Veedrac

The simplest explanation is surely that China is new to having a highly educated workforce and the ‘failed for some fifty years’ claims the article makes don't mean much given it.

I don't think you need to entrain market arguments or whatnot to this, when it's only in the last decade China was a strong advanced manufacture player, and turbines are the kind of project that you probably wouldn't expect to go much faster than that regardless of the demand.

I have libertarian enough tendencies to think that if a person wants to self-operate, or pay for an operation that doctors are telling them is not justified given the evidence, then they should have right to do it. But I don't think that's what people normally mean when they say that eager screening causes harmful overdiagnosis.

So your suggestion for indeterminate scans is more scans?

The solution to imperfect evidence is consistent and calibrated risk estimation of both disease and intervention.

This is fair, but I think it's better stated as you did than couched in language suggesting it's a matter of principle.

This style of argument has always bothered me, because the correction to misdiagnosis or mistreatment is not to stop looking, it's _git gud_.

For sure, we have to be realistic about what processes will systematically have error, and if we can't stop a doctor from doing bad things with a piece of data we should shield them from it, but the tools to make scalable, calibrated risk estimates based on large data dumps is getting better every year.

Wow this paper is bad. I was expecting little and received genuine crackpottery.

It's hard to critique this paper directly because its claims are so incoherent and decorated with so much obnoxious verbiage[1] that people aren't going to believe me when I point out what the claims actually are.

Regardless, this is their paper:

First, they conflate the substrate with the presentation layer. Then, they point out that Turing equivalence means you can run an LLM on anything, with a pointless aside where they nerd out about making a logic gate in AoE II. This lets them conclude that you can use anything as the presentation layer.

Then they claim that it's natural to ascribe human-like attributes to outputs from some presentation layers, like abstract letter symbols on a computer screen, but not to most other things, like patterns of goats on AoE II, or LEGO. Yes, this seems to imply if your partner writes something heartwarming to you using LEGO, you're meant to laugh at them and point out how LEGO isn't intelligent so this isn't evidence of anything.

Then they do a thing where they say that assuming substrate independence is true (or false) prevents proving whether substrate independence is true or false, and from this, but just by vibes AFAICT, make it sound like everything one could learn about attributes of a system from its outputs is circular.

Then they write a bunch more incoherent text and mercifully then it ends.

[1] 'from an epistemic perspective, we argue that a generalised conclusion such as that necessarily requires a well-designed experiment' — the whole thing is like this.

You don't have to argue the opposite of a prediction to disbelieve sources with poor track records.

I broadly agree those are good to consider (a typical requirement for my own designs is to fit the KIK format as-is), but I think you're being a little too absolute. Enclave deserves some recognition here for getting pretty close. It's just a sidebar and a cam, totally normal lock components. The simpler cylinder-in-cylinder designs are also mostly just hardware that multi-shearline locks already have for master keying.

A few locks like you describe are Enclave, andy pugh's, Built Different Design's, Carl L. Lambert's, and Michel Robert's.

You can't normally describe it as profit because stocks by rule trade at a fair price, but it's surely reasonable to consider in this case GME making a profit from the squeeze given they were selling a good well above fair acquisition price.

Quote from bottom of page.

We understand that some may wish to sensationalize a public crisis like this, but we would implore anybody to consider the ethics of the situation before publicizing this matter to a wider audience than is already exposed to it, or interjecting with prying questions.

I wasn't saying you would have that experience, I was saying that the reason people act like transducers are unique is that transducers are an unconventional place on well worn ground.

Ultimately, yes, everything bottoms out, most special tricks seem less special the more you understand about them, because it's programming and Turing Equivalence is the bedrock the whole field rests on. But the average person learning about transducers is not going to spot how closely related it is to other things that already exist.

I'm happy to elaborate on any part of the terminology if you're curious, but tbh I mostly wrote it for myself because I thought the framing was novel and wanted it noted down somewhere.

Fundamentally, there are two ways of representing iteration pipelines: source driven, and drain driven. This almost always maps to the idea of _internal_ iteration and _external_ iteration, because the source is wrapped inside the transforms. Transducers are unusual in being source driven but also external iterators.

Most imperative languages choose one of two things, internal iteration that doesn't support composable flow control, and external iteration that does. This is why you see pause/resume style iteration in Python, Rust, Java, and even Javascript. If that's your experience, transducers are a pretty novel place in the trade-off space: you keep most of the composability, but you get to drive it from things like event sources.

But the gap is a bit smaller than it might appear. Rust's iterators are conceptually external iterators, but they actually do support internal iteration through `try_fold`, and even in languages that don't, you can 'just' convert external to internal iterators.

Then all you have to do to recover what transducers give you is pass the object to the source, let it run `try_fold` whenever it has data, and check for early termination via `size_hint`. There's one more trick for the rare case of iterators with buffering, but you don't have to change the Iterator interface for that, you just need to pass one bit of shared state to the objects on construction.

Not all Iterators are strictly valid to be source-driven, and while most do, not everything works nicely when iterated this way (eg. Skip could but doesn't handle this case correctly, because it's not required to), but I don't think transducers can actually do anything this setup can't. It's just an API difference after that point.

Do you not... remember? The US life expectancy is 79 years. 7.9 years ago was late May 2018. The best LLM was... wait, there weren't any. There was ELMo, an embedding model. It wasn't just not smart at agentic coding, it wasn't even just not smart at writing code snippets, it wasn't even just not smart at answering questions of any kind, it wasn't even just not good at producing a coherent output, it wasn't even just not good at producing coherent sentences, it was _not even the point where people thought unconstrained text output was a thing machines did_.

There is no step along the ladder which has remotely evidenced or supported that the next step is going to be ten, twenty, a hundred times harder than the last step on the ladder, but a constant chorus of people singing at every moment, each moment wrong, that the next step is the one.

It turns out there is literally no amount of being publicly right about a longshot bet sufficient for people to conclude you hold your beliefs because you think they are true.

More than anything it's a supply limit. Solar is consistently scaling about as fast as any manufacturing industry scales. The TAM is just big.

I think the idea is that in an always-on display mode, most of the screen is black and the rest is dim, so circuitry power budget becomes a much larger fraction of overhead.

There's a very simple solution to this problem here. Instead of wink-wink-nudge-nudge implying that 100% is 'human baseline', calculate the median human score from the data you already have and put it on that chart.

If you don't like the word 'actuality', I can rephrase. Many worlds is just the claim that physical reality materially evolves in correspondence with the Schrödinger equation.

If you want to quibble over what it means for something to be material, go ahead, but unless you can tie it to some specific claim being made about QD I don't really know what the exercise gets you.

Many worlds is just the claim that the Schrödinger equation holds in actuality.

I don't think QD makes decisions 'uniquely'. Take this quote,

The step from the epistemic (“I have evidence of |π17〉”.) to ontic (“The system is in the state |π17〉”.) is then an extrapolation justified by the nature of ρS⁢ℰ: Observers who detected evidence consistent with |π17〉 will continue to detect data consistent with |π17〉 when they intercept additional fragments of ℰ. So, while the other branches may be in principle present, observers will perceive only data consistent with the branch to which they got attached by the very first measurement. Other observers that have independently “looked at” S will agree.

https://pmc.ncbi.nlm.nih.gov/articles/PMC9689795/

Emphasis on "the other branches may be in principle present" — the claim at least in this paper can't be that all branches agree, just that they agree locally.

My confusion is that this is just Many Worlds / the Schrödinger equation, and Quantum Darwinism doesn't seem to add anything that wasn't already obvious by inspection. But after reading more, I think that's kind of the point? It's ultimately just an argument for why the Schrödinger equation produces these locally classical regions, plus a bunch of overly flowery prose and dressing up in invented jargon that can mostly be ignored. I think the article failed to ignore that second part and ended up confused.

Maybe I am just out of my depth, but I don't understand what problem quantum Darwinism is solving. The Schrödinger equation already explains why observers seem to agree: the ones that don't are separated from each other.

This article is making some pilot-wave-like claim on top of quantum Darwinism that while the Schrödinger equation is real, all the 'real realness' exists in some pointer to a specific location inside it. Why does it do this? Where does this claim come from? At least collapse theories allow that the thing the Schrödinger equation is modelling is actually real up until the part God gets out his frustum culler.

This is missing the primary reasons insider trading is bad, which are that it's an information theft incentive against employers, and worse, that it's a sabotage incentive.

From what I've seen, models have hit a plateau where code generation is pretty good...

But it's not improving like it did the past few years.

As opposed to... what? The past few months? Has AI progress so broken our minds as to make us stop believing in the concept of time?

This rule isn't internalizing an externality.

Have you been around a Waymo as a pedestrian? Used one recently? I have never felt as safe around any car as I do around Waymos.

It can feel principled to take the critical stance, but ultimately the authorities are going to have complete video of the event, and penalizing Waymo over this out of proportion to the harm done is just going to make the streets less safe. A 6mph crash is best avoided, but it's a scrap, it's one child running into another and knocking them over, it's not _face jail time_.

the companies owning and operating those AIs would go out of business as no one would be able to afford the products made by the AIs

What do you think money is...?

Money is a way to indirectly trade labour and goods. If a job is automated, that labour doesn't disappear into the aether, it's still in the tradable pot of total goods and services. You cannot empty a pot by filling it. A world where a company though automation has made there nobody else to productively sell to is a world where _by definition_ they own all the output that they could otherwise have traded for.

The volume of space from the ground to 50,000 feet is about 200x smaller than the volume from the Karman line to the top of LEO alone (~2,000 km).

Volume is the natural way to assume space scales, but it's incorrect. Two planes can fly parallel, side by side. Two satellites cannot orbit side by side.

In the limit, if Earth had a solid ring of infinitesimal width, it would take zero volume but all orbits.

The paper says we are 2.8 days away from a collision. It doesn't say we're '2 days away from kessler'. In fact, the paper explicitly warns against your interpretation.

We emphasize that the CRASH Clock does not measure the onset of KCPS, nor should it be interpreted as indicating a runaway condition.

The stock market learns from experience, because it's made of people who learn from experience.

Imagine an investor's experience with TSLA. From the beginning, they're flooded with news reports about 'fundamentals' this, 'fundamentals' that, about how Tesla would imminently collapse, how it's a scam, yada yada. Said investors _constantly_ see themselves being right and those skeptics wrong. Tesla is in fact disrupting an industry. They really are just continuing to scale. Marginal profitability keeps going up. Their cars keep getting better. FSD keeps getting better. The competition that people kept pointing at kept failing to materialize. None of this seems to change the skeptics' byline.

Tesla is actually in a materially worse position than it was a few years ago, by many metrics, but the stock price isn't set by 'fundamentals', it's set by the people setting demand for the stock. With TSLA, this is disproportionately going to be people who have learned to and gotten rich from ignoring the people loudly telling them why investing in Tesla is a bad idea.

A market will correct eventually, but corrections either require people to change their minds or run out of capital. Neither has happened yet, so the market can't correct.

Wikipedia has a section on this that I thought was presented fine.

https://en.wikipedia.org/wiki/Lindley%27s_paradox#The_lack_o...

Indeed, Bayesian approaches need effort to correct bad priors, and indeed the original hypothesis was bad.

That said. First, in defense of the prior, it is infinitely more likely that the probability is exactly 0.5 than it is some individual uniformly chosen number to each side. There are causal mechanisms that can explain exactly even splits. I agree that it's much safer to use simpler priors that can at least approximate any precise simple prior, and will learn any 'close enough' match, but some privileged probability on 0.5 is not crazy, and can even be nice as a reference to help you check the power of your data.

One really should separate out the update part of Bayes from the prior part of Bayes. The data fits differently under a lot of hypotheses. Like, it's good to check expected log odds against actual log odds, but Bayes updates are almost never going to tell you that a hypothesis is "true", because whether your log loss is good is relative to the baselines you're comparing it against. Someone might come up with a prior on the basis that particular ratios are evolutionarily selected for. Someone might come up with a model that predicts births sequentially using a genomics-over-time model and get a loss far better than any of the independent random variable hypotheses. The important part is the log-odds of hypotheses under observations, not the posterior.