It's a shame, because there may well be a kernel of truth to some of it, but it's dipped so deep in LLMage that it taints the rest.

English is not my first language, and you nailed it. I used LLM to "polish" it. Probably too much. But I am open for questions if you like :)

A few days ago I got the nature scapes but with a, "This would make an awesome prompt huh?" as the tagline and a link to more AI shoe-horned in.

Bizarre that the US legislature is going to bat for the right of a corporation to sell illegal content in a foreign country.

The UK should call their bluff on this one, appeasement isn't going to work forever.

It's what corporate capture looks like. She's either peacocking for corporate donors or trying to court musks/MAGA's favor.

Also, there's a difference between 1 rep effectively shitposting on social media and the US government actually doing something.

Regardless I agree, it's not a good look

Luna is hardcore MAGA so she's definitely throwing out red meat.

It's more bizarre that Apple and Google are allowing X/Grok apps that generate non-consensual AI images in their App stores.

That's not bizarre. Nobody in their right mind should assume that Google or Apple are capable (let alone willing) of holding themselves accountable.

They are subject to the same bullying and will do exactly as they're told. They are not moral stalwarts.

If you're arguing down that route, LLMs can bulk-apply style attributes exactly where they're needed. Every element precisely described, no need for CSS and style-sheets at all.

And then you'd wind up with a needlessly noisy approach, and then you will reach for Tailwind to do basically the same thing but in a more terse manner. ;P

The only one I can think of, literally the only one, is grouped icons.

And even that's only because browsers ended up in a weird "windows but tabs but actually tabs are windows" state.

So yeah, I'd miss the UX of dragging tabs into their own separate windows.

But even that is something that still feels janky in most apps ( windows terminal somehow makes this feel bad, even VS code took a long time to make it feel okay ), and I wouldn't really miss it that much if there were no tabs at all and every tab was forced into a separate window at all times with it's own task bar entry.

It's not like grouped icons wasn't technically infeasible on win95. And honestly, whatever they are more useful is quite debatable. And personally, I don't even have a task panel anymore.

The real stuff not on Win95 that everyone would miss is scalable interfaces/high DPI (not necessary as in HiDPI, just above 640x480). And this one does require A LOT of resources and is still wobbly.

I'm not sure what you mean by "Technically feasible", but it wasn't supported by explorer.

You could have multiple windows, and you could have MDI windows, but you couldn't have shared task bar icons that expand on hover to let you choose which one to go to.

If you mean that someone could write a replacement shell that did that, then maybe, but at that point it's no longer really windows 95.

Ironically that might have passed, because this didn't break the version, this broke all versions when the global referenced changelog was published. It wasn't the new version itself that was broken.

But testing new version would have been downloading the not-yet-updated working changelog.

There are ways to deal with this of course, and I'm not defending the very vibey way that claude-code is itself developed.

Ah, that's an external file. That explains it.

I've been finding the reams of markdown that claude generates too difficult to read comfortably, so I switched to asking it to generate PDFs with typst instead.

Without even writing a "skill", it nails the execution, and then generated this booklet of examples with only a minimal prompt. ( It made up all the examples itself. )

Part of what makes Typst so nice to work with is the speed, the almost instant compilation means you can very quickly iterate.

Here's an example:

We have some tests in "GIVEN WHEN THEN" style, and others in other styles. Opus will try to match each style of testing by the project it is in by reading adjacent tests.

The one caveat with this, is that in messy code bases it will perpetuate bad things, unless you're specific about what you want. Then again, human developers will often do the same and are much harder to force to follow new conventions.

I'm gutted PlanetHalfLife isn't listed/shown here, I used to visit that site religiously around then.

Yeah HL introduced me to modding :) I used to visit it everyday waiting for the next chapter of They Hunger to get released.

I'd completely forgotten about that! They Hunger was a masterpiece. Not one I'd ever want to re-visit though, because nostalgia is best left not closely examined.

But a golden age, of txt file walkthroughs downloaded from gamefaqs with news mods and patches off PCGamer discs to explore because there's no way I was ever downloading those on a 56k modem.

I think the older games and mods are still pretty playable if you can get over with the slightly janky control. I recall TH has nasty jump traps which almost turned me away back then.

I wish I could relive the golden age.

Heck yeah. Remember when new games came out and you were actually genuinely surprised/impressed, over and over? For years in a row? Those were great times indeed. Sure, there were a lot of crappy games, but overall there were so many crazy awesome moments. Newer stuff is very iterative by comparison, IMO.

Yeah, back then both the games and communities are great, and there were diversities. But it may just be that I'm getting older.

Do you have a legacy code-base in mind?

I'd happily demonstrate this kind of workflow on my day job if not for company trade-secrets.

That's as legacy as it gets, 20+ year old code base with several "strata" of different technologies and approaches.

Claude Opus handily navigates around it, and produces working bug fixes with minimal guidance.

I'm not going to claim it's 20x or 50x yet, there's still navigation and babysitting involved, but it's definitely capable of working on complex problems, I just don't trust it to run it in YOLO mode.

The key thing is, you need domain knowledge. You need to know where to correct it, and which direction to point it in.

It's not magic, and it will have bad ideas. The key picking out the good ideas from the bad.

Hi eterm, this is very relevant to me as I'm building a self-hosted open-source tool for legacy code comprehension (AI/ML final project).

You mentioned "navigation and babysitting", could you share what that looks like in practice? Do you have to spend time reconstructing context or correcting Claude's misunderstandings? Do you still need to interrupt colleagues for some tacit knowledge, or has that changed?

I don't know. There's lots of options. At the extreme ends it would be interesting to see these agents work on something like boost, or metamath/set.mm, to choose deliberately obtuse candidates. Perhaps a web browser.