withinboredom

Might be worth extracting the system prompt and then patching it. TBH, that's what I was expecting when I saw the gist.

When I talk to my niece in the US and she says they have shooting drills instead of fire drills ... I think the US might be doing the wrong work.

I suspect it happens when the model's adaptive thinking was too conservative and it could have thought more, but didn't.

Is there not a setting to change the system prompt itself? I vaguely remember seeing it in the docs.

If this is the “straw that broke the camel’s back” as they say… it seems it is more about the field than anything else. It’s a strange hill to die on… there are much larger changes happening on a daily basis… or is this like a bike shedding effect, where it’s such a small trivial and pointless change that it is worth fighting over? Something everyone can understand.

I dunno. The only reason I’m even on the mailing list was to report a bug several years ago…

Because someone added a new user field? Does that need governance?

The mailing list doesn't seem to make it out into such a controversial issue. It's an optional field that doesn't require a real birthday.

Ah. That’s like a 15-line rite-of-passage plugin you write once and never have to worry about it again. Filter content going into the database and use relative uri for the same site. Configure everything else via environment variables.

I moved away from Wordpress altogether earlier this year because I got tired of babysitting MySQL.

This is why I prefer the AGPL over the GPL. But isn't this the entire point of open source? So long as it is attributed/following the license, who cares if they're selling it or not?

I would say it was a collapse of ethics, not morality. Most people have morals (their own belief system on what is fair), but their morals may not be ethical (rule-based morals to achieve fairness). I personally attribute it to cars and the internet.

The internet removed consequences. You can say the most vile thing imaginable to another human being and… nothing happens. No social cost, no awkward eye contact at the grocery store, no reputation hit in your actual community. Just a dopamine hit and a notification count.

Cars did something sneakier. We spend hours every week sealed in a metal box, alone or with the same people. No random encounters, no friction with people who think differently. Just you, your podcast, and whatever is important in your tiny echo chamber.

Put those two together and you get people with deeply held morals and zero framework for applying them to anyone outside their bubble. Ethics requires seeing strangers as real. We've engineered that out of daily life.

> a dev/test/prod workflow involved copying filesystem content, database content, and changing URLs that got saved in the database.

This just sounds like deploying web software. You always have static assets that need to be deployed, the code/binary itself, and database migrations.

Email isn't guaranteed to have a faster round-trip. https://groups.io/email-provider-status -- sometimes goes into hours of latency.

My blog does a proof-of-work before submission (withinboredom.info) in your browser. It'll use a fair bit of cpu power, but should only take a few seconds to complete. For an attacker... that's quite slow and self-limiting.

A $150 hold would clearly be noticed by the victim, so the attacker wouldn't even try it.

> The key difference is with cloudflare your site is globally distributed by default, and you get to buy into the whole ecosystem, if you want.

The real question nobody asks: do you even really need global distribution?

The process itself (including other threads) can call setenv whenever it wants.

Please read the https://news.ycombinator.com/newsguidelines.html

I literally can’t even continue this thread.

This is about going to the moon. Space-x is over budget and extremely late. It has nothing to do with the management there, only that it is better to come up with a solution without them.

environment variables can change while the process is running and are not memory safe (though I suspect node tries to wrap it with a lock). Meaning if you check a variable at point A, enter a branch and check it again at point B ... it's not guaranteed that they will be the same value. This can cause you to enter "impossible conditions".

I'd prefer if we just wrote off space-x and pretend they don't exist.

timoutμs is even better. People will learn how to type great symbols.

Unless you and I are looking at different web pages… it only goes back to February, not December or January.

It’s not a mystery… I can tell you what I do most days, and probably 80% of it is communication. An AI could do that. That communication is to learn what is going on up, down, and across the org. I mostly want to make sure we aren’t doing redundant work — though sometimes that is useful, and making sure timelines aren’t slipping. Oh, and dealing with conflicts.

The other 20% is writing: policies, SOPs, audits, grants, performance reviews, etc.

I could probably automate over half my job in n8n in a weekend… hmm… actually might try that.

Probably patent licensing shenanigans kept it holed up for awhile.

The funny thing is that AI can probably replace the exec’s job before it can replace a devs job.

Well, I don't see 4.5 on there ... so I'm not sure what you're trying to say.

And today is a 53% pass rate vs. a baseline 56% pass rate. That's a huge difference. If we recall what Anthropic originally promised a "max 5" user https://github.com/anthropics/claude-code/issues/16157#issue... -- which they've since removed from their site...

50-200 prompts. That's an extra 1-6 "wrong solutions" per 5 hours ... and you have to get a lot of wrong answers to arrive at a wrong solution.

Hold my beer.

The chart doesn't cover the 4.6 release which was in the end of December/early January time frame. So, it's hard to tell from existing data.

Opus 4.5 ~= Opus 4.6 high. Opus 4.5 was nerfed just before or after the release of 4.6.

Not just complicated, but even simple ones if the current software is too “new” of a pattern they’ve never seen before or trained on.