groby_b

Sure, but did it need to be written by AI? Does that solve for the right constraint?

Yes, to get funding.

But they don't want to, because figuring that out is your job as a supervisor.

If you outsource that work to customers/patients, you'll end up with the car dealership model, where the sales rep begs you to give a 10 on every single question including on the interior design so they don't get fired.

That's the part most of this discussion misses. Supervisors exist for a reason. Congrats on your flat org structure, you fucked up an important feedback channel.

There are other phones. They allow publishing PWAs, it turns out.

"you just need radiators which are passive

"Just" is doing a lot of work there. SpaceX is planning to launch 100 GW of compute annually, that comes with ~ 2.5 square kilometers of radiator (assuming an optimistic 800K radiator temp and emissivity of 0.9, double sided)

Go for advanced carbon composites, you can do that with just 5,000 metrics tons or so of material. That's 34 starship launches just for the radiators. We haven't solved assembly, we haven't brought up power panels or core compute. Planned launch cadence that SpaceX hopes to reach end of this year: 12/year.

"Altman doesn’t appear to be a beacon of corporate ethics."

Do those exist? I'm usually happy to see a mild candle flicker in the ethics window.

This is a general belief of our current version of capitalism. As much work as possible is outsourced to the user, the gains are captured by the company.

See also self-checkout etc.

It does. It's called a PWA.

In retrospect, I think my wording might be misleading - I didn't mean to imply the US is doing significantly worse than developed world, just that none of the developed world does significantly worse than the US either.

I.e. I was responding to the original "in 'most of the world' EMS quality is ass.", trying to say that you only get that outcome if you compare apples and oranges (developed vs developing)

8-10 minutes in urban areas is pretty much universal (for the developed world). Rural areas are where differences come in, and even there standards for the developed world are somewhat on the same level.

So, KYC propaganda?

Haven't yet seen a developed nation do worse than the US. If you prefer to compare against developing countries only, sure, winning by lowering the bar to the ground is an option.

This project was likely not written just to be convenient for you.

And if it's got value for you, hey, it's open source. Spend the 15 minutes and 10 cents, it's a bargain.

Write an LLM script to review them. Tell it to find at least three severe issues. Set to auto-reply.

He who brings the slop cannon shall be drowned by slop rain.

They're not just "too zealous", they're ludicrous.

I've had it reject looking at pages served from my local network because it "can't find it with my search tool" and had "ethical concerns about consent for access".

The People's AI Concern Front has gotten the classifier they want, and it's made Claude hilariously useless. I am waiting with bated breath for their next set of revenue numbers. (And happily hand my money to competitors instead)

I mean, you don't have to annotate them :) Would I love to read annotations? Sure - but it's a boatload of work for you, and if it's just meant as a repo for you and your friends, no need to do a ton of work just because an Internet rando asked.

But given the sudden wide audience, a quick "here's what this is for" at the top might be helpful.

Good for China.

Society cannot work with too many corrupt civil servants. Yes, "autocrats", "civil liberties", and yet - the guy slurped up $325M to put his finger on the scale, not to change the model of governance.

I wish we in the west took corruption more seriously, but I suppose we're more interested in cage fights on the lawn these days.

"Truly robust engineering isn’t about what works for most; it’s about gracefully handling the edge cases"

Not really. Truly robust engineering includes a cost-benefit analysis of which edge cases you handle. We don't live in a world of unlimited time & money.

Yes, but we've traded LLM psychosis for agent psychosis, followed by loop psychosis. (Or was it the other way round? Also, we did dark software factories already, right?)

Turns out that if you try to be "15 minutes into everyone else's future", you get better and more frequent trips than any psychedelics could provide.

I think it'd be interesting to hear what you think the goal of the site is.

Is it just rehosting the list, plus a reformatted copy of the papers? I was hoping you'd have at least annotated them with what you'd learned?

The thing is that most architectures aren't amenable to emulation by things you wrote stoned and drunk.

I think this misses a few crucial steps.

Assuming you discover a capsule from the 21st century chances are overwhelmingly that either things have gone great, and you could just ask your AI to create an environment for that old binary (or reverse engineer it), or things have gone badly and you try to bootstrap computing.

The "things have gone average, and I want to run a 200-year old binary" case is I think the least relevant.

But even if we stipulate its relevant enough, this does not solve the issue. The subleq machine does not actually reproduce the hardware environment. A lot of software depends very much on the characteristics of the hardware it runs on. (Timing issue, undocumented but interesting side effects, etc). It depends on bugs in the containing OS. (See Win32 backcompat, e.g.) It depends on quirks and behavior of its input/output devices. Heck, the VM does not even define what "reads a keypress" reads - scancode? ASCII? UTF-8? Value according to numerology?

There's a lot of software that's not amenable to "it's just a single set of numbers in a capsule" with just that underlying VM. Likely most software.

SUBLEQ is a somewhat interesting architecture if you think about "bootstrap compute from scratch", but that's a very different problem.

"Sub-Saharan Africa had the highest cold-related excess death rate"

That's an interesting choice. It conflates cold-related excess deaths in the equatorial lowlands (which struggle to ever get below 70F/20C) and the southern highlands (like Lesotho, which routinely goes below freezing in June/July). Both are sub-saharan.

Of course, that may just be a bad summary, but it puts it onto the "should probably verify the results before I trust it" pile of papers, something that's sadly growing at an ever-increasing pace.

Given it's Meta, I'd put roughly zero trust in the opt out actually being respected.

Pangram is basically a made-up number.

I've tried it on large docs I've written well before the AI times, and that are nowhere available on the Internet (so it can't be a corpus issue) - and it is happily classifying me as 60%-80% AI.

Ah, but you only tell them the things they already know and want to hear.

That filter is crucial in any kind of advisory/consulting work. You're there to confirm biases, in 95% of the cases.

It's not quite that clear-cut. Franklin was pretty clear on the helical structure in both research notes and papers, but she didn't quite nail the overall structure (2 strands with opposing winding, complementing bases).

Fundamentally, she suffered the curse of the experimental scientist - waiting for actual data before being willing to build a model. Watson & Crick postulated ahead based on partial data.

He's under the building doing plumbing or electric one day.

I believe I've encountered that guys electrical work, and it ain't sage :)

(I.e. there are a few of those "fix it all" guys, but they're not always code compliant. They do get stuff done, though)

You understand there's a difference between human lives and the value of a stock, right?

Because otherwise, I don't think this conversation is going to lead anywhere.

It really doesn't. Same as we don't say "written with vi" or "written by Emacs", even if it is intuitively clear one of the two is better.

Relatedly, is it OK with you if somebody bets a few million dollars on your early demise?