When Hitler invaded Poland, it took all of two days for basically all of Europe to realize that they were about to replay the Great War (which we now call WW1).
Of course it took longer for it to blow up into a truly global war (Pearl Harbor etc), but a conflagration across Europe is hardly a "small regional war".
Declaring war is one thing, but if you look at how leaders actually responded it's another (notice the 8 month gap from the declaration of war, into actual fighting). They were still willing to negotiate with Hitler, because most western leadership also wanted the communists to be destroyed and thought Hitler would do just that without attacking them. They were willing to push for this literally until the tanks were invading their streets.
Once Hitler invaded France the "phoney war" turned into a real war. [1]
Once I became a Dad, getting socks for Christmas suddenly turned into one of the most thoughtful gifts possible. A self-care item. The flip was very sudden and surprising.
I actually chose "socks for Christmas" as a metaphor because the types of people who like socks for Christmas are grown-ups, and grown-ups tend to actually view municipal finance as important and interesting.
I feel like my papers are better for having gone through peer review, and I'm a better researcher for having had a few rejections. Of course the reviewers can't hover around in your lab watching everything you do. But even if reviewers can't check the validity of the evidence in your paper, they do a pretty good job ensuring that the claims you make are supported by the evidence you present. That's a valuable if imperfect guardrail! What would be the alternative?
Peer review has never been an indication of truth, validity, or utility.
It's only ever been an opportunity for other scientists (ideally more competitive than they are today) to see if they can spot some methodological problem.
That's completely upside down take. The problem with peer review is not that it does not allow good papers to get published (that rarely happens, almost all good papers get published!), but that shitty papers get published!
Yes modern AIs have an entire opening database and generally have cached the first 20+ moves of the game (for most common openings) from a database of very deep searches identifying the best move. This is absolutely a form of opening prep for AIs.
That said, even without that database a modern AI will completely topple the best human at every common chess variant. Humans cannot defeat modern AIs in chess like games.
Like my answer below, that's wrong. Even I have achieved a few draws or even wins against Stockfish in training games, and I am FM strength. From time to time you are happy to reach a simple rook endgame which happens to be won and the engine doesn't anticipate that (horizon effect). You still draw or lose 90% of those but you win 10%.
Either the engine was misconfigured, the hardware you were playing on was glitching or you are omitting something. There is no chance in the world that you can beat stockfish in standard time control.
Just because you can not do it it does not mean that others can not do it. If you search for Lichess games where strong players play against (edit: strongest!) Stockfish (which, admittedly is not the full throttle Stockfish) you will find that Stockfish by far does not win all the time. Such is a claim which only inexperienced chess beginners and Stockfish fanboys make. Stronger players know that Stockfish is relatively better, and by a far margin, but – obviously – does not win all the time due to the huge drawing range in chess. Admittedly, winning a game gets more and more difficult with every year. And, to make you happy, I have never beaten Lc0.
> If you search for Lichess games where strong players play against Stockfish ([..]) you will find that Stockfish by far does not win all the time.
I'm sure some of those games are actually stockfish v stockfish or something similar. Its pretty easy to run stockfish or lichess locally and copy the moves from each engine back and forth.
@josephg (for reasons I do not know there is no reply link below your post)
Sure, some people are cheaters. Some are not. There is no personal win in cheating against Stockfish. Usually strong players do it for training purposes, or to entertain their watchers when they stream. I actually remember having seen one who did that, and he drew. That was a party.
Yes. I hear this claim from above: "Some humans can beat stockfish."
Evidence given: "There exist some small number of games on lichess.org played against stockfish where the user won."
My counter argument is that games on lichess against stockfish don't imply a human beat stockfish. It could just be that stockfish (or other bots) can sometimes beat stockfish. And some humans surely use bots to play on their behalf in order to cheat in online games.
I don't know if any humans can beat stockfish. But I don't consider that to be strong evidence.
> My counter argument is that games on lichess against stockfish don't imply a human beat stockfish. It could just be that stockfish (or other bots) can sometimes beat stockfish. And some humans surely use bots to play on their behalf in order to cheat in online games.
Also, Lichess' Stockfish runs in the browser (with all the slowdown that entails), plus is limited to one second of thinking time even on the highest level. It also has no tablebases and AFAIK no opening book. Even if you _can_ consistently beat Lichess Stockfish level 8, there's still a very long way from there to saying you can beat Stockfish at its maximum strength, which is generally what people would assume the best humans would be up against in such a duel.
People generally don't play unencumbered engines anymore because the result isn't interesting.
Well, there is nothing I can do to prove to you that I did, as I can not travel into the past taking you with me. I know, I did win two or three games and drew approximately 25 out of approximately 500 training games. But I can not prove it. You have to believe or not.
I believe you. I just suspect stockfish was misconfigured, it wasn’t playing at its highest skill level or something similar was going on. That seems more likely. (I’d love to know for sure though).
Yeah his claim is quite absurd really. If it was a weaker stockfish (bad hardware, older version etc.) then maybe. Modern stockfish pretty much crushes any and everyone. A draw alone would be extremely impressive, and maybe doable with enough luck from a top player. But even that is very far fetched nowadays. Let alone actually winning.
Elsewhere in the thread he revealed that he achieved these results around the year 2015, which means we was playing against Stockfish 6 or earlier, estimated to have about 400 less ELO than today's Stockfish 18. Stockfish 6 didn't even have NNUE, so the real issue seems to be that he thinks his results from 2015 hold any relevance to the chess engines of today.
No not at all! You can find plenty of videos on YouTube of humans taking down 2015-era stockfish. Usually it involves exploiting specific weaknesses in the engine, for example bringing the game to a stalled position where the game nearly reaches the 50 move rule, and then the engine makes a disadvantaged move to avoid a draw.
Especially pre-NNUE, chess engines were often not fully well-rounded, and therefore a human with specific knowledge of the chess engine's weaknesses could take it down with enough attempts.
Would you be willing to bet money that you can beat a properly setup stockfish, no piece odds and even time controls? I'll give you literally any odds you name and let you try an unlimited number of times until you give up. 100% serious.
P.S: You should not take this bet. You will lose. You are mistaken if you think you beat stockfish.
If you're betting against modern stockfish, respectively, that's a terrible bet.
There are some games of knight odds Leela playing superGM's.
For example, Hikaru Nakamura went 1 win, 2 draws, and 13 losses against LeelaKnightOdds at 3 minutes + 2 sec increment:
https://www.youtube.com/watch?v=pYO9w3tQU4Q
So that's a score of 2 out of 16. Which is apparently actually very good. I know Fabi played a lot of games too, and also lost almost all of them.
And that is with knight odds lol. And stockfish is ever better than Leela, but generally less aggressive and more methodical.
You clarified in another post that you had won back in 2015. I have no clue the strength of engines back then (I imagine still very strong of course), but a decade of growth is a lot. They're completely insane nowadays.
I doubt that. Stockfish 11 years ago as you claim (which around then was rated approximately 2800), maybe. Stockfish today? Stockfish on Lichess is 3000 and that's not even running at full capacity. A fully supported Stockfish running on top hardware is currently 3650ish. It can avoid known draw lines and stalemate lines, and could absolutely crush the likes of Magnus.
Further, if the engine does not use an opening database and the thinking time per game is the same, then the engine will usually make the same moves, so you can learn from your errors. There are just a few chess engines which "learn" per default and therefore change their moves, like BrainLearn.
I have achieved these results around 2015, sitting at home, relaxed. I was not in a match situation observed by millions. Such a situation can knowingly lead to blunders like Kramniks overlook of mate in 2.
I also sometimes "cheated" by aborting the game when I was tired and continuing it the next day (if at all). That's what the player in a match can not do.
I also sometimes restarted a game at a specific position. Can also not be done in a match. Finally, they used better hardware in these matches. I had eight threads on my old Laptop and I used four of them. The Laptop itself was bought around 2005. Between 2000 and approximately 2020 I trained every day and I was on my peak. I am still around 2400 on Lichess today, without training.
So, I hope it does not sound that extraordinary any more. It isn't. Maybe it is now, but not then.
2015 stockfish is quite a different beast from 2026 stockfish. Stockfish didn't even add NNUE until 2020.
Based on what data I can find, it's estimated that the difference between the 2025 stockfish (stockfish 6) and today's stockfish (stockfish 18) is nearly 400 points.
That's the difference between Magnus Carlson at his peak and someone who doesn't even have enough rating to qualify for the grandmaster title.
So yes, the fact that you beat stockfish in 2015 doesn't sound extraordinary, because AI today is vastly stronger than it was when you achieved those results. What sounds extraordinary to people is your belief that you could repeat those results against today's top chess engines.
Out of sheer curiosity, I did a bunch of research to understand just how dramatic a 350 point rating gap is in real word chess. Magnus Carlson, for example, has a 98% win rate against players >350 rating points lower than his own, with zero recorded losses.
In fact, there is only one game I could find in all of Chess history (Anand vs Touzane, 2001) where a super GM (rating >2700) dropped a classical game to someone more than 350 points below theirs (gap: 402 points). (it's estimated that there are between 2000 and 3000 classical games in history played between Super GMs and players >350 points below them) And it could easily be that Anand was ill, or suffering some other human condition which made his play significantly worse than his typical play for that game - which you would not see from a computer engine.
In other words, the Stockfish that you beat in 2015 would itself be expected to get 3-5 points (that is, 6-10 draws and 0 wins) in 500 matches against the best chess engine of today. The delta in strength is immense, and it is reasonable for everyone else in this comment thread to assert that you would have zero chance at all of picking up a draw against Stockfish 18 in a fair game of any time control, regardless of how many matches you played.
I do not know the time controls anymore, but I always use the latest Stockfish with all available threads. No opening book, but I do not repeat lines to take advantage of that, because I play to train calculation. I guess hash was the (for my setup) normal 4096 MB.
Latest Stockfish with all available threads and no opening book is still well beyond any human. Elo ratings get a bit silly with computers, but we're talking an Elo of well north of 3000.
For reference: The last serious match between the top human player and an engine was Brains in Bahrain, Kramnik–Fritz 7, in 2002 (already that should tell you something). Well, actually a broken and buggy version of Fritz 7, but that's another story. It was a 4–4 tie. On the latest CCRL list, Stockfish 18 outranks Fritz 8 (the oldest Fritz version on the list) by 947 Elo points _on the same hardware_. (For comparison, Magnus Carlsen's peak rating is 65 points higher than Kramnik's peak rating.)
Add to that 24 years of hardware development, and you can imagine why no human player is particularly interested in playing full-strength engines in a non-odds match anymore. Even more so in FRC/Chess960 where you have absolutely zero chance of leading the game into some sort of super-drawish opening to try to hold on to half a point now and then.
I expect this would hurt humans more than machines... but I'd love if someone with better knowledge than me of the current state of chess engines could chime in.
Engines have a significantly bigger edge in frc. Humans generally know enough about openings to minimize mistakes for the first 10 or so moves (such that computers playing humans generally are trained to make a couple highly dubious moves to get far away from theory in the opening). engines on the other hand don't go based on theory and are just as capable in 960 as they are in regular chess.
That's hard to answer because the advantage is essentially infinitely large. Engines never lose or draw against unassisted humans. Any modern chess engine, if it plays 100 games against any human (even magnus), will have a record of 100 wins, 0 draws, and 0 losses. This is true both in standard chess and Chess960.
A fair match has never been played between humans and computers. Let's say we have a fair match:
* 100 games, to have some statistical relevance.
* One move per day, so that being tired is no disadvantage (engine can ponder all day).
* Human has access to endgame tablebases and opening databases, like the engine.
* Human can make notes and has a software like Chess Position Trainer, which can min max, like the engine.
If the human is a GM with Elo 2700+ I predict 25 draws and 5 wins for the human. The engine wins 70 games.
> Rating Stockfish against a human scale, such as FIDE Elo, has become virtually impossible. The gap in strength is so large that a human player cannot secure the necessary draws or wins for an accurate Elo measurement.
You're way off the mark here on modern engine strength.
There are many examples of top players playing Leela Knight Odds. And none of them even got remotely close to a decent record. Usually a few draws, and maybe a win. But almost entirely losses.
And that is with knight odds. Without that, zero chance.
reply