Pain

Needlessly savage

Your name says something about you

“Your username says how you die”

I do food reviews while I’m high off that zaza on my yöutube chånnel 👌👌

It’s interesting to see how an AI evaluates such interestingly complicated positions and gives the playing AI a chance to mess up the AI bar with very complex moves.

“Akshually I am winning in this position 🤓” “No, Akshually I am winning in this position 🤓” How I imagine the conversation goes in between transistors.

@@Lius525 ayy zaraki my man you got bullied by yhwach in the last episode

reminds me of the batman pointing at batman meme

“Stockfish vs. Stockfish is a perfect game according to Stockfish”

Evil 😂😂😂

This had me dying!!🤣he’s so casually funny.

6:05

Lol💀💀💀

@@SFH2042THANK U

💀💀💀

After thinking for 20 seconds the eval went from -0.8 to 0.0 wow

For the wages of sin is death, but the gift of God is eternal life in Christ Jesus our Lord

@@fn_cxrsed2984 hello?????

@@fn_cxrsed2984 hail satan

this is why chess will never be solved. there will always be some inacurracy thats actually a brilliancy if u calculate it long enough

question, why these games where there are so many great games generated on "teesec" (tcec)? and its quite clickbaity too, quite a few (apparent) fortresses you can say "LEELA CANNOT FIGURE THIS OUT"

ratio

Watch out + Lock your doors + Lock your windows + Block all pipes and entrances into your home + Secure the door with Wooden Boards/Planks + Prepare and stock food and water + install CCTV Inside and outside your home + lock your bedroom if you're gonna sleep + the fog is coming + don't look up + stay in the light + keep light sources open + they are coming + watch out + be alert + stay safe + behind you

hi lol i'm dumb

Lichess better

Stockfish seeing this 😢

alphazero is hardly what one would call a bot. it uses a neural network for decision making that emulates the neurons in a brain. furthermore it learned the game from scratch as it was given only the rules of the game and then proceeded to play millions and millions of games against itself ( reinforced learning, each trained iteration would be a bit better than its predecessor, so its like it went through an evolution from the worst chess player ever to an immortal silicon overlord chess engine that revolutionized the way chess is played).

@@Pintkonan I know?

Here is the PGN use it to analyze the game anywhere... 1. Nf3 Nf6 2. d4 e6 3. c4 b6 4. g3 Bb7 5. Bg2 Be7 6. O-O O-O 7. d5 exd5 8. Nh4 c6 9. cxd5 Nxd5 10. Nf5 Nc7 11. e4 d5 12. exd5 Nxd5 13. Nc3 Nxc3 14. Qg4 g6 15. Nh6+ Kg7 16. bxc3 Bc8 17. Qf4 Qd6 18. Qa4 g5 19. Re1 Kxh6 20. h4 f6 21. Be3 Bf5 22. Rad1 Qa3 23. Qc4 b5 24. hxg5+ fxg5 25. Qh4+ Kg6 26. Qh1 Kg7 27. Be4 Bg6 28. Bxg6 hxg6 29. Qh3 Bf6 30. Kg2 Qxa2 31. Rh1 Qg8 32. c4 Re8 33. Bd4 Bxd4 34. Rxd4 Rd8 35. Rxd8 Qxd8 36. Qe6 Nd7 37. Rd1 Nc5 38. Rxd8 Nxe6 39. Rxa8 Kf6 40. cxb5 cxb5 41. Kf3 Nd4+ 42. Ke4 Nc6 43. Rc8 Ne7 44. Rb8 Nf5 45. g4 Nh6 46. f3 Nf7 47. Ra8 Nd6+ 48. Kd5 Nc4 49. Rxa7 Ne3+ 50. Ke4 Nc4 51. Ra6+ Kg7 52. Rc6 Kf7 53. Rc5 Ke6 54. Rxg5 Kf6 55. Rc5 g5 56. Kd4

@@nikunjsharma6043 legend

@@nikunjsharma6043 how to use it

​@@PrakritiSenpai lmao take a ss open Google lens copy text and use

@@Mihir_hun_vaii where to use and how to use Lmao ik how to copy text from comments Lmao ³

Ok

Ok

Ok

Ok

Ok

No no this just means that stockfish should judge you but alphazero is better at judging your 438 rating ass

taught me :)

Good for you man😀😀

@@stevekaye4438 thx bruv

Wait till you reach 2000. Gotham will decrease in value when you go up in rating

@@johnora3857 Bruh I'm 1400 and know 3 openings 🥶

I never noticed, but yeah, you're absolutely right.

I swear, cleaning them is always a pain. He probably has some really expensive scratch resistant glass.

@@aryamanrai1972 - all glasses come with scratch resistant lenses at no extra cost when you go to Specsavers - lol!!!

New York City Central Park chess is like this sometimes

ya! I'd love to see it do humanistic commentary as it analyzes. maybe a few gens of engines later and we see friendly personalities tagged into some of them

​@@pauls5745 Or they hook Mittens up with ChatGPT to trash talk us

yea I actually burst out laughing when I heard Levy say Queen to H..1 😂

I seen H4 too but don't understand why he did that in between move instead of going there right away?

We already have Stockfish v. Stockfish, but yeah, the other engines vs themselves would be interesting.

it hasn't even been 1 min until the video was uploaded lol

lmao...Its been 1 minute since the vid has been posted and you already watched it. ;)

Man be going into a movie theater saying this is the best movie i watched 1 min in, and then stands up and leaves

Dick riding for free...

Yeah that's the joke actually, if it was needed to be said lol

Good hans reference lol.

I wonder if this is just going to be a meme forever in the chess world

Thx 👍

Humans play chess just differently, instead of strict math and exhausting and rating all the options 30 moves ahead often it's more about how things feel, what sort of presence they represent and abstract advantages seeing the exchanges and situations just a couple moves ahead. For humans it's a move that feels wrong from the outset, a complete shift of focus away from the active pressure followed by some pretty hard to see board evolution. Sure you can easily see it cuts away the queen's diagonal, but do you want to do that? Now? Aren't there more pressing exchanges which could cost you the game if you don't follow them? So the brilliance is quite hidden in the sense that it's 1) seemingly low value 2) resulting in kind of a complex and hard to quantify board evolution 3) fear of losing an advantage.

and plus does someone knows the details of the hardware that these 2 engines were running on? i am pretty sure stockfish had weaker one or was that some other game??

Thanks for saving my time

@@JustAPokemonCommentingOnVideos and stockfish always makes a comeback anyways...soooo, I'm just not impressed

@@JustAPokemonCommentingOnVideos Like, AlphaZero got to use some fancy Google custom TPUs. No amount hardware can equal that. And the insane thing is, we now know that Stockfish almost held the position.

He was held hostage by the fish

Ever since I started watching engine games I lost interest in watching Carlsen

Í was excited because I saw it a move before b5 . I didn't see h1 and wouldn't know how to go from there, but im proud we saw it and would've played it

@@nicholasparker2086 GTE from just reading a comment. Your elo is around 1000-1200, am I right? cheers!! btw.

Hans Niemann speaks for himself

Wow man you’re so funny and original

Agreed! It's not a clean German pronunciation. With the NYC connection I like to imagine it's Yiddish but maybe it's Slavic.

@@SethKBaldwin No, it is simply the pronunciation of an English speaker.

Wow

True i hat that clickbait

Could also mean "stockfish simply got crushed"

That annoyed me too. I was hoping to see a brand new innovation in computer chess!

@@soupisfornoobs4081 Yeah and as always he changes the title after about an hour to make it less clickbaity

@@soupisfornoobs4081 He simply wanted the video to perform well in the first hour with clickbait and than changes the title for the viewers who would have watch the vid anyway later

I was shocked that I saw it as a newbie.

That's happened many times at TCEC.

@@seasideman I take it stockfish won?

@@JB_inks Yes

@@seasideman I looked, I can't see any results

@@seasideman Alphazero was never available to the public so it has never played at TCEC

It's not really a bait. AlphaZero would likely have gone Qd6 as black as well. So when deciding between Qa4 and Qf4, it basically just calculated whether it prefers black's queen on d6 or d8 and decided for d6. Inducing a move like this is a somewhat common tactic in high elo games, but more often with pawn moves as they can't move back obviously. Generally, engines don't do "mind games". Engines always assume their opponent plays just like them when doing their calculations.

It's hard to say, the folks working at Google certainly know what they are doing. However, AlphaZero in this video was using a method called MCTS, the Monte Carlo Tree Search algorithm, and it was way more efficient than the brute force calculation that StockFish was using at the time. Current day engines use a new method called NNUE - stands for Efficiently Updatable Neural Network - and basically this just means that they have combined the brute force and MCTS methods together. (It is a little more complicated than that, but that explains it well-enough in simple terms) As strong as the MCTS method was, it was still not the most optimal way of doing things. It cut down computations by having a "stored memory bank" (think preparation but an entire game's worth vs the first x moves) that it could reference, and it built this reference by playing against itself. You could certainly feed additional databases into this to make it stronger, but it would still have a blind spot or two somewhere and would need to rely on it's own computational power to solve positions, which brute force methods were still better at doing. So now we have engines that have this huge network of databases that they can reference, they store the games that they play to increase the size of said database, but they also still have fine-tuned brute force calculation methods to try and evaluate complex positions when they don't have a reference to the position on the board. Honestly, it would be really cool if one of the major NNUE competitors (Stockfish, Fire, LCZ) would start releasing these databases in a form akin to move explorers so that we could see if this position "has ever been reached before" in that computer's database - this would allow us to illustrate the differences between the "MCTS" and "Brute Force" methods much easier.

@@LunaticSoldiers ah thank you. I guess it's weird to think that every single possible outcome of chess is theoretically possible to calculate but absolutely not probable. Then again my phone I'm typing this on would be an impossible miracle two decades ago so who knows.

@@willowdrakon We used to try and do that with computer engines back in the day! :) Sometimes they would take all day to make one move. Some would take even longer, encroaching on multiple days or even weeks if left to their own conventions. Simply put, some moves are actually just bad and don't deserve a second thought, and the first huge computer milestone came when we were able to teach engines how to identify what moves to ignore. It increased computational efficiency by allowing them to ignore the millions of branches that stemmed from a bad move. I'm not sure how well-versed you are on the MCTS algorithm but it was and is still a very huge computer milestone for lots of AI projects and not just Chess. Specifically, it enables us to emulate the human approach to problems. We stand on the shoulders of giants, our ancestors, who have already figured out information prior to us and we are steadily finding new discoveries to add to that ever-growing list. the MCTS algorithm does just that, but since it is a computer, it can simulate generations much faster than we would see it happen in the real world. Likewise, we also have the ability to establish a framework that the AI can start with. An example would be that the first generation of AI samples learning to walk would probably all mostly fall over, but one might take two or three steps. We can then create a second generation where they all start with that three step foundation. Now, the best iteration of that generation may have taken 12 steps. We can rinse and repeat this until we have a generational number that the hypothesis demands. In Chess, the approach is similar but you don't really have to add generations to the list because you can just store the results of the games vs itself in it's own database and, provided your algorithms make sense, they can just make decisions based on those algorithms. When in a match, the first step in the algorithm is to perform a "tree search" to see how many times this position has been reached, and then select the move with the highest winning percentage. It is extremely efficient in terms of computational power, but it has a big oversight; what if the opponent has a move that destroys the position, but has not been played yet? This is where engines will still take a brute force approach to compute their chances based on their algorithms. It is really fascinating stuff, honestly.

They probably saturated their network in those 4 hours (aka, the network won't get stronger with more training). And saying "4 hours" was much better for PR purposes as seen by the fact that you still remember it even if it was 4 years ago.

No.

None?

"mush"

And now bcs u kept improving it every day, it's now much better than what Google could ever do

@@-zelda- Because Deepmind stopped developing AlphaZero, and everyone knows what happened when they worked on it. Just imagine what will happen if they restarted to work on it...

​@@awmdanger9677 Nothing would happen, look at Leela, it's an engine based on the same ideas as Alphazero but much stronger and it can't catch up to Stockfish

@@-zelda- Google deepmind team has world's best dl Engineers and if they stopped building alphazero doesn't mean that they can't beat stockfish.

@@awmdanger9677 It doesn't mean that they can either lol again, Leela is way way stronger than Alphazero (the one made by "world's best dl engineers" lmfao) and it is still weaker than the fish

I hope your making a joke that he’s an engine cause he uses one