This is interesting! I assumed the same thing, so I just skimmed through the game with an engine. The world, on average, was definitely not cheating. As early as move 7 Magnus was outright winning!
But there's an interesting meta in that Magnus played far more passively than he normally would. And so I think he also expected he was probably playing an engine by proxy, and wanted to keep the position completely under control. If he knew the world was legit, they probably would have lost!
I'm still trying to reconcile how it came to be that the world didn't cheat though. Lowest common denominator amongst 140k+ people paired with inevitable chatter of 'Hey best engine move is blah' seems unavoidable.
I think the assumption that more than 50% of people are cheating in online chess is not correct. Another Grandmaster and ex-world champion Anand recently also did a match against 70k people and won.
That's not the assumption at all. The percent of cheaters in online chess is approaching an asymptotic 0 (as a percent of all players) simply because the sites, and chess.com in particular, have gotten very good at culling them.
But things like this are social. I didn't follow this (or even know it was going on somehow) but it seems very safe to assume that somebody and probably multiple somebodies were regularly pointing out and discussing engine moves.
So my only real assumption is that a significant chunk of people would end up deferring to the engine moves rather than their own preference. Of course my implied assumption there is also that a significant chunk of people were involved in the social aspects of this, but I think that's also a fairly reasonable assumption.
Based on a quick skim of the article, I don't think this was, for example, Twitch Chat picking moves, which might enable the social aspect you're referring to - although I'd like to point out the difficulty inherent to being in a room with many thousands of people, all spamming chess moves, and trying to find the one engine move :P
It's not that. It's that once you know the best move you start to see why. And then it becomes hard to get it out of your mind. It's the reason I don't recommend computer assisted game analysis, unless you just want a quick blunder check.
A very non-zero chunk of people also probably would not have even understood that that's cheating if it wasn't really clearly laid out in the interface somewhere. For instance computer assistance in the largest correspondence chess league is legal.
There are writeups about this. The sites score each move, it is extremely unlikely to pick superhuman moves multiple times in a row. Once or twice maybe, but not most of the times.
They do a lot more than just that, but a lot of the process is kept confidential. There's a huge cat and mouse game between cheaters and the sites so they try all sorts of things including only using second tier moves, and even 'blundering' but only in winning positions such that the blunder doesn't risk the outcome of the game. Some cheaters also only use the bot in certain parts of the game. In the extreme case you might have very strong players who are cheating (like in a money event) using only the comp eval. So they don't have an easy job at all.
Here, for example, is one of the more well known and easy meta-indicators of cheating: Humans spend more time on difficult moves than on trivial ones. Cheaters will typically spend a comparable amount of time on a trivial forced move, and the start of an exceptionally deep combination. And an even bigger tell is that said exceptionally deep combo may be followed by a couple of forced moves, yet he will again take just about the same amount of time to play those moves.
Perhaps it is worth considering that this was Freestyle chess and not classical chess. Which means the traditional book moves with which chess engines are trained goes out of window. I am not saying Stockfish cant beat Magnus in Freestyle chess but it makes sense to believe that Chess engines are better at classical chess when compared to freestyle.
But then again, with 24 hour time to brute force every possible combination, I guess chess engines may be better at freestyle when compared to classical chess, due to the sheer amount of creativity and calculation involved.
It's actually the opposite. Humans fare much worse at freestyle chess because they don't have any opening theory and are unfamiliar with the patterns that arise from nonstandard opening positions. Engines don't care much about opening theory one way or another
Source: am rated 2000 fide (partly because I struggle with openings)
Chess engines haven't needed book moves fed to them for a very long time now. Start up any modern engine and let it analyze, and it will only be considering strong theoretical main lines from the first second.
That said, they have been trained almost exclusively on games that started from the normal starting position. But then, so have humans.
It might be natural to jump to immediately think the majority was cheating, but as you rightly point out if they were cheating Magnus would have lost. Human players cannot compete with even a couple hours compute on stockfish let alone 24 hours.
It's impressive that Magnus might have won if The World hadn't forced a stalemate.
> In the Chess.com virtual chat this week, players appeared split on whether to force the draw — and claim the glory — or to keep playing against Carlsen, even if it ultimately meant a loss.
Some irritating aspects to this story. It's a chess.com fluff piece and frankly it would have been disastrous to their bottom line if Magnus would have lost as of course it would mean a significant percent of the world was cheating. I would not be at all surprised if they had measures in place to stop that if it arose.
Freestyle chess has been almost universally known as Fischer random. But of course Fischer being who he was history needs to be sanitized. The ap story is also wrong in its description of how it works since pieces are only randomized along the back rank not "all over the board"
Then you have so called experts racing into this thread to pronounce that cheating had nothing to do with the outcome. On the contrary the accuracy of the world is suspicious and I don't believe chess.com would ever permit Magnus to lose this match so it makes sense that, despite strengthening its position a strong "draw by repetition" faction magically appeared to prevent that possibility.
Magnus Carlsen would get crushed by an engine running on an iPhone 1. Meanwhile the world has access to iPhone 16s. The entire concept is flawed. I'm guessing someone made money off it, though.
> Magnus Carlsen would get crushed by an engine running on an iPhone 1.
Did a quick sanity check here - this seems about right - Carlsen might be at least competitive with Pocket Fritz 4 at similar hardware performance to the iPhone 1, but that discounts the software improvements chess engines have seen over the past couple decades.
I don't know enough about chess, and will take your word for it. What it suggests to me is a deeper question: How do you get 143000 people to all fall in line behind a single machine, or person, making the best decision for them?
If you had a military-like organization and turned 143000 people into calculators led by one (talented) person or a hierarchy, then yes, they would crush Magnus.
No they would not. If you imagine running a computer chess engine on 143,000 humans, it's not even remotely close to the amount of compute you need to win.
Humans don't win by calculation the way computers do. When you have multiple humans working together on chess they don't add up to an ultra-smart human. You are simply as smart as the smartest human in your crew, and that's it.
You could absolutely form a system to harness the power of that many people. It would not happen spontaneously, but it is possible given enough effort. Calculation and memoirzation plays a huge role in chess.
Now I want to see a YouTuber hire 1000s of humans to make a human CPU. Each human can do a single simple task like a redstone block. What would their equivalent CPU clock speed be?
Cheating obviously does happen but on the whole chess is kept alive by people who do it for fun. What would be the point of beating Magnus with a computer? Would anyone get satisfaction from that?
Isn't this the promise of LLM, that with enough data the best answer will surface. The problem is that it needs training against a Magnus Carlsen which is definitely not going to happen.
