anitil

These days I wouldn't be surprised if the models are smart enough to get the analysis correct.

However sometimes I think that AI use shows a revealed preference around things. In this case, the actual data just ... maybe never mattered to them at all? People wanted a shiny chart and they got a shiny chart.

This has been something I've experimented with my own baby langs, even banning precedence altogether and requiring explicit parens in any expression.

I find it annoying to have to remember precedence, even if it does lead to more concise code. I guess there's some elegance in something like `if (a == b = c + d << e & f)` but having to remember what this means overrides any benefit for me. And if you get it wrong you'll end up adding the parens anyway.

I haven't played with this version which seems like a reasonable compromise, but I always enjoy reading zig updates and how they really think through things that other languages assume away

Sorry I didn't mean it to be cynical, it just came to mind

I actually remember that in an old interview (that I can't find sorry) Grant Sanderson said he moved in to YouTube because he thought it was a more promising career than being mathematician, it's interesting to see his opinion change.

It's interesting as the appeal seems to be that he's narrowly targeted the training of his models at this specific industry, including side-stepping the rights issue by training on their own footage. It's a cool approach

I read your 'Leaving Tech' blog and it made me sad because it reflects a lot of what I'm feeling. Luckily I'm still employed, though it feels fairly tenuous at the moment

I thought it was clever!

This video is by Sholto David (who had made somewhat of a career in exposing academic fraud in bio sciences[0]), and is accompanied by a blog post [1]

The interesting issue in this case is that imaging was made by a third-party company, and it appears that this company appears to have photo-shopped the results. I imagine it must be devastating as a researcher to put time and money and effort in to this research to be let down in this way

[0] https://www.justice.gov/opa/pr/dana-farber-cancer-institute-...

[1] https://forbetterscience.com/2025/09/22/in-bad-company/

I was about to say that - it's really a mess

Even putting aside the ugliness of errno, this is pretty bad. I'm at the point with libc where I try to use as little of it as I can, though I'm not writing a lot of C these days

That's very fair. I enjoyed undergrad maths because it kind of blew my mind but I never considered pursuing graduate studies

That may be true, but mathematicians still need to eat, and for the they'll need jobs. 'Will there be more or less demand for mathematicians in the future?' and 'What will the work entail?' I think are valid concerns

I believe it's lying with extra steps

I think he took some time off to write up some academic contributions that he discovered as a result of working on the mechanism

I have only listened to the visa episode, I'll queue up those three too. Any other recommendations?

Related talk/podcast [0] (via @simonw [1])

He speaks very clearly about the topic, I had no idea he was in this field

(edited for clarity and links)

[0] https://www.youtube.com/watch?v=HzraLR2VkNM&t=2259s

[1] https://simonwillison.net/2026/Oct/7/ben-affleck/

Is there any chance you'd add an RSS feed to your blog? Unless I missed one?

Ah amazing thankyou! I'm really enjoying it and it has cleared up a lot of things I'd vaguely heard about but never fully understood

I actually didn't understand that paragraph. Is it saying that instead of averaging between two screws it instead averaged across 600? Edit: Or that the nut had 600 threads?

What really surprises me about these precision machines is how recent all of these inventions are, only a couple hundred years

Interesting, I'll give it a whirl tonight

I'm new to solvers and have only used them in anger exactly once, but during my research it seemed that the received wisdom is that hand-rolled solutions were typically faster, so this is a surprise to me. I suppose in the same sense that 'you can write assembly better than the compiler if you really want', which is probably also mostly false these days.

Do you have examples that are public or that you can talk about?

That's very cool. I don't quite understand your solution, are you saying that because you used the i:j relationship as a constant it meant you only needed to store positions rather than a whole board? I've never been interested in 8-queens but I'm keen to play with it now

This is one of those moments that makes me wonder how anything ever works at all

He (Han van Meegeren) was also the subject of a 'Cautionary Tales' episode [0] which I would recommend

[0] https://timharford.com/2021/03/the-art-forger-the-nazi-and-t... (sorry I couldn't find a transcript on that page, I'm sure there's one somewhere)

I tried archive.org but the captures I found were all paywalled as well

If you're interested to see some high resolution images of the tapestry there's one here [0] and Stefan Milo recently did a tour [1] which I really enjoyed - it captures a bit of the wonder that I imagine seeing it in person must evoke

[0] https://www.bayeuxmuseum.com/en/the-bayeux-tapestry/discover...

[1] https://www.youtube.com/watch?v=qIh5pjzonGg

I believe it's actually linked in the OP about half way down. What a crazy thing for an agent to do - to turn a minor disagreement into a maybe-crime.

Really nice explanation. When I saw the title I immediately remembered the tsoding video as well so it was nice to see you mentioning it