This is horrifying, terrifying, and sadly well-known even to those who superficially monitor such things.

Popular media: More than one US film/tv studio has "lost" or "suffered a mysterious fire" in un-digitised archives, destroying the lot, during battles to preserve. The BBC sued Bob Monkhhouse for preserving material it destroyed. In Britain, it has been no better. Fans of the British TV series "The Avengers" can only see old episodes because armies of previous fans descended on rubbish tips and, at great risk to themselves, collected as much film as possible.

General history: Places like the John Ryland's Library and the British Library have suffered with rescuing archives at risk of becoming submerged or destroyed by mould. The Archimedes Palimpsest was partially destroyed by one collector filling in the pictures with coloured pens and by another collector allowing the book to be severely damaged by mould.

The National Archives have mysteriously "lost" a great many files over the years and are only digitising those they've retained at an incredibly slow rate. I know because I've personally forked out several hundred to get just two scanned, all because politicians far prefer frippery to archiving. We've absolutely no idea how many of the manuscripts held in other archives are still in usable condition because nobody bothers to check.

It's not just limited to archives, of course. The US has, over the last couple of decades, demolished numerous buildings within the US that are over 300 years old because malls produce profit and ancient structures don't. (They also then complain they have no history...) The Space Shuttle is to be taken to Texas for a PR stunt, which will require it being dismantled and those things aren't designed for that. There is no guarantee any of it will survive the journey. All because PR matters and preservation does not. Other countries? The Louvre... well... probably best not to talk about that utter disgrace. In Egypt, 3000 year old gold artefacts are routinely melted down so the conservators can pocket some extra cash.

It's at times like this that Kenny Everett's general comes to mind.

I have had testing centers send me cameras in the mail to take a test. This was for an online college, and they sent two cameras. One was meant to be mounted to the monitor and the other was supposed to be behind me but offset to the side a bit. They combined to give a full field of view of my desk and the surrounding room. This is a pretty good solution, and these cameras are quite cheap now. Requiring the user to pay for them so they don't have to come into the testing center isn't asking a lot.

I want an accountant who knows how to game the tax code. Not one that knows how to game the testing process. Those are two very different things.

Capitalism is very wasteful, but so far the waste of capitalism has always outperformed more centrally controlled economies. That said, I agree it likely won't scale forever. The size and power of large consolidated global companies, the easy movement of wealth by extremely wealthy families, and how easy it is for the wealthy to wield populism as a weapon to maintain political power have arguably already pushed capitalism to its limits.

This LLM "spasm" is capitalism at its best. Massive funding into hundreds of new companies, 90%+ of which will fail in the next decade, is how capitalism has pushed economic growth forward for the past few centuries. That isn't the problem; the problem is our inability to spread out the wealth created by capitalism. It doesn't seam like our current combination of political and economic system in the US can handle this problem. When the status quo breaks, the result is almost always worst than the bad situation that caused the collapse, so if we are at the breaking point it's going to get a lot worst before it gets better. But perhaps our grandkids will appreciate us tearing the current system down.

That won't scale. Our world runs on plenty of systems that use statistical models to automate decision making. We prioritize how often we verify these systems. We already can't verify every single decision made by statistical models, but when it is something as benign as your Google search results we are fine with only verifying a small sample to do quality control. The same will be true of most code written by AI coding tools in the near future.

I said that I didn't need OpenGL or DirectX in the 90s because I was quite happy fine tuning my own graphics libraries with C and x86 assembly language. I also didn't want to learn early game engines for similar reasons. But I obviously couldn't keep up with progress by using the last generation's tools. I assume the same happened for early software developers when early compilers first started gaining popularity.

I don't think AI is going to replace software developers, but I doubt there will be many developers in 10 years that do almost any programming without heavy AI assistance. It's likely they'll be some niches where AI assistance isn't useful, but the other 98% of developers won't be able to match their colleagues capabilities and productivity without AI coding tools. Just like I couldn't keep up with modern game developers if I was still writing an entire game with my own C and x86 assembly language libraries (although in all honesty, I've moved to the healthcare industry now).

Well, the obvious ones:

No built-in instruction-level or block-level parallelism
Array/vector operations are highly inefficient
Multiprocessing is pretty feeble
No CSP, you have to use OS primitives which are often unsafe
No formal contract system, best you can do is show statements don't do anything bad, you can't show functions do what you intend
Heavy software verification is difficult to impossible

That reminds me, I should really invent a language called "browns" so that Microsoft can put a # at the start.

On the other hand, Microsoft trying to write an OS in Rust will reveal absolutely every defect present in the language.

Unless they write their own, which is quite likely as they'd rather have defects they can ignore.

Abandon hope, all ye who press enter here.

Yeah, one of the things I like about Claude (and Gemini 3 as opposed to 2.5) is that they really clamped down on the use of "Oh, now I've got it! This is absolutely the FINAL fix to the problem, we've totally solved it now! Here, let me write out FIX_FINAL_SOLVED.md" with some half-arse solution. And yep, the answer to going in circles is usually either "nuke the chat" or "switch models".

Now, yes, there are predictions that you could get a supermassive black hole launched into space, especially during a galaxy merger if the velocity of the smaller black hole exceeds the escape velocity of the combined galaxy.

But I'd be wary of assuming that it's a launched black hole, unless we can find the merger it comes from. There may be ways for such a black hole to form that cause the stars to be launched away rather than the black hole being flung, and if a galaxy isn't rotating fast enough to be stable, one could imagine that a sufficiently small galaxy was simply consumed by its central black hole. Both of these would seem to produce exactly the same outcome, if all we have is the black hole itself and a velocity.

I'm not going to say either of these is likely in this case, or that astronomers haven't examine them (they almost certainly have), but rather that we should be cautious until we've a clearer idea of what the astronomers have actually been able to determine or rule out.

Yeah, when thinking of the typical air fryer market, think "working mom with kids who wants to serve something nicer than a microwave dinner, but doesn't have the time for much prep or waiting". You can get those mailard reactions that microwaving doesn't really get you, nice crisping and browning of the surface that you normally get from an oven, without having to wait for an oven to preheat. I don't think anyone disputes that an oven will do a better job, but the air fryer does a better job than a microwave, which is what it's really competing against. They're also marketed as easy-clean, which again is a nod to their target audience.

How costs build up is really staggering. I'm getting into the business of importing 3d filament. In Iceland, it currently sells for like $35/kg minimum. The actual value of the plastic is like $1. The factory's total cost, all costs included, is like $1,50. If it's not name brand, e.g. they're not dumping money on marketing, they sell it for $3 for the cheapest stuff. Sea freight adds another dollar or two. Taxes here add 24%. But you're still at like $5/kg. The rest is all middlemen, warehousing, air freight for secondary legs from intermediary hubs, and all the markup and taxes on those things.

With me importing direct from the factory, sea freight only, I can get rid of most of those costs. Warehousing is the biggest unavoidable cost. If I want to maintain an average inventory of like 700kg, it adds something like $5/kg to the cost. Scanning in goods and dispatching user orders (not counting shipping) together adds like $2,50. And then add 24% tax (minus the taxes on the imported goods). There's still good margin, but it's amazing how quickly costs inflate.

Your "junk" is someone else's "must-have".
Your "must-have" is someone else's "junk".