choeger

Where does all that money go to, though?

Is there a rich caste of doctors or pharmaceutical shareholders that don't need to work and live off these dividends? Or is the system so inefficient that most people in it aren't contributing to actual health care?

If only someone would properly finish U8 ...

Yeah. I wouldn't be surprised to learn that even the electrons are poisonous there...

U.S. law. It's pretty obvious that neither Amazon nor Google are good options for serious actors that are not the U.S. government. So if they want to make business outside the U.S., they need to dance around the fact that in the end they bow to the will of Washington.

> Why is Microsoft so eager to also be able to know this?

A database of pretty much all Western citizen's faces? That's a massive sales opportunity for all oppressive and wanna-be oppressive governments. Also, ads.

So a question to the experts here: What's the catch?

You're absolutely right. Competency has lost its value.

When was the last time you heard someone praise someone else's competency?

How much did that reimplementing cost and when will the savings exceed that cost?

The question is: What's ownership? How do I ascertain that I own a device and not, say, the guy who just robbed it from me at knifepoint?

From a government perspective, I think the issue is anonymity. In the long run, governments cannot accept ownership of a thing without being able to attribute usage of that thing. From that perspective, as much as you cannot anonymously own a warehouse, you cannot anonymously own a programmable radio device.

From the corporate perspective, it's even worse: They cannot accept you using a device freely if they license you software or data. They would probably be fine if you could prove to them that you were not violating the terms (or vice versa, they could prove when you did), but that probably has a massive impact on privacy.

It's not necessarily that Mozilla could not do it. Just look up Mozilla's revenue sources.

Proposal: Also implement the efficient Hindley/Milner variant used by the OCaml compiler. IIRC it is inspired by garbage collection and much faster than the original.

Version ranges solve the problem of transitive dependencies if libA needs libZ 1.0 and libB needs libZ 1.1 how am I supposed to use both dependencies at the same time when my language doesn't allow for isolation of transitive deps?

You're very optimistic in the potential of these tools. I tend to agree, but I think that they will find their master in formal systems. If productivity raises as you're predicting, the world won't accept 99,9% correct software anymore. There will be demand for 100% correctness.

Regarding the potential economic gains, they're exactly the salary of software engineers. That's a decent amount but not massive.

Compare this to civil engineering, architecture, and craftsmen. None have been replaced because machines let amateurs do something resembling their job.

A parser is trivial. It's XML and you have a schema.

What you want is a compiler (e.g., into a different document format) or an interpreter (e.g., for running a search or a spell checker).

That's a task that's massively complicated because you cannot give an LLM the semantic definition of the XML and your target (both typically are under documented and under specified). Without that information, the LLM would almost certainly generate an incomplete or broken implementation.

Yet, AI agents don't replace software engineers.

Imagine a software company without a single software engineer. What kind of software would it produce? How would a product manager or some other stakeholder work with "AI agents"? How do the humans decide that the agent is finished with the job?

Software engineering changes with the tools. Programming via text editors will be less important, that much is clear. But "AI" is a tool. A compressed database of all languages, essentially. You can use that tool to become more efficient, in some cases wastly more efficient, but you still need to be a software engineer.

Given that understanding, consider another question: When has a company you worked for ever said "that's enough software, the backlog is empty. We're done for the quarter with software development?"

Is there any indication whatsoever that there's even a glimpse of artificial intelligence out there?

So far, I have seen language models that, quite impressively, translate between different languages, including programming languages and natural language specs. Yes, these models use a wast (compressed) knowledge from pretty much all of the internet.

There are also chain of thought models, yes, but what kind of actual intelligence can they achieve? Can they formulate novel algorithms? Can they formulate new physics hypotheses? Can they write a novel work of fiction?

Or aren't they actually limited by the confines of what we as a species already know?

Separate compilation also enables easy parallelization of type checking.

LLMs are to a certain degree compressed databases of their training data. But 42% is a surprisingly large number.

It's only logical that this happens. Just because we can nowadays throw a massive amount of compute on a problem doesn't mean our models are good.

Why are people using transformers? Do they have any intuition that they could solve the challenge, let alone efficiently?

> People who say GenAI isn't vastly speeding up development are just using motivated reasoning, aren't using the tools, or are just bad engineers.

Then where is all that new software? Where are all the SaaS solutions, the games, the new tools?

The productivity increase that comes from working at a higher level of abstraction is a real thing, but it's not going to change the industry.

I wonder if this will eventually lead to increased density and if that then leads to congested bike lanes. Will the cities of tomorrow regulate traffic between individual buildings?

Make no mistake, bikes are much, much, better for urban centers than cars. But the overall problem isn't cars, it's individual traffic in densely populated areas.

Certain policy here in Europe simply assumes that people stay in their surroundings ("15 minute city") and rarely, if ever, visit parts that are farther away individually.

Public transportation, however, is naturally biased. It can be much quicker to get 10km north-south than 5km east-west, or the other way around, depending on the city. And, of course, public transportation is often lacking quality compared to individual traffic. (Taking a bike across a bicycle road vs. getting into a crammed subway train in July, for instance.)

Uh, nope. Your language needs to be able to define these types. So they belong into the stdlib because they are useful, not because they are special.

You might add syntactic sugar on top, but you don't want these kinds of things in your fundamental language definition.

All this has been known in the PL design community for decades if not half a century by now.

Two things are incredibly frustrating when it comes to safety in software engineering:

1. The arrogance that "practitioners" have against "theorists" (everyone with a PhD in programming languages)

2. The slowness of the adoption of well-tested and thoroughly researched language concepts (think of Haskell type classes, aka, Rust traits)

I like that Rust can pick good concepts and design coherent language from them without inventing its own "pragmatic" solution that breaks horribly in some use cases that some "practitioners" deem "too theoretical."

What I don't get about lunch in school is where this assumption comes from that it has to be a warm and complete meal, ideally with starter and dessert.

I am really wondering about that. School is more like an office job than a construction site or other heavy work and thus would not a sandwich and a salad be more appropriate? It would be much easier to offer good quality this way, especially when it comes to fruits and vegetables. The stuff you get when cooking with a limited budget for hundreds of people is seldom good or healthy.

Impressive. Really cool what you did there

I'd add a mechanical brake and teach my kid how to use it. And maybe a failsafe switch that shuts the thing down when no one sits on it?

I can confirm nearly everything you say, and I'd like to add that it's a cultural phenomenon. We don't seem to value competence anymore. I cannot recall when I've heard someone say something positive about another person's competence. Be it a craftsman, industry worker, knowledge worker, or even a teacher. There doesn't seem to be any value in doing a good job.

Lovely. It is well written and has a design I can 100% agree with.

I am curious, though: How will separate compilation work? I think it's an important practical feature but polymorphism (which I simply assume from the presence of type classes) and flexible file/module mappings make it extremely difficult to implement.

There's little value for parametric polymorphism at this level as it requires parameters (not only type parameters, but also value parameters, as otherwise the type parameter are dangling).

It's better to think of these types as different. Maybe "primitive types" or "shapes".

You talked about e-ink in high school in 1950?

So you asked an LLM to translate. It excells in translation. But ask it to solve and it will, inevitably, fail. But that's also expected.

The interesting question is: Given a C compiler and the problem, could an LLM come up with something like Prolog on its own?