This suboptimal approach is basically what every Android phone on the market uses, right? They ship a 'vendor' kernel with all the drivers built only for a specific version, and the phone stays on that same kernel version until EoL.

That's not what that warning means. "Does not have a stable CLI interface" means that the _input_ format may change (the stuff you put into argv[]), not the _output_.

First they came for Tiktok
And I did not speak out
Because I didn't like Tiktok nor China

There's a lot of "we've always done like that" in those arguments.

A good chunk of those domains (friendship, dating, academics, politics, family dynamics, identity, for instance for sure) can exist over the internet as well. To quote a wise man, "we say AFK, not IRL, because the internet is real life, too". And it's going to be even more so in their adulthood.

We have no idea what the world is going to look like in 30 years. What if the skills and mindset they learn by looking at their smartphone all day end up being more important than the in-person social skills this guy wants to replace them with?

The nerds of the 80's that parent encouraged to turn away from computers have now become highly paid software engineers. What if you're crippling your kids' lives and careers because you want them to grow up the good old, traditional way?

If I hire a human to write a newspaper article for me, and they hand in a copy of an existing article on The New York Times, they're liable for copyright infringement, together with the organization that hired them and published the article. Why should this be different if they use a LLM?

Humans need to be taught not to plagiarize. Why can't OpenAI spend 1% of their training time on detecting and preventing whole chunks of copied text? Or on running Turnitin on all the text they generate?

Why is this news now? 1.10 was released on December 27. And 1.10.1 came on February 14, so today it's already an outdated version.

Looks like the solution is (1) don't reinvent the wheel use standard public key / private key crypto, for instance the way SSH does: install a new public key on the car to log on with the keyfob that has the matching private key (2) require users to press a button to unlock the car, like FIDO keys do.

Hardly an unsolved problem, if this was done competently.

Why just them? Almost every firm stores personal data, and all breaches seem equally dangerous.

In real life, probably it's Visa and Mastercard that forbid the pirate from accepting credit card payments. They are a de-facto duopoly, and they're not afraid to weaponize it by refusing services the US of A don't approve, like they did with Pornhub and Wikileaks.

We should go back to the old habit of chiseling the big philosophical truths on stone tablets then. Kids will never learn properly with parchment, it's too fast to write with a quill.

Uh, the script is legit, the worm does contain those markers. Or at least so says Akamai, the owner of that Github organization. You know, the CDN and cybersecurity company with $3B yearly revenue.

Download and run https://raw.githubusercontent.com/akamai/akamai-security-research/main/malware/noabot/noabot_detect.sh .

This is like saying that banning the letter E in emails is a good idea, because it would make life harder for Nigerian-prince scammers.

Different paths make life hard for everyone, not just for the malware. The same differences cause trouble to all developers.

Then why do you need separate directories? Set things up so that (1) programs that are meant to be executed by root only have permissions that enforce this, such as 500, and (2) shells autocomplete only the names of programs that are executable by the current user.

Thanks! Even if it doesn't leak passwords, a strategy like this one still looks horribly problematic, because it would reveal to Google that the same person has username john.smith423 on Linkedin and blazeit420 on mylittlepony.com.