I'm guessing their issue isn't about the vowel, it's about the number mismatch between the singular article "an" and the plural noun phrase "frequently asked questions".
An half-air/half-oil filled watch that looks like a smartwatch, but is fully mechanical. No bling, somewhat understated, but still quite visually interesting with a modernist design.
And all kinds of interesting technical quirks like using a magnetic coupling to transfer motion from the air filled half to the oil filled half, tiny bellows that open and close to allow the oil room to expand, the little temp gauge etc.
It's very expensive, but not cartoonishly expensive. And the expense isn't tied to speculation or hundred year old pedigree like some other watches, instead it feels like you're paying for people who really enjoyed skirting along the edges of their craft in a time-intensive way the same way a hacker does
(I don't think the pick has stayed the same over time though: early days would have been some Casio calculator watch, then the Apple Watch/Pixel Watch, and now this)
One of the top stories on HN yesterday was about a company that paid 4-5 average people's wages per person for a team that sat on their butts 8 hours a day and wrote meeting scheduling software for a decade. This was done so they could then sell, not even the software, but... the right to their institutional knowledge for an additional few thousand years worth of average wages.
And of course they're permanently deleting the fruits of that decade's worth of work with 1 week's notice.
And this is the 2nd time the team's leaders have run this play, with the same buyer paying each time: seemingly they can just leave again and keep doing this ad nauseum. (Clockwise)
If you put the value we assign to software engineering in terms of other things it really doesn't make sense either. At least what these people did is something mechanically interesting, unique, and enduring vs the average CRUD app.
I suspect (but don't know) that this had to be edited somewhat heavily or generated in isolated chunks: I've generated a lot of fiction with Claude and it has a chronic issue of overusing any literary device one might associate with good writing once it appears in the context window
I think if you left it to its own devices, some of the narrative exposition stuff that humanized it would go off the rails
Yeah, there's a lot more work and personal touch that went into this (and the previous piece) than just "write prompt -> copy/paste into substack".
It's really interesting to hear about others that have been exploring generating fiction with Claude. I clearly need some more work based on some of the comments, but it has been really interesting discovering and coming up with different techniques both LLM-assisted and manual to end up with something I felt confident enough about to put out.
I'd be curious to hear more about your experience!
I run a product that generates interactive fiction (for search engine reasons I don't mention it in my comments, but there's a link to an April Fool's landing page in my post history where you can try it)
Because it's productized I need to "one-shot" the output, so I focus a lot on post-training models these days, but I've also used tricks like running wordfreq to find recently overused words and feed the list back to the model as words that cannot be used in the next generation.
Models couldn't always follow instructions like that (pink elephant problem), but recently they're getting better at it.
Just this week famous SF-local "Purple Ferrari With A Duck Man" (I don't know his real name) went through what seems to be a psychotic break and ended up in an armed standoff with police
There're some early comments saying he was having apocalyptic delusions reinforced by Gemini: this really seems to be growing as a class of issue.
What's strange to me is that, while subtle delusions are hard to deal with, delusions where the model is saying "we are at war now with those who destroyed the Earth" seem like they should be very easy to catch with a classifier, and so do the series of prompt that go this far (you can boil the frog with LLMs, but getting it to encourage violence typically requires some pretty sharp prodding.)
we're one more rl run from Codex not trying to satisfy the type checker by replacing an index accessor with a BFS of all the keys in the API response and matching the correct property via regex.
Honestly I got hung up on understanding why for Typescript the declaration of `keys` shouldn't be:
let keys = (keyof typeof users)[]
Like I get it's a contrived example and maybe I'm missing some nuance to it, but if we're obsessed with type-safety why are we treating a array of strings as an array of keys (which are more like atoms)?
I thought the answer might be we're looking for duck typing-ish behavior, but then the proposed take signature doesn't work either?
reply