov after someone Dubov had been in contact with tested positive for the plague. The big Carlsen-Firouzja game is round 1, so that's something to pay attention to. Giri-Tari is also quite a wild piece sacrifice, following some new theory that seems to have largely emerged from Carlsen loving around in these online rapid tournaments.
|
|
#
?
Jan 16, 2021 16:34
|
|
|
https://lichess.org/blog/X-S6gRUAAGjNX4ki/new-puzzles-are-here it seems like lichess has billions of new puzzles, i went from seeing puzzles that 10000 people have already played to puzzles that only 100 people have played.
|
|
#
?
Jan 16, 2021 18:25
|
|
|
Hand Knit posted:Wijk an Zee started this morning. There was one last minute change, with Donchenko in for Dubov after someone Dubov had been in contact with tested positive for the plague. I had Dubov in my fantasy chess team. 
|
|
#
?
Jan 16, 2021 19:39
|
|
|
Hand Knit posted:Wijk an Zee started this morning. There was one last minute change, with Donchenko in for Dubov after someone Dubov had been in contact with tested positive for the plague. Finally the Bong Cloud is theory
|
|
#
?
Jan 16, 2021 19:49
|
|
|
Helianthus Annuus posted:https://lichess.org/blog/X-S6gRUAAGjNX4ki/new-puzzles-are-here Is there out of the box software that can generate puzzles from PGNs of your own games? I'm not sure whether lichess is using something that's practical for a home user of not.
|
|
#
?
Jan 16, 2021 22:52
|
|
|
https://github.com/linrock/chess-puzzle-maker check this out and lmk
|
|
#
?
Jan 16, 2021 23:24
|
|
|
That stockfish analysis feature on lichess is a game changer. I haven't played in a while, so this is all new and wonderful tech to me! My last online chess experience was the Internet Chess club in the aughts, and that client was basically IRC with chess bolted on. I wish, though, that it provided more justification when it points out an inaccuracy. The blunders (??) it points out are clear, but the (!?) inaccuracies can be a little hard to follow.
|
|
#
?
Jan 18, 2021 16:26
|
|
|
Inaccuracies are sometimes difficult to figure out at a beginner/intermediate level because half of them are simple things like dropping a pawn, and the other half dont lose any material but put you in a worse position if the opponent knows how to punish.
|
|
#
?
Jan 18, 2021 17:06
|
|
|
former glory posted:That stockfish analysis feature on lichess is a game changer. I haven't played in a while, so this is all new and wonderful tech to me! My last online chess experience was the Internet Chess club in the aughts, and that client was basically IRC with chess bolted on. I wish, though, that it provided more justification when it points out an inaccuracy. The blunders (??) it points out are clear, but the (!?) inaccuracies can be a little hard to follow. The inbuilt engine tends to wildly overstate evaluations and isn't good at subtleties. I wouldn't put all too much stock in moves that aren't outright blunders, but if there's one that's confounding you don't hesitate to post it here and ask.
|
|
#
?
Jan 18, 2021 17:42
|
|
|
it just means stockfish found a better move than the one you did. if your move was only a little bit worse than stockfish's? "inaccuracy." if it's a lot worse? "mistake." if it literally loses the game? "blunder!" theres a guy on the lichess forums who claims it's tied to the specific scores that stockfish produces, which seems plausible to me. https://lichess.org/forum/general-chess-discussion/what-are-the-differences#5
|
|
#
?
Jan 18, 2021 17:45
|
|
|
That's helpful to know, thanks guys. I'll be sure to post anything interesting as I blunder my way down the ladder. Ol' grey mare she ain't what she used to be. Ain't what she used to be.
|
|
#
?
Jan 18, 2021 18:50
|
|
|
new chess engine hit the streets a couple weeks ago. looks similar to leela, but improved. stay safe out there! https://github.com/dje-dev/Ceres
|
|
#
?
Jan 19, 2021 07:01
|
|
|
Me: *comes up with elaborate multi-check plan to discovered attack into capturing the Queen Analysis, at the start of the plan: 
|
|
#
?
Jan 20, 2021 01:41
|
|
|
Artelier posted:Me: *comes up with elaborate multi-check plan to discovered attack into capturing the Queen well, did it work?
|
|
#
?
Jan 20, 2021 02:03
|
|
|
i played a few games against that new maia1 bot and it does seem pretty decently human. managed to get off one of my favorite halloween gambit lines in against it, and this is a hilarious line that human players in blitz & bullet absolutely fall in to sometimes. at least at my elo
|
|
#
?
Jan 20, 2021 02:05
|
|
|
The poor bot-- that's harassment! My kids have been getting pretty into chess lately and my knowledge of the game is from The Streets (I suck). I'd like to grab a couple of books that might teach me some proper methods that I can share with them, but my daughter will likely want to read some of it, too. So I'm thinking a book with a bit of a narrative to it rather than a bunch of puzzles or a chapter on a passed D pawn would be nice. Do you guys have any you could suggest? I've got A First Book of Morphy in the cart already because it sounds pretty great. I read some of Weapons of Chess in school and hated it. That's about my only exposure.
|
|
#
?
Jan 20, 2021 02:51
|
|
|
Seirawan's books fit the bill. I'd go for the tactics book as the best place to start, but good book on openings is just about worth the price just for the stories of his earliest days in chess as a kid. Opening 1. h4 to get his "cannons" out, queen raids, etc. I think most GMs might not admit that kind of thing, but that's where we all start.
|
|
#
?
Jan 20, 2021 03:00
|
|
|
fart simpson posted:well, did it work? Yes but halfway through Analysis was like Why Queen When Forced Checkmate is right there in 11 moves?
|
|
#
?
Jan 20, 2021 03:00
|
|
|
former glory posted:Do you guys have any you could suggest?
|
|
#
?
Jan 20, 2021 04:17
|
|
|
Captain von Trapp posted:Seirawan's books fit the bill. I'd go for the tactics book as the best place to start, but good book on openings is just about worth the price just for the stories of his earliest days in chess as a kid. Opening 1. h4 to get his "cannons" out, queen raids, etc. I think most GMs might not admit that kind of thing, but that's where we all start. Looking now and I think I'll grab his Chess Duels book. Seems like what I was looking for - thanks! Sub Rosa posted:I don't think books have much going for them in regards to beginner chess. Maybe check out ChessKid accounts for them I was thinking more that the book might teach me some proper thought processes to pass onto them vs. my ad-hoc approach, but this site looks like a great one. I'm going to give it a shot and see if it's at their level. I would be nice to pin them against some other beginners vs. me being the punching bag after dinner. 
|
|
#
?
Jan 20, 2021 04:52
|
|
|
This is the book for you:
|
|
#
?
Jan 20, 2021 04:56
|
|
|
lol This one got me. White to move: 
|
|
#
?
Jan 20, 2021 06:14
|
|
|
isn’t that just took takes e3?
|
|
#
?
Jan 20, 2021 06:29
|
|
|
Yeah. I don't usually pick up on sacrifices like that where the piece is pinned against something other than the queen or king. These generated puzzles are a lot of fun on lichess.
|
|
#
?
Jan 20, 2021 07:42
|
|
|
former glory posted:Yeah. I don't usually pick up on sacrifices like that where the piece is pinned against something other than the queen or king. These generated puzzles are a lot of fun on lichess. yeah the thing about puzzles is if you do them you start to quickly see stuff like that and it translates into real games
|
|
#
?
Jan 20, 2021 07:43
|
|
|
This one broke a GM and blew my mind. The solution he found was really interesting, but I can't reproduce it. I know that he won the queen by promoting the pawn to a knight, which put the king in check. Loading it now in stockfish, the optimal solution it finds is different. e: white to move
|
|
#
?
Jan 20, 2021 18:13
|
|
|
1. Qh7+ The king has to protect the queen so either Kd8 or Kd6. Depending on that, either Qh8+ or Qh2+ and the next move you promote to queen. At least it looks like that without setting the board and checking. edit: No, there's the bishop... hrmmm
|
|
#
?
Jan 20, 2021 21:16
|
|
|
Your initial win condition is promoting the pawn. Without black's queen moving, that's going to require giving check in such a way that white's queen supports pushing b8=Q. That's gonna require a check with the queen on either the b-file or the 8-rank. I guess my first thought is Qe4+, since black blocking with the bishop is met with Qb4+. But it's hard to see all the way through. Because this is a study, I suspect the answer will include (1) at least one quiet king move, and (2) sacrificing the pawn on b8 allowing white to create a skewer. e: puzzle sense is screaming for 1.Kc1 maybe followed up by 1...Qf4+ 2.Qd2
|
|
#
?
Jan 20, 2021 21:26
|
|
|
|
| # ? Jun 11, 2024 21:39 |
|
|
I'm not able to reproduce the outcome the GM got because I wasn't paying full attention until the end. I guess there's quite a bit of leeway with how the king moves. In a Stockfish study just now, it came up with this sequence that promotes the queen safely. In both this sequence and the one I witnessed, they block the initial Qh7+ with the bishop. Not sure why that's so much better yet. Qh7+ Bf7 Qh4+ Ke6 Qh8+ Kg5
|
|
#
?
Jan 20, 2021 23:06
|
|