HN user

Hizonner

3,669 karma
Posts0
Comments1,371
View on HN
No posts found.

If there were a well-known URL that retrieved all content added to a site after a given date, and a meaningful number of sites implemented it, I bet the biggest scrapers would use it.

Except, of course, that then sites would start lying and sending abridged, poisoned, or nonexistent data.

Nice things cannot be had, no matter how many levels you go down.

> The default system prompt is opinionated (fine) but it has shit opinions (not fine). It took me a while to figure out why my agent kept saying “Use ABSOLUTELY NO COMMENTS” when dispatching subagents.

Okay, so change it? Any LLM is opinionated, this system prompt enforces consistency across different models which seems reasonable.

OpenCode embeds scatters bad coding policy all over all of its inbuilt prompts, including tool instructions. The only way I've found to fix them is to edit the source code. And, yes, they're also huge and repetitive.

A framework's native prompts should include only information about how to interact with the framework itself.

Somehow nobody ever bothers to mention to the kids that it's not required.

How many schools still do it, though? Honestly you could tell me it was almost universal or very rare, and I'd have to believe you either way.

Of course, Canada was doing the freaking Lord's Prayer in schools until freaking 1988. I don't know about other countries, but wouldn't be surprised.

... and once you take more than the tiniest step beyond the defaults, the difficulty of configuring all that complex optionality exceeds the difficulty of coding your own, and coding the integrations you actually need. And you still get screwed when the vendor decides to make a change, or refuses to support some niche thing you need in favor of some niche thing a few more users need. Not to mention the security nightmare.

SaaS was mostly a bad deal before "AI".

Not a bad idea, except that *WEB PAGES SHOULD HAVE NEITHER ANY ACCESS TO NOR ANY CONTROL OVER THE HISTORY, PERIOD, AND SOMEBODY NEEDS TO BEAT THE MORONS RESPONSIBLE FOR THIS "WEB PLATFORM" BULLSHIT SENSELESS*.

It's good that a police department has chosen to do this with the misfeature, but the fact that there are non-abusive applications is not an excuse.

Nobody has tried to limit knowledge of chemistry or physics unless it was directly about doing something illegal, to the point of basically being a detailed recipe. Usually not even then. And when they have tried they've had basically zero success.

The ability for a handful of companies, simultaneously very powerful and easily susceptible to pressure from other powerful actors, to do the same sort of thing with the next generation of core learning and engineering tools, is freaking terrifying.

every country on Earth spies on the rest,

It's entirely possible an EU country did this; they're only vaguely guessing Belarus or whoever. In most countries, it's a big deal if the spies are caught spying on the domestic government.

quite a few private entities do as well.

It's a risky game, doing that. You don't get any of the professional courtesies, and you're not usually eligible for the prisoner exchanges.

How is it that any NSO employee is still able to travel outside Israel without getting arrested? Seems like they're involved in criminal conspiracies in like half the countries in the world.

Why would Craigslist stop Flock-related posts from going through? The only answer I can think of is something along the lines of a National Security Letter.

That's because you lack imagination. 99 percent chance they are blocking you because they don't want "divisive political rhetoric" on the platform. Allowing a surveillance state is "apolitical" as long as it doesn't involve rocking any boats or making any noise.

... and NSLs don't do that. It would really be nice if people actually understood what NSLs were before blaming everything on them. Trust me, they are bad enough without inventing stuff.

Or you have a child who is questioning their sexual orientation or identity and is targeted by an online bullying ring.

It is far more common for that child to be targeted by parents, and maybe by people they know in person, especially because of the lousy social environment their parents have pushed them into, and therefore to have limited offline support systems, and you are now trying to take away all they do have.

Um, the nginx binary would have to be in the hands of hundreds of thousands of server operators. And the set of server operators is rich in the kind of person who would attack it. Not to mention the huge number of leaks you'd get.

Maybe if it's some server-side software that you only use yourself...

I don't think such regulation should be restricted to AI. "Google-style B2C blanket bans" routinely ruin people's livelihoods in other ways, including, of course, when Google does it. There are way too many companies nowadays that are way too central to how way too many people live their lives.

If you want to be a piece of critical infrastructure, you need to deal with the implications of that. It's not OK for private entities to be able to "unperson" people in important parts of their lives for what amounts to convenience reasons. If them not being able to do that raises the price of the service, so be it.

Honestly even banks, which are already highly regulated and at least have more nominal competition, still have too much leeway to cut off customers based on error-prone statistical methods, without recourse or explanation.