Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[silverspawn]

Carlsen tweeted this: https://twitter.com/MagnusCarlsen/status/1566848734616555523

Which is a pretty obvious implication that he played against a cheater when you get the idea, especially since he's never withdrawn from tournaments before, isn't blaming the tournament itself, and usually doesn't react this way to losing.

People commenting on this give me very misanthropic vibes. There are so many comments which are expressing outrage at the fact that you would accuse someone else. It's such an attractive way to do maturity signaling, and man do people love to do maturity signaling. I see almost no-one doing proper Bayesian reasoning here. If you did, I'd say cheating looks very probable.

I weakly predict Hans' performance to noticeably decline in the remaining tournament since he won't have the balls to keep cheating but isn't actually as good.

[silverspawn]

I am also biased because I think I have good social detectors for this kind of stuff and Hans gives off super bad vibes to me. So I want to be right to validate this nebulous ability.

[silverspawn]

It is disgusting how everyone is piling on Hans, a 19 year old with not a shred of evidence he cheated.  These people not only have zero evidence of Hans cheating, but they can't even think of how it would be possible for him to have cheated!  Still, they continue to launch accusations.   Hans' play wasn't computerish.  He beat Magnus fair and square and Magnus got salty about it.  That's what happened.  Magnus tweeted out, making it subtly OBVIOUS that he thought Hans cheated while pretending he was being classy.  Not cool.

[silverspawn]

Imagine being a 1700-1800 player, making a new account, and then joining a <1500 tournament. Pathetic.

[silverspawn]

Hah, I spent 3$ to share my take on the probe-is-the-strongest-sc-unit as a donation and artosis agreed. I feel validated.

[silverspawn]

I do feel really bad for spending money in frivolous ways, but it was worth it.

[silverspawn]

Speaking from my mafia experience, Hans' interview was similar to someone drawing scum for the second time. But most people are eating it up.

[silverspawn]

seems like hans is losing his game with the white pieces

[silverspawn]

The World in your Head from Steven Lehar is super fascinating. It's all this evidence that is obviously there and obviously important, yet everyone  (including me) by default ignores completely!

It's all about what the brain actually does in terms of visual processing. We don't just have a 2d image, we have a spatial 3d representation of our environment at all times. To see this, you need only stand up in your room and then close your eyes. While the image disappears, you still have an "amodal percept" or "invisible model" of your environment. You know where each object is. You can observe how this changes when you turn around.

And when you begin to examine all of the visual functions in depth, like all of the gestalt laws, it becomes a neverending stream of evidence that the brain uses holistic, analog computation to do all that. Which for me is an absolute goldmine of evidence. This will become a major part of the sequence.

Standard computational models in the literature seem to be like, here's one isolated thing the visual system does, here's a model that solves it, probably with questionable runtime complexity like quadratic in the number of points on the visual field. Often only works in simplified settings and pure cases. A different such model is required for every detail. I've read one of them so far that only deals with how to group points together, which is already quadratic complexity and probably one of the easiest problems.

Steven Lehar is like, here's a paradigm that solves everything at once with some significant but manageable complexity.

[silverspawn]

It begins to feel more and more insane to assume that the brain only uses bits for computation. I mean why on earth would it do that? Why should that be the default assumption?

[silverspawn]

Hans losing another game with white. sweet affirmation.

[silverspawn]

Getting caught up in this drama is really stupid because I should have been asleep 100 minuts ago, but I can't help it now.

[silverspawn]

This case so wonderfully demonstrates the difference between maturity signaling and pattern matching and the much harder work of actual bloody rationality. Yes, most of the time the less sensationalist take is the more rational one. "Don't trust conspiracy theories" is a good heuristic. It works... until it doesn't work, and this is one such case.

[silverspawn]

There are no actual shortcuts to rationality techniques in the general case. You actually have to examine the evidence.

[silverspawn]

So the overarching goal of my sequence can be summarized as "convince people that the brain uses holistic effects" where "holistic effects" means "computations using elementary objects significantly more complex than bits". Anything like a self-untangling repulsive wire where every atom pushes on every other atom on the surface, and the sum of these interactions is computationally relevant, is holistic. Conversely, digital computers (at the level of electronic gates) probably never have anything that depends on more like 8 inputs, and usually it's 2. I actually don't know what the highest number is, that would be interesting. But even if it's larger than 8, the function by that object is still fairly primitive.

One of the things I think I'm missing is the intuition for *why* the brain wouldn't use holistic effects. Everyone seems to think this is obvious, but why? Whenever you *can* use holistic effects, they're extremely computationally useful. By making a wire surface repulsive, you're essentially getting a ton of computational power for free. I mean, imagine simulating the repulsive force between every pair of atoms!

The obvious answer is, "there's no way to use holistic effects in a useful way". Like, yes, the raw computational feat done by an untangling wire is massive, but it's not useful since intelligence isn't about untangling wires. And okay, that *could* certainly be the case. Like, it wouldn't be a-priori shocking if it weren't possible to use holistic effects flexibly as part of intelligence. But... it also wouldn't be shocking if it were possible. I mean, consider how hard it is to build something even slightly interesting out of logical gates. How do you know that a similar amount of effort can't manage to utilize holistic effects for their advantage?

One way is to use myself as a data point since I didn't use to think that the brain uses holistic effects. Unfortunately, my reason was a mixture of "never thought about it" and "I'm sure I'd have heard about it if it did". Probably that's many people, but I don't want to straw man the opposition by assuming it's everyone.

[silverspawn]

It seems to just be a sticky equilibrium that neuroscience has run into. Apparently there was this thing called the neuron doctrine which was roughly about this question (kind of says "neurons are little modular units"), and then there was some evidence in favor of the doctrine, and ever since then the concept of holistic effects was abandoned. That remains to this day even though it seems like we now have lots of things pointing in that direction.

[silverspawn]

One of the key problems, maybe the biggest problem, is that people don't even have a conceptually crisp understanding of "holistic effects". It requires some thought to realize what is even asked, so if even getting to the question is a hurdle, that probably plays a big role in the failure to answer it correctly.

Wikipedia says,

The neuron doctrine is the concept that the nervous system is made up of discrete individual cells, a discovery due to decisive neuro-anatomical work of Santiago Ramón y Cajal and later presented by, among others, H. Waldeyer-Hartz.[1] [...]

This isn't rigorous enough to be well-defined. Might be true as stated.

[silverspawn]

Even the term "holistic effects" is non-universal; everyone seems to use slightly different language. I've also heard "computational relying on inner states" and "spatial computation across a spatial medium".

[silverspawn]

woah merge hype!!

[silverspawn]

doesn't matter what anyone says, the insistance of background music in youtube videos is one of the worst things on earth today

[silverspawn]

I want god-tier powers to stop it

[silverspawn]

Every single instance is a personal offense and a self-insult. It's offensive because it suggest that that I'm so shallow that I require constant entertainment to sustain my attention, and it's a self-insult because it implies that the material is not interesting enough by itself.

[silverspawn]

So the merge is in 14 and a half hours. That's 7.30, so I'll still be asleep. how anticlimactic

[silverspawn]

Merge goes extremely smoothly, and ETH... goes slightly down? I don't get it. Shouldn't there have been a risk premium?

[silverspawn]

Everyone said it wouldn't go up much, but I thought it would go up like a few percentages.