npm is a problem. It's this massive, unvetted self-publishing repository without any easy way to verify the origin of packages, and the packages largely get installed directly to production on billions of sites every day without any vetting or review.

It's crazy, like something out of the 90s.

Yes, supply attacks like those carried out against npm are pretty common in general, at the state actor level. There've been a couple fun ones in recent years. But the openness and lack of basic precautions surrounding npm in conjunctions with common development practice just makes it a recipe for disaster.

"lossy compression"

Yes, just like human memory.

If I read a bunch of books from a series and extrapolate based on them to form something similar, it's not plagiarism.

If I read your book, then write a book using a similar voice, style, and plot, and do it in a different language - it's not plagiarism if I offer citation. Likewise, if I do so with a verbatim copy in another language. It's an independent effort.

Ultimately, it boils down to what you can get away with. Considering how trivial it is now to re-implement things, I'd say the chance of license enforcement is close to zero for anything open source except in extremely rare situations where there's a lot of money involved.

The UK has reliable rainfall data from the 1860's (165 years), and less reliable records going back further. But compared with 1,200 years, you have the right magnitude.

If you can substantially distinguish how tokenized abstraction is any different than natural learning, I'd buy it. But as it's not, I don't think that's a meaningful argument.

You realize that's extremely easy to do with AI, right?

If I can run software, I can run run function/system tracing and introspection on it. Simply that. The difference between this and looking at the code is almost negligible, except in terms of scope of difficulty.

And with AI, I can automate the entire process.

I'm sure there are techniques which will be readily implemented in at least one of a half dozen different agent platforms - assuming there's any merit in it. It's too stupidly easy to build things now to really keep anything "unique" private. People will figure it out and do as they will to get things they want to use.

I personally have an agent framework that's a combination of capabilities of different agent platforms that does things the way I want to. I haven't shared it, though judging by the quality and utility of some of these other platforms, I just might once I iron out a couple little bugs...

Yes, it was miserable nonsense.

Antrhopic seems to be spiraling of late, doing a lot of things which are shooting themselves in the foot.

- This "accidental" code release (I'm not convinced it was an accident and not a fancy PR stunt)
- The complete nerfing to useless of Claude Max plans (less usage, heavily throttled to the point where even getting close to quota has been impossible, and waiting 30m+ for a simple prompt response often takes longer than doing it myelf)
- Consistent API outages for the past several weeks during US business hours.

In this specific case, the only reasonable response to the leak was to embrace it: "our client is now open source, I guess". Put it in a repo and control it. They'd have gotten a lot of community involvement, as it's the forefront client, and would've effectively nix'd many moving to opencode, codex, or hermes agent. Sure, there are a lot of features in there which make them look not-so-benevolent, but it is what it is.

Now they've got a PR nightmare to contend with.

Maybe the legal experts could sit down and work out how to modify licenses (including the GPL/LGPL) to be for non-commercial use only? As long as an entity wasn't making money using FOSS, it could use it just like now. Individuals and non-commercial projects wouldn't be affected. But if you're a business making money using FOSS? Not without paying for it you're not. Yes, this would go against the free-software principles. But principles don't pay the bills every month, and none of these changes would prevent anyone from staying with the existing licenses if they wanted to.

The first thing I think of as a problem would be a company setting up a separate entity that wouldn't make money, just make services available to the company using FOSS to get around the fees. The trick to preventing this would be to phrase the terms so that that entity truly had to pay it's own bills without having the company using it's services pay anything either directly or indirectly. Not even by doing things like providing hosting "free". I'd have to sit down with a bunch of rules lawyers and game out all the ways to funnel money into that entity and how to block them, but what's life without a little challenge?

Anecdotally, the only crashes I've ever had on MacOS were due to HDMI and cheap USB dongles. Find a combination that works and you're not going to have problems. I've been a consistent MacOS user since 2017.

Haven't had a single issue in years, on Apple Silicon.

These things include:

- booting
- rebooting
- using basic high quality hardware (asus/mb/msi boards w/ corsair/crucial memory, nvidia GPUs, seasonic PSUs)
- installing drivers

I've seen crashes on W10/W11 on each of these, sometimes (often) requiring "repair" that fails, and a reinstall (of the OS). Multiple machines.

I just won't do it anymore.

On linux, sleep/hibernation is more like a Polish revolver with one bullet loaded. When it works, it works. When it doesn't, it really doesn't.

Hey, believe it or not, that is actually the OS crashing.

The crash might occur in the driver, but it's still the OS crashing.

These driver crashes on Windows typically lead to having to reinstall/"repair" Windows. It takes a lot of time, and is a frequent occurrence. It's more common than it used to be in the W7 days by far.

I've been doing this for 30 years as well, and you're full of crap. Even with new, reputable (high end) hardware, it's a common problem.

This is a lot of cope. Sorry - there's nothing historically or linguistically accurate about that paper. It uses liberal misinterpretation of the word 'regulated' to infer government control, and grossly over-extends how militias have been regulated and mustered for the 300 odd years prior to the Constitution, and for 150 odd years after. It's doublespeak, a reinterpretation and recast of original intent and meaning.

My guy... have you been on youtube lately?

Ignoring for a moment that militias were actively prosecuted and pushed underground during the 80s/90s/00s, "guntube" quite clearly shows that there are organized and well equipped (how we say 'regulated' in today's parlance) militias out there still. They're just not registered 501c3 organizations. When the founders wrote the US Constitution, "militia" was every able bodied male who could muster arms. This is well established historically from the English tradition.

It isn't that we think gun-lovers are going to go ape-shit and shoot everyone around them, it is that a proportion of gun-lovers will do this. Which ones? Why you just have to ask them.

Why don't you do that, then? And look at the shooting death and mass shooting statistics and demographics, while you're at it. It isn't the people you're concerned that it will be, at all.