> while being only marginally better.

It's only marginally better in the things it's actually comparable to. A\ models are MUCH better in many more things; eg: things Kimi/etc. didn't distill.

For those things the difference is like a cliff.

That's a baseless claim that borderline reads like shilling. Do you have any proof of that you wrote there?

Safety from what? Competitors? That sounds like a product decision. They're puking on any requests that could be used to create LLMs or competitive products.

I would guess prevention of using Claude as a pentesting or hacking platform. This could mean that every script kiddie out there would be a massive risk.

To prevent their models from doing harm in dual-use contexts including CBRN or by accelerating research in authoritarian-backed AI labs.

Anything to prevent mecha ai hitler. At all costs

> I plan on using this as a sort of benchmark for future AI discussions: "how do you plan on separating data from instructions?"

I'm reminded of the thing that HN drools over constantly with LISP languages.

To a corporation?

> It's similar to the gates quote about RAM (I think it was 64KB?)

640, and Bill Gates said he either never said that, or at least never remembered having said it. I think there is no evidence anywhere that he did.

https://www.computerworld.com/article/1563853/the-640k-quote...

That exact quote? No, never. He said something like: current computers at the time had 64kb of RAM, so the OS was designed with a limit of 640kb, and he believed this would give them 10 years of future proofing. As it happened, that limit was reached much faster, in about 6 years.

MS-DOS didn't create that limit; the physical memory map of the 5150 did. So Microsoft (and Gates) would not have made that decision.

You are right. The quote must have been slightly different then. I'm sure about the 10 years part.

How much time is `(random-uuid7-bytes)` taking?

An insignificant amount for the comparison (why I didn't mention it), it's a fast implementation and the JVM C2 JIT has kicked in by the time the first batch has completed.

I can't believe I had to scroll down to this far to see someone making this point.

Also INSERT speed instead of SELECT? Typically most time is spend in SELECT or UPDATE.

Although not as prominent as insert SELECT and UPDATE both benefit from page cache locality, assuming rows that are stored near each other are often selected/updated together.

> But then why not just use Rust?

The BEAM?

> With cronie the system needs to be running at the scheduled time, but systemd timer tolerates this and runs the service as soons as the system is available.

Cronie doesn't have a `@reboot` meta-trigger?

Is your deliberate misspelling of "Macintosh" spell-check or the same sort of intransigence that compels some people to misspell "Micro$oft" thinking they're clever?

No, sorry, it was just a typo.

The last time when I had an Apple computer was 2 decades ago.