Welcome to BlueSky! Honored to be your first comment. On the licensing point, many publishers have long-standing text and data mining addenda to their licenses and I've yet to see a principled reason why feeding PDFs to an LLM isn't covered under that, but agree on the "mostly for hobbyists" thing.

Yes, being a crank about one specific thing can go either way: bsky.app/profile/dept...

I get the regrets about peer review as it's often practiced, but if you show me a piece of work and tell me you've shown it to some people, they pointed out issues, which you've addressed, I'll take it more seriously than if you just show me a piece of work.

Hmm... One could argue that this is what preprint servers have done.

Reflexive loathing is definitely common on social media, more so than in person.

There's insurance and there's catastrophic risk insurance.

If you're just talking about liability payouts to harmed consumers, maybe. But if they're required to have insurance, the underwriters (who actually understand risk management) will not write policies unless the insured are actually minimizing risk. See fire insurance in CA, for (a weird) example.

Silly me talking about browsers. I should have said apps, and I guess that means app stores will have to fill the role of browsers. Who's working on this?

What a valuable contribution! I'm sure they've never thought of that and now that you've given them the idea, they'll see how brilliant it is. You must know there's not a whole literature on the effect of incentives on review, because you would have expected them to be familiar with it if there was.