AlotOfReading

If you're manually calling malloc in Rust, your code is almost certainly wrong. It might be called as an implementation detail when you create new objects, but std can generally be used as if it were managed automatically.

The problem is that Meta jumped headfirst into the circular financing of datacenter build out, so they're just as screwed if the infrastructure becomes less valuable.

So take 16 independent computers and have them each solve the problem separately. You'd still be saving 80% compared to the paper's benchmarks. It wasn't close.

McSherry does a lot of interesting work on making monotonic/incremental distributed systems efficient (e.g. Differential and Timely Dataflow). Those kinds of systems scale much more linearly.

I would be pissed if my programming language changed as quickly as Claude code does. Languages need to move slowly and carefully, and zig is on the faster end of language development regardless.

I really don't see what's supposedly awful about that loop, but if you want to count down to x instead of 0 you just do:

    for (size_t i = size - 1; i >= x; i--)

User numbers aren't the only factor. Microsoft has a much larger presence in commercial email than Apple. I suspect an outbound email from a personal provider is far more likely to be destined for an outlook inbox than one on iCloud.

They're not the newest kind of literature. Arthurian legends and religious canons are two examples of newer forms, neither of which I think would be typically described as "recent." I could also use the novella, and the anecdote as examples instead.

There are many, many factories that still don't have internet access on the floor, and commercial inference generally has response latencies measured in seconds. I struggle to imagine a factory spending hundreds of thousands for the local compute to run a large model either, given how cheap they are about expenses.

I'm also skeptical that you can cleanly differentiate between "safety critical actions" and "actions", though this is less of a practical concern given how laissez-faire some manufacturers are. For context, I work on safety critical robotics (in automotive).

By what standard are novels "recent"? The earliest novels we have originated not long after the first books (aka codices) appeared. The first modern novel was written at the same time as the king james bible, over 400 years ago.

Only the first is written by Banks, to clear up any ambiguity. They're very different books, just similarly named.

Unfortunately there really aren't any other books like the culture series. You might enjoy Banks' "The Algebraist" and the completely unrelated though similarly named "The Alchemist".

No one says "I gave up on eating because restaurants kept serving me spicy food". You just order different food. A short story that's a couple pages long isn't going to waste them describing the color of buttons, and not every novelist is Tolkien.

As a follow-up to this comment, I sat down and implemented a better method myself, then looked at the specification code to discover they did exactly the same thing.

They're definitely a thing. The scam is people showing up and painting them without approval, demanding payment afterwards.

It really should have been defined in terms of fixed point like HAWK was. Supposedly you'd need a large number of bits for it, but I'm not sure I find that argument particularly convincing given that floats are a kind of fixed point in log space, with a bunch of asterisks that don't help falcon much.

You don't have to assume that. We can take each randomly oriented noodle and orient it correctly in O(1) as a preprocessing step. Since the complexity would be additive, the overall complexity remains O(N).

Dropping spaghetti is an O(1) operation. Once that's done there's a straightforward O(N) sort by removing noodles in the order that they're intersected by horizontal plane descending from max_noodle_length to the table.

I think it's interesting to note that in one year we've gone from they're not even close [0] to arguing whether open models are only as good as sonnet or opus.

[0] https://news.ycombinator.com/item?id=44623953

The numbers in this are insane, but if they work out that's a huge financial moat for Nvidia, on top of their existing software and hardware moats. Building an (independent) competitor seems basically impossible if you're also having to financially backstop your customers.

You can write programs of a similar spirit in most languages. C has a programming culture that's much more tolerant of it than say, Python though.

Zig could probably support a similar contest as it grows up.

   ...most popular projects require some level of support for dynamic code execution before reaching their usable state.

None of your examples require arbitrary script execution. You can specify them all declaratively, like Bazel forces you to do. I don't think that package managers should be doing the job of a build system though.

How does embedded require deep understanding of CS? I work on embedded systems (safety critical no less) and while I can point out how they're actually distributed systems or x theorem is applicable all day, my coworkers don't actually need that info to do their jobs.

C and the various C++ subsets called "C with templates" would be much less common than they are if "deep understandings of CS and DS&A" were required in day to day work.

    There are rather concrete problems that strictly prevent it from being possible to efficiently map graph (object) database access patterns to a relational database.

Do you mind going more into that? Naively, it seems like prolog/datalog describe graphs pretty well and they're inherently relational. Relational databases have typically just optimized for row-oriented OLTP uses instead of columnar OLAP, but there's nothing inherent preventing one or the other. They're duals of each other.

"good" can take lots of different meanings. Generally though, I want as little code as I can get away with. A majority of code lifecycle cost isn't in writing it.

Of course? I'm still better than sonnet or opus, just slower and much more expensive.

Sometimes it takes me a day or more to find the one line fix or abstraction necessary, while claude can hammer through a hundred line fix in under an hour.

Some manufacturers are attempting to address this via apps and account syncing, which I think is still much worse than carplay/AA.

The argument structure is interesting, and reminds me a lot of Solomonoff Induction, but constrained into NP by the assumptions. I'm not sure the front half is enough to support the back half of the paper arguing that the current LLM craze means firms are actually running collusion detection algorithms, even unintentionally.

Leaving aside any quibbles about the quality of Tesla's offering, look up the story of CARIAD [0]. The scale and cultural inertia of OEMs is hard to overstate. That change can't happen quickly.

Toyota had the better idea, keep the software folks an independent business unit with internal competency/ownership, and only allow them to fix future platforms. Woven is apparently having great success inside Toyota. The side effect is that those changes still aren't apparent even years later.

GM had the same idea, but killed it recently.

[0] https://www.germanautopreneur.com/p/cariad-volkswagen-softwa...

Yes, they did because various countries have talked to the US about expanding it. The problem is that South America is an enormous place, whereas Panama is a narrow isthmus. It could have been done with some amount of money, but that opportunity ended in 2010 at the latest.

Pretty much every major manufacturer has a bay area satellite office these days. Toyota has a research center in San Ramon and a woven office near the broadcom campus in Palo Alto. GM has an office in SOMA and in mountain view. Ford has an office literally a block away from Tesla HQ. Mercedes used to be near Cariad at moffet field, but now they're down by SJC. Nissan is in Santa Clara.

Need I continue? And Tesla is generally understood to pay on the lower end among those companies.