To be fair stockfish level 8 on lichess is also not even close to the full power of stockfish 15. I think it's like 400-600 points lower? That tells you how crazy strong we've made engines lol
@@scoutbane1651 Hi Scout.I never played an "engine", but it occurred 2 Me what if a weaker engine had been given the advantage of "more points. Not that I'd learn -just a curiosity of it's method.
I'm not a fan of 1-minute games. I don't think they're good for learning. I would try the same format but in a 3-minute game with increment so there's no flagging. And to save time, just start with a +5 point advantage.
Stockfish used a few pawns, a Queen, some knights and bishops and a rook to checkmate you, everything else was in its deep reserve. So when you took out a pawn or rook or knight it used what was left to launch its standard attack, but when you removed its Queen you effected its core attacking forces, (I.E., removing forces it had no indention of using had no effect on its attack.)
AI knowledgeable is arguably more useful than being good at chess if your only goal is to beat engines. The technical restrictions are the real bottleneck of engines skill.
Haha I really felt that one! You could try the same challenge but with increment, I reckon you could shave a couple of points off then. Or, you could steadily add points of material to your side, converting pieces to better pieces etc until you could win and see how that compared to deducting points from SF.
@@ChessVibesOfficial can you do this with stockfish 16 where stockfish 16 has 10 mins thinking in total time? Has any human in the world succeded with that
Brilliant, that’s how I used to handicap the mixed levels of tennis players on some ‘match play’ training nights… if someone lost a game they start the next at fifteen love up, and if they lost the next game they’re thirty love up, or they drop back by one spot if they won. Kept everyone competitive.
Day 11 or something (idk at this point) of posting my chess puzzle: The FEN: 2K5/PP6/pk1p4/npb2n2/2rpb2p/4ppp1/8/6rq w - - 0 1 Its white to play and win. I'm not giving in until this puzzle gets featured lol
How about beating Stockfish in correspondence chess, that is limitless time control? It usually moves pretty fast anyway, so you could walk us through the thought process, and see how much handicap that would take. I think it would be easier.
Guess before watching the video: I've beaten stockfish with the queen removed before and I'm around 1600 rating so I'd say you can do it with a 5 point lead, definitely with a queen though. Edit: Oh wait it's bullet, I'm changing my guess to you being a queen up. EditEdit: Yup, removing the queen makes Stockfish really docile, always ok with trading pieces for some reason.
I think you should play classical against Stockfish and play as long as you win, not taking away any material. That way your content problems would disappear =)
Idea for future video: Try to beat Stockfish with unlimited number of take backs. EDIT: I.e. no time control. No nothing. Just try to beat the engine (at a really high level) and do whatever it takes (except using another engine of course).
My idea too. Try thousand of different things and if that's not enough, then exploit that Stockfish will sometimes pick worse moves just to be less predictable.
I'm not sure if there are any known lines where Stockfish loses when playing from turn 1 with any sort of reasonable time control. There are a few where Stockfish loses from a forced opening.
I did that as a child multiple times with Chessmaster 2100 (?). Today when you try, you just instantly realize how vast of a game chess is and how little you know about it…
@@G.Aaron.Fishersometimes (not often) there are openings were stockfish things that the wrong player is better. (I don't know if there are any known in current stockfish)
Absolutely loved this video, a really interesting exercise! The 1mn time limit does make it super hard, I’m sure a 10mn time would give you more wriggle room. I’m the old days of computer chess, I found I would outperform the computer at longer times, but I’ve no idea whether that would still apply. Anyway, we’ll done!!
I actually did predict that it would take a queen missing for you to win. I think your strategy of using every opportunity to exchange pieces was a good idea and makes good chess sense. I also think that playing only a 1 minute game, particularly with no increment, was really a disadvantage for you... Maybe try 1 minute with a small increment instead?
Basically the only thing that prevent stockfish from crushing you even 20 points of material behind is that he is considering you as his equivalent. He is only playing lines that he can't refute himself. But if he "know" that your not able to calculate any of those lines he would spend ~100ms per move and proceed to demolish you or Magnus Carlsen. All he has to do is avoid complications and play instantly. Stockfish is a 3900+ ELO bullet player (according to CCRL 40/2 format which is comparable to bullet) and crush all the other engines with a large margin in this time control. I'm pretty sure no human would never beat him if an adaptive strategy depending on the opponent estimated strength was implemented in stockfish.
As an example Hikaru Nakamura made a RU-vid series where he casually crush 2300+ rated player sacrificing his queen against a piece, in 3+0 time control. But he know the good stategy like not exchanging too much etc. So imagine what an engine would do in bullet. He would smash the piss out of us
Your thumbnail makes it appear that you played more games than is material on the board to remove from stockfish. My guest is 0-2, advantage stockfish, 3 even odds, 4 or more advantage Nelson. For me, you would need to double those numbers.
Try beating Stockfish 8 on even material in one single game but you undo thousands of moves trying different things or maybe even trying to make him do something stupid if he decides 10 times on the same decision (there's some RNG in which move he chooses because otherwise you could just let him play itself and learn his one line and then you could beat him without outside help).
Maybe instead of removing stockfish's pieces, you can add more pieces for yourself. It would put up for a bigger challenge because you'd have to trade more when you're up material.
It'd be cool to see the same type of video, but with increasing time odds instead of piece odds. For example, the first game you play Stockfish in a 10 minute vs 10 minute game, then if you lose play a 10 minute vs 5 minute game, then 10 minute vs 3 minute, 10 vs 1 minute, 10 vs 30 seconds, then 10 minutes vs 15 seconds.
You’re handicapping yourself too much by only giving yourself a minute. Stockfish does more calculations in a second than thoughts you could make in a day.
The ELO score is configured such that for every 400 points of separation between the ELO ratings of any two competitors, the lesser ranked one will statistically have a 10% chance of victory. As you´ve played near enough to 100 games, meaning a 1% win-to-loss ratio, and _Stockfish_ has a rating (you said) of 3000, that means that your rating is about 2200. Seems low, and I think the discrepancy is the crucial competitive attribute that computers do not get distracted or fatigued; however, because you are handicapping _Stockfish_ each time, the gap in ELO should be even greater.
Level 8 is not max Stockfish; it's made to make mistakes. Plus, no browser based Stockfish will ever be good. For best results, you should use profile built Stockfish on fancy CPU with AVX 512 support in a minimalistic browser. If you don't have the hardware, borrow it from a friend (you probably do).
You want ideas against Stockfish? Try playing Stockfish, but Stockfish has no pawns! It's 8 points of material, instead of 9. Are pawns truly the soul of chess?
Ok nice event, (honestly) but why the misleading title then... Just do such nice content without clickbait and false promise. Unfortunately, it's so modern and abused and not nice for a viewer who want's to decide what to watch. (which you can't if you get an intended wrong impression before clicking) IMO it's not good for long term viewerbase. Luckily I am subscribed to you for quite some time though :-) just some feedback.
This is actually possible without talking material from Stockfish. Level 8 Stockfish is designed to blunder once in a while. That wouldn't be possible against actual max stockfish.
5 points, maximum. it depends on what exactly you remove, but you are surely good enough that you can win, a rook up. Just trade all the material. The problem is, the time control is so short that you can not mechanically complete the game. Thus the experiment is not, strictly speaking, about chess at all, but about how fast you can move your hands. At the very end "how did i not win ?" because you dawdled. it's bullet. 1 plus zero. all such games become not-chess near the end.
Play Stockfish over and over until you win again, but instead of removing 1 point of Stockfish's material for each lose, add 1 point of material to your pieces
How well did you do, guys? I'm hopeless. I drew wtih queen, knight and pawn ahead and won with a queen and two knights up. That's how much material I needed to beat Stockfish. My FIDE rating is currently 1892.
@@brianbuckman6908 I beat Stockfish in my first ever game, so I had to start giving it material. I lost when he was up 12 queens. (I'm 100 rated on Lichess)
How he says "good luck Stockfish" dead serious in the very first game with the 3000 elo engine and no changes this confidence is refreshing lmao. Must be a masochistic experience
I did it without a queen for a stockfish, was pretty easy, just trading down and down. The hardest part was not blundering anything in a stupid way :) update: obviously not in a 1 minute game
I think you should redo this video, giving yourself extra time (3 minutes + 2 seconds per move?) and stockfish the same amount of time. Could be really interesting in my opinion
Remake the video but it’s you vs stockfish with just a king , keep playing until you lose, add 1 piece of material every time you win until stockfish has enough material to win
Can you manually limit Stockfish's calculation time, or depth? See how shallow a depth you have to take it to before you can win, or how many milliseconds of thought time it needs to still beat you.
10:04 im not a im, nm, gm, fm or anything else i literally have 500 elo but couldnt you go to Re1 with any rook to protect both rooks? or is it pinned or sum
One minute is too short in time. Nelson had to make lots of pre-moves. Two minutes would be more reasonable, where Nelson could win with a 4-point handicap.
Iirc Hikaru could do it with 2 points, but one hour game and he lost most and only won a few (also he managed to draw a few against Komodo with just 1 point advamtage which was impressive), so I assume Nelson can do it with 4 or 5 and if more than 5, then Stockfish 8 is probably harder to beat on 1 minute vs on 1 hour.
Dude I'm beating Komodo max on com without odds every second or third game if I make him play sth just very, very slightly off like Caro or Scandi ;p Hikaru only drawing Komodo with pawn odds in multiple games is actually pathethic unless it was bullet :D
You should play stockfish right up until the moment where you think the game is hopelessly irretrievable, and then switch sides. Just to see if stockfish can get itself out of the mess it created.
Does Stockfish care at all about time? I know all computation takes time, but it doesn’t think on human time, so it might get all the moves done in a minute even if you give it 30.
In the game the bishop forked the rooks the rook on a1 could have moved to e1 to protect the rook on e3 preventing losing the chance to capture the bishop.
