The point is you just need to scrutinize the theorem. Not easy either, but still significantly less work than writing the proof.

This. Almost all languages can at least call into c. Which should allow you to do whatever you want.

I disagree to some degree. Tests have value even beyond whether they test the right thing. At the very least they show something worked and now doesnt work or vice versa. That has value in itself.

I agree. I feel there is not a time in history where such a huge diversity of music was produced.

Well the first naturally led to the other. So you can distinguish them, but they are not separate.

In order to get from the first to the second, you need the artificial scarcity laws, and we ought not to keep those.

I disagree. You simply increase the supply of labour by double digit percentage points. Thinking this will not affect the price, all else being equal, is magical thinking.

You're ignoring the other side of the ledger. If the supply of labor increases, but then those people get paid money, then they spend it and create additional demand for labor.

How do you suppose a country with 100 million people can have the same standard of living, if not higher, than a country with 10 million people despite having ten times the supply of labor? Or for that matter that large populous cities can have higher paying jobs than small towns?

Just thinking out loud here, perhaps behavior like this has been more normalized because of the total shitshow that C is. Which followed all these supposedly correct rules.

I don't understand this comment. In everything it does this is a MacBook, not an iPad. It doesnt have a touch screen. It has a fixed keyboard.

IPad pro exists with an M5 as well. And is prices basically the same as an MacBook air.

fair point --- as I noted, I'm not holding my breath, and I don't think it likely, but maybe the above is why some folks think this might happen.

Is there anything truly low latency(sub 100ms)? Speech recognition is so cool but I want it to be low latency.

Agree about the latency requirement.

There's https://kyutai.org/stt, which is very low latency. But it seems not as hackable.

Parakeet does streaming I think, so if you throw enough compute at it, it should be. The closest competitor is whisper v3 which is relatively slow, maybe Voxtral but it's still very new.

The python MLX version of Parakeet indeed support streaming: https://github.com/senstella/parakeet-mlx It requires modification of the inference algorithm. In this implementation, I see the author even uses a custom metal kernerl to get maximum performance. The Parakeet model batch inference logic is simple. But for streaming, it may require some effort to get the best performance. It's not only the depencency issue.

There's a minimum possible latency just given the structure of language and how humans process phonemes. Spoken language isn't quite unambiguously causal so there's a limit to how far you can go for a given accuracy. I don't know where the efficiency curve is though. It wouldn't surprise me if 100ms was pushing it.

Yeah the metric would be the total processing latency after that. I've found that VAD is honestly harder to get right than STT and if that fails, STT only gets garbage to process. Even humans sometimes have issues figuring out when exactly someone is done talking.

On macbook pro - parakeet.cpp is very low latency, under 100ms (76ms) for 60s audio.

I guess the insider information is knowing who the wallet belongs to?

I seriously doubt a person with average intellect can become a world class mathematician, let alone a decent one. Just on grid. I have seen people in college that were tremendously hard working fail math classes and just not understand it. At some point saying they should just try harder is cruel.

The logic in my claim is that the overwhelming majority of people will reach nowhere even remotely near to their genetic potential in literally anything. You can see this in any endeavor where performance can be objectively measured - chess is the obvious one. A 2000 rated player is not much more than a strong amateur, but that already leaves one in like the 90th+ percentile for a game that millions of people work and study at.

It's not like the other 90% of people lack the intelligence or whatever else to be much stronger than they are, but it requires extensive dedication, work, and suffering that many just uninterested in tolerating for the sake of improving at a single domain. I think your example largely proves the point. Anybody of average intelligence can obviously excel at undergraduate math if they're willing to dedicate themselves to it, but many people aren't. If somebody was failing at math it's probably because they were just treating it like you might e.g. literature, and trying to do cram sessions relatively shortly before each exam, whereas by the time somebody gets to stuff like diff eq math starts turning more into a puzzle game that requires developing things on a subconscious/intuitive level.

This was not undergraduate math in my case, but I still don't agree.

I don't think anyone of average intelligence can excel at undergraduate math. It of course depends on the degree and the school. Can anyone with average intelligence excel at an undergraduate math course in community college for their psychology degree? Probably yes. But an undergraduate math course at oxford as part of a maths degree? Not for sure not.

I think you are severely underestimating how much intelligence factors in how fast and even what at all you can learn. Take the opposite end of the spectrum. The US army rejects candidates with an IQ of ~85 or lower. Because they have found this group cannot contribute meaningfully. Let that sink in. Just a drop of 15 IQ points means the US army has decided you cannot be effectively taught anything to a minimum degree of competence. Now consider that the average IQ of a mathematician is 130 (https://realiqtest.webflow.io/posts/iq-by-occupation-a-compr...).

I'm speaking of math majors, or fields with a heavy math requirement, of course. Diff eq is not required anywhere, as far as I know at least, for non-technical majors. As for the army, I wouldn't just hand-wave away soldiering. You're talking about people being put in high pressure situations with ever-shifting dynamics, potentially against a human adversary, where lives are at stake. And they think everybody except the bottom ~32% of society is fit for this task.

I'd certainly expect an average person who dedicates his life to mathematics to end up with a higher than average IQ largely because while IQ is a useful measure, it's not an independent g factor. Studying mathematics is going to absolutely help train your brain in many areas that are also beneficial for performance in IQ exams. So for instance, some studies have shown that each additional year of education, relative to a fixed base, can causally contribute 1-5 additional IQ points. [1] So our person in question would almost certainly expect to see a significant and measurable IQ increase.

And FWIW I'm rather on the opposite extreme of those who argue for some sort of tabula rasa. I fully acknowledge dramatic innate differences between individuals, but I'm largely arguing that such differences only become major factors for people who approach their genetic potential in something, which most people will never get even remotely near, simply because the amount of dedication and sacrifice it takes is something that very few people are willing to accept.

[1] - https://pmc.ncbi.nlm.nih.gov/articles/PMC6088505

The data on this site is extremely dubious.

Well the central post that the commenter made about the army’s iq requirement is trivially fact checked to be untrue. The army doesn’t administer iq tests as part of screening. They do asvab which tests _knowledge_ which you can study for. They have correlated outcomes in that a high IQ usually means a high asvab but they aren’t identical (you can for instance top out an asvab test and practice shows meaningful improvement whereas there is no top iq and if you can practice for it the test is flawed.)