I was that way too! Having watched Gotham for two years though, I've definitely improved and most things make sense now. Chess is a beautiful game, keep watching and learning!
As the engines play a pair of games against each other in the same opening, I think it would be much more interesting if you would show the second game aswell to understand if it was easy for an engine to win or if it was a real masterpiece.
20:13 for anybody wondering, that !? symbol is actually called an "interesting move" and isn't shown on game review. it means "interesting, but perhaps not the best move", "move deserving attention", "enterprising move" and "risky move"
So it’s more like the cross between a brilliant move and a blunder right like a chance move? If you accept or miss to capture then you win but if the intention was disrupted and your opponent does the opposite then it’s a blunder unless it is already a winning position but for a quick checkmate.
I do really love having stockfish analyze the games because I see the question mark or brilliant move and I feel like I'm watching the smart guy in an anime breakdown what the protagonist is doing.
Video idea: what's the least amount of advantage it would take for levy to beat the best chess bot. Starting with a 0.0 advantage and then slowly increasing the advantage like a .5 or 1 increment. Like if you would like to see it.
@@איתןשי i searched for it, what i found was hikaru vs different levels of stock fish. What I'm proposing is levy vs best bot but levy starts with an advantage
It would be very interesting for Levy to read through some of the code for stockfish (or any engine) with someone who can explain the code. I am curious to see a strong player's perspective on how it evaluates positions and makes decisions.
They all use neural networks, or commonly called AI. The decisions are made based on multidimensional matrices that are refined from random values through training. As a result, I expect there is little chess to be learned in the code. There are some methods to visualise the weights such as showing not only the top move but multiple lines, or heatmaps. But ultimately neural networks still are very much black boxes, and analysing them largely relies on inspecting the results, not the process itself. The code for a bot that essentially makes random moves would look no different than stockfish.
Levi talking about all this brainy shit gets me nodding along and excited to play some chess! Then my 200 elo ass goes and hangs my queen every other game.
I recently played against this Torch Chess Bot and at the end of the game it moved its pawns to spell out the words "I'm trapped in this box, HELP ME!" What do you think it could mean?
So about that "a chessbot that is going to be the answer to, hopefully, in the long run, stockfish" part: Why would you want a proprietary piece of software, developed by a predatory company to win over a community driven, open source engine? I hope that if torch ever gets good enough, it will only motivate everyone to put it back in it's place again.
16:30 haha! that's exactly what I do in those kinds of positions, I actually learned it from playing stockfish, they can't mate you if their own pawn is blocking them.
I remember when Kasparov had a melt down over a move Deep Blue played against him in a game in 1996/97. Can you imagine him playing Torch today, his head would explode. Literally, figuratively, metaphorically and a bunch of other 'lys'.
After watching levy for 2 years, I can proudly admit that i started at 1 000 000 elo and i am now 100 elo. I would like to thank levy for all his help.
I don't believe for a minute that this Torch is stronger than Leela, let's see how it does in a controlled environment like TCEC before we crown it. This is just marketing.
Why not? It’s developed by a very successful company with many veterans in the field. Every time a company with this many resources developed a chess engine it has become number one (Deep Blue and AlphaZero)
I know a decent amount(slightly above average) about how the bots are programmed to do what they do but i still always genuinely enjoy your videos about the chess bots and you just make it seem like an action movie every time. Also stockfish evaluating the other engines games shows just how dominant it is 😂
i believe the reason its called torch is because it is run on python(python has a module for AI called PyTorch) Sorry for my trashy explanation im not sure how to prperly explaining this.
I do like seeing more top level engines, it makes it more interesting. I want to see more stockfish level engines battle it out. But it sucks that Torch isn’t open source.
Not a software engeneer, but.... Didn't in this very same video Gotham say that the French Defense is basically unplayable at the AI level? And then showed a game Torch won against the current #2 where the #2 plays the French? So... Torch isn't the new #2, I see.
French is definetly playable. Infact most classically trained engines (not NNUE) prefer french over c5, or c6. But I think engines prefer exchange french which kills the fun.
Sometimes levy just breezes passed the most unhinged engine moves, and other times he acts like the most natural moves ever are insane. And I think that makes these videos even more entertaining
Levy is so good. Here is an essay why: Levy talks about chess, chess bots, grandmasters and alot more! He puts alot of effort into his videos to make his viewers comfortable. He explains every thing about the bot or GM at the moment. You should watch him, because he is happy for even 1 like. Thank you for reading this, i appreaciate every view.
Hey Levy (GothamChess)! You're amazing at chess! Your videos are so fun to watch, and I've learned a lot from you. Your explanations make complicated stuff easy to understand, and your game analysis is awesome! Thanks for inspiring me to play more and get better at chess. Can't wait for your next video!❤
Hi Gotham! I just played a chess game and got my first brilliant move! Can I contact you somehow and ask you to make a video about it? I am 300 ELO so it might be very fun, I absolutely love content you make, especially 200-300 ELO chess. Thank you early.
What I learned 1. When ur pieces are hanging, push anything. 2. Trade if ur losing at 6:00 3.Play moves that look like blunders 4. Try sac Da Ruuuuucccckkkk!!! 5. Do mental, emotional and physical damage to ur opponent 😂
The final SF vs Torch as 1000 games (500 pairs bias opening , with White try to win and with Black try to Draw, so 750 is almost the maxium points u can get). SF wins 640 vs Torch 360 (+280 pairs) SF 3862 vs Torch 3500 elo.
Hi Gotham I love your videos but wanted to ask will you be doing free tutorials on RU-vid like your openning videos ( in your playlist ) , endgames ( also in your playlist) ??