Oh and prime video has ads now.

If your concern is that some hacker could put an illicit binary in ~/.local/bin, then your security problems are much deeper than your PATH order.

Related: I'm (pleasantly) surprised to see the rash of BBS-related content on HN recently, as I've been working on my own (rather niche) BBS retrospective article for the past month or so.

It doesn't feel like there is a great modern alternative to this.

Never say 'no' to something unless there's an obvious safety issue.

Especially don't say 'no' just because you don't understand the request: that's the time to show them that you're curious about what they're thinking and give them practice explaining their reasoning in a safe and comfortable situation. Kids need freedom to explore their world.

At first glance, this looks like another one of those "LLM riddles" that humans _think_ should be easy for an LLM to answer but is actually quite difficult because of how they work in the first place. The answers to such riddles ("should I walk to the carwash" or "how many R's in strawberry") reveal the weaknesses in our expectations of LLMs in general, not weaknesses or characteristics of any particular model.

I'm not sure I see this as much different than asking a bare model its own name: without a system prompt or post-training, it doesn't know, it's just a bag of weights and will hallucinate an answer to that the same way it will anything else.

I'm sure you already realize this but to be very explicit, You're not getting an actual random word out of an LLM this way. You're seeing the bias in each model's training set around how often they've seen "random word" followed by "lantern" or "zephyr" during training.

* All the posts are walls of text. In many of the posts, all of the paragraphs are just about exactly the same length.

* None of the posts contain personal stories, experiences, or anecdotes.

* All of the posts are hyper-focused around advocacy of "old tech" and decentralization. Which is great, but real blogs with this many posts have at least _some_ variation in subject matter and quality. This one's surprisingly uniform.

* The blog is relatively new, posts began about two months ago and there are 17 posts, that's a little over 2 posts a week. Sure, there are bloggers who post that often, sometimes daily, but they are usually much shorter posts, tend to be shorter, or do it for their job.

* The picture of the author on the About page is very obviously AI generated.

* The author has a GitHub account that had virtually no activity until March of this year, and basically all of the commits were co-authored by Claude.

To be clear, I'm sure a real human is behind the site/posts (as opposed to a fully autonomous agent), but I'd bet dollars to donuts that each article started out its life as a very short prompt.

Containers protect against "I don't trust this curlpipe to not crap all over my dotfiles," rather than, "there might be a sandbox escape attack in this random file I downloaded."

If a VM is not sufficient for your threat model, I'm curious what is?

What harnesses do you recommend?