concreteblock

I'm not the OP, but that is part of my belief system. It may seem ridiculous, but I'd be happy to discuss.

> But you didn't choose a compelling form of words to advocate your interest to other people. Why not?

I thought it was a good question and the wording was compelling enough.

I think it's fundamentally a question of incentives. A mathematician's career prospects and status depends heavily on being able to prove things that others can't. Why spend extra time helping your 'competitors'? Not saying this is ideal, but it seems to explain many behaviours in the community, and it seems unavoidable given the ever-increasing competition for jobs.

The heisenberg uncertainty principle is a different type of uncertainty than the uncertainty modeled by probability distributions.

First part makes sense to me. But second part doesn't. I remember learning and successfully using the p=h* f/c formula in high school physics, but what is the justification for this?

And if the formula only holds for photons, why can we say that frequncy = constant * momentum for other particles?

I may have missed this, but doesn't this just raise the question:

Why is the fourier transform of position equal to momentum?

I.e why is position conjugate to momentum?

More generally, why would the fourier transform of an observable be another observable?

If you unwrap a cone you get a circular sector. Similar idea.

Who sold him those shares?

IMO if all it takes is a few simple substitutions like “physics” -> “known laws of physics” to make the article or title make sense, then it’s unfair to say that the author has missed the point. C.f. Reading the strongest possible interpretation and all that.

That's a good point. I guess our current mathematics is not good enough to say much about the macroscopic behaviour of large interacting models.

You are missing the point the article. Author is not trying to argue that AI is 'harder' than physics, like a freshman cs major might argue with their physics friends.

Author is talking about how our physical theories, such as QFT, currently have more predictive power than any theories we currently have about machine learning/deep learning.

(Author has a PhD in theoretical physics).

The title is 'clickbait' for sure, but not neccessarily incorrect. After all 'simple' can mean different things to different people. And the author clarifies what he means by 'simple' in the rest of the article.

Titles have to be short, and as such they can't hope to represent the contents of the article completely accurately. If you wanted to do that you would have to make the title equal to the article's contents.

Based on the parts which I've read so far, a more accurate title would be 'Why some currently hot parts of AI not well understood, and some parts of Physics well understood?'

I think the original title is an ok approximation of this.

Why would you even bother responding based on just the headline of the article? The author is not using 'simple' as a synonym for 'easy to learn'.

I don't really care about winning any global arguments. I see bad logic and I try to point it out. If you didn't like that some of your points were easily rebutted then you shouldn't have written them. Leaving them unchallenged makes your position seem artificially strong.

I couldn't resist bringing up the poker bankroll example because I think your in-game-poker example was poorly chosen. To me, it looked like you came up with a situation where the criterion obviously had no hope of being applicable and then used it to argue that the criterion is useless. E.g. I could find a whole list of things for which calculus is not applicable, but that would not be a good argument for 'calculus is useless'. The example I gave is at least closer to the assumptions of the Kelly Criterion.

I think the main thing I wanted to do was to correct the misconception that Kelly is only maximizing expected log utility, because it is a shame if someone (including other readers) thinks that the Kelly Criterion is just a fancy name we gave for the argmax of E f(S) where f happens to be the logarithm.

After all this, you (and other readers) might still conclude that the criterion is useless. But the set of justifications, and maybe the certainty, in that position, should change.