Beej Jorgensen

We’re harmmaxxing!

Here’s mine: the world would be a much, much better place if LLMs did not exist. Also, it would be stupid–today–to not use LLMs to find flaws in the Linux kernel. The only other option is to get your machines compromised.

Is it therefore smarter than you, Sam?

You promise?

This is true. But you have to balance out how much you care about following specific people. My friend was dying of cancer, and she posted to FaceBook and I wanted to keep up. It was her preferred way of maintaining contact. Would you have told her to switch to something else?

She passed away a couple months ago. So maybe you’re right that now would be a good time to cancel the account.

I don’t use the platform but still have a presence so I can be contacted there. But half the messages I receive are unsolicited porn spam. Kids are getting that shit pushed to them!

I was about to move all my photos to Proton, and then… I guess it’s a vps and Immich after all that. Sheesh.

Got mine in '92, I think. Still on my shelf.

Make still bests the author’s examples in certain use cases. I use it all the time in those cases and it’s fast and simple. Use the right tool for the job.

I don’t disagree. There’s a lot of pretty poor material out there. 😅

The problem is that the bot doesn’t know what people tend to be confused by, so I don’t think it’s a matter of the prompt. The verbosity itself isn’t even really the issue; the clarity is the issue.

I, too, use LLMs all the time to explore spaces. I just strive to help people explore particular spaces more efficiently than they can with LLMs.

And another thing. I had to write some C++ material recently. Conversations with Claude went something like this:

“I’m a senior dev who hasn’t used C++ in 20 years. What’s new?”

[pages of output]

“Tell me more about the rule of zero”

[pages of output]

“[More questions]”

[pages of output]

“[More questions]”

[pages of output]

Then, once I understand it (and have written some sample code to mess with) I write a concise, direct, effective paragraph or two. You could read all AI output if you want. But I think I did it better.

You know when you finally see the answer, and you think, “Why didn’t anyone just say that in the first place?” When I write, I look for that thing to say. LLMs are pretty crap at it.

Amen to that. My goal has always been to produce the most optimal educational content. I don’t sell ads, track anyone, or charge for access. The only purpose is to write the best stuff. And currently it’s really easy to write better than an LLM. I don’t know what that looks like in 10 years, but the author’s right–pick up those shovels and produce quality human works.

And will LLMs vacuum it up immediately? Absolutely. Can’t stop it. I’ve put half a million words out there and they’ve eaten every one. But that doesn’t mean you can’t still best it, 'cause you definitely can.

It probably is honestly questionable whether walking up or helicoptering up is safer. https://en.wikipedia.org/wiki/Didier_Delsalle

Reproduction of the material we agree is against the law.

A model is very obviously a derivative work, as it has 0% original content input. Under fair use law, it also very obviously fails all four tests. Computing the letter frequency has 0% original content input, as well, but is definitely not a derivative work. Do you disagree?

but we’d first have to have models trained without violating the rights of all the authors of the works in the training data. Like if you trained a model only on GPL software, then the model and its output should also be GPL.

I’m not so convinced on that one. Copying code without permission, yes, that’s illegal. Distributing copied code without permission, also illegal. But processing code (copied with permission) into weights, that doesn’t feel to me like a violation. If I have a legit copy of something, I should be able to distribute computed statistics about that copy.

EFF points out some hazards: https://www.eff.org/deeplinks/2026/10/rewriting-copyright-law-generative-ai-would-hurt-everyone

Human-generated code absolutely is, correct. But AI-generated code is still untested. AI-generated art does not have copyright protection currently. My money is that AI-generated code will have copyright protection through some mental and legal gymnastics. Because money.

Well, depends on the person. And I think the era–I’m getting the impression what it takes to be a dev in 2020 isn’t quite the same as what it took in the 90s. But in any case, I said toward the senior dev level in that timeframe.

Slopmaxxing ftw.

I want to climb Everest, but it’s so much work. I’ll add my name to the ranks of the greats by taking a helicopter.