WWII sent a very clear message. You can annex Austria. You can invade Czechoslovakia. You can take over Lithunia. But you don't fuck with Poland

Well, I mean, you can fuck with Poland a little bit. You just can't take over, like, too much of Poland.

Reminds me a little of the old Jonathan Shapiro research OSes (Coyotos, EROS, CapROS), though toned down a little bit. The EROS family was about eliminating the filesystem entirely at the OS level since you can simulate files with capabilities anyway. Serenum seems to be toning that down a little and effectively having file- or directory-level capabilities, which I think is sensible if you're going to have a capability-based OS, since they end up being a bit more user-visible as an OS.

He's got the same problem every research OS has: zero software. He's probably smart to ditch the idea of hardware entirely and just fix on one hardware platform.

I wish him luck selling his computer systems, but I doubt he's going to do very well. What would a customer do with one of these? Edit files? And then...edit them again? I guess you can show off how inconvenient it is to edit things due to its security.

I just mean it's a bit optimistic to try and fund this by selling it. I understand he doesn't have a research grant, but it's clearly just a research OS.

Omegle is a bit of a unique case due to their persistent non-action. Most places, if people start grooming children or broadcasting child porn, they'll start banning offenders at the very lest. Omegle, nah.

At one point, they put a warning splash screen "Careful: there are pedophiles that use this" or something like that, but they took the warning down after a while. And eventually they did officially say that you can't use the site if you're a minor, but of course it was just enforced through the honour system.

Those are literally the only two actions they ever took to address criminal content and behaviour.

Easier compared to what? Easier compared to sysvinit, of course. Easier compared to all the other alternatives? Six of one, half a dozen of the other, on balance, I would say.

But SystemD has inertia behind it now. If you run into problems, there are probably 1e10 web pages out there that will help you fix it. That's why Debian solidified on SystemD: not because it's any better than any of the others, but because it's the same as everybody else.

That's less fun. I believe you've either got to put everything on one SLAAC network (no static IPs), or you've got to use DHCPv6 (with a smaller network size) instead of SLAAC.

This isn't just the same guy that's released the other videos: it's the original video. I don't know why this community has been flooded with year-old posts.

I don't think "greed" is quite the right word. "Greed" would be the right word if they were trying to make themselves more profitable. But they're not: they're trying to make themselves profitable at all. That's not about greed, but about surviving. You can't survive unless you stop hemorrhaging money at some point.

Maybe the question is "Why do investors invest so many hundred of billions of dollars into companies that cannot be profitable without becoming super-shitty? And why do users join them knowing that they're going to become super-shitty one day?"

Just set your tab width to 1. No more argument now.

If it helps you to visualize, one somewhat common/popular form of personal knowledgment management is a wiki. Like Wikipedia, except it's personal (or for a small team). You can keep track of references and also make notes about things, but it's also about connecting ideas together. Just like on Wikipedia, you can have a page about, let's say LLMs, which includes all the software and approaches you've tried, results, sample snippets, references to repos, but as you're writing about what you've tried and what worked, you might also have links to other wiki pages, like programming languages, build tools, test tools, etc. As you document more and build more knowledge, your articles all get meshed together in one well-organized network. Ideally it should be easy to navigate if you come back to a technology later and need to get back up to speed.

I have a similar kind of idea. I think if it had been a free/open source/community project that made the headlines I would have been all like "this is so awesome".

I guess what I don't like is the economic system that makes that impractical. In order to build one of those giant GPTs, you need tonnes of hardware (capital), so the community projects are always going to be playing catchup, and I think quite serious catchup in this arena. So the economic system requires that instead of our posts going to a "collective hive mind" that aid human knowledge, they go to some walled garden owned by OpenAI, which filters and controls it for us, and gives us little bits of access to our own data, as long as we used it only in approved ways (i.e., ways that benefit them).