And to think, we could have had George RR Martins instead.
HN user
dyates
https://davidyat.es
[ my public key: https://keybase.io/dyates; my proof: https://keybase.io/dyates/sigs/s4XED2xJpVZPACFlXOcy50Xse8G3-dDxy7F5K3xe_qE ]
I had the opposite experience when Gemini 3.1 first came out. It didn't show up as a model option in my fully updated Gemini CLI, and I subsequently figured out I had to install this Cursor-lookalike thing called Antigravity to try it. I'd like to stick with my existing editors, thanks.
On a related note, the AUR package previously named antigravity has been renamed antigravity-ide[1] after some lively discussion, and the new thing lives at antigravity2-bin.[2]
I'd agree with sph about having one workspace per activity. I've never had a rigid workflow with lots of permanent named workspaces, but I have a workspace-naming script that lets me label my numbered workspaces after they've been set up.
Other things that help include a fuzzel-based open window searcher and, to be honest, restrained use of Niri's flagship scrolling feature. Most of my workspaces most of the time are the same size as my screen, with the scroll used very sparingly for usually temporary overflow.
I guess it also helps that I never used the i3 scratchpad so I don't miss it.
I switched to Niri at the end of last year after over a decade on i3.[1] Having horizontal scroll unbounded by my monitor size and workspace count unbound by the number of shortcut keys I have configured has been very freeing, and the graphical stuff is nice too.
My only remaining pain point is that its X compatibility layer, xwayland-satellite, does not yet support drag and drop between X and Wayland programs.[2]
[1] https://davidyat.es/2026/01/28/niri/
[2] https://github.com/Supreeeme/xwayland-satellite/issues/133
LLMs are not doing searches, they are doing statistical computation of likely results.
This was true of ChatGPT in 2022, but any modern platform that advertises a "deep research" feature provides its LLMs with tools to actually do a web search, pull the results it finds into context and cite them in the generated text.
In my last conversation with a Google support person, I was sent a clearly LLM-generated recommendation to switch to a competitor's product. Either they're not doing this, or the support person wasn't using Gemini.
ChatGPT's image generator has been able to do this since last year. That NBP still can't is baffling. They should at least train it to respond to requests for transparency with a solid colour pink background.
Interesting read! A lot of AoC challenges involve navigating 2D grids, which can map quite nicely onto the text adventure model of connected rooms with compass direction exits (a grid of straightforward little passages, all alike). This insight led me to attempt Day 6 from last year's Advent of Code in Inform 7[1], though I ultimately admitted defeat on the second half. I've always found Inform 7's Mathematics Textbook English syntax quite charming, though perhaps I would have a different perspective if I'd ever attempted to build anything substantial with it.
[1] https://davidyat.es/2024/12/23/aoc-2024-part2/#day-6-python-...
I'm trying to complete each challenge in a different language. I've written up the first five days on my blog[1]. Have completed a few more than just these, but I don't know that I'll get all of them done by Christmas.
Congratulations on the achievement. I recently hit my 10 year blogging milestone.[1]
I've also written about some similar thoughts to those in your linked "Content is King" post.[2] Long may the individualist personal website live.
I wrote a short blog post[1] about this pattern a few years ago. I still don't understand why so many apps use this workflow, especially ones that previously didn't. Just let me deal with the project creation stuff on the first write instead of requiring all this up-front commitment!
I used my Priv for five years, best phone I've ever had, barring some battery & heat issues. Showing people a normal-looking phone and then sliding out the keyboard was a fun party trick. If there were any modern phone like it I would buy it in a heartbeat.
I think the lockdown concept as implemented in 2020 was a great illustration of the problem of scale. It's trivially true that if you as an individual avoid all contact with others during a flu season, you won't catch the flu. You can probably extend that to your family, maybe a bit further, but at a certain scale it just becomes impossible (and undesirable) because of individual variations in occupation, lifestyle, risk and infinitely many other factors. And because someone has keep the wheels of society turning. There was no lockdown for supermarket cashiers.
My own government got very excited by the prospect of dictatorial central planning that lockdown mania created an enabling environment for and immediately set about writing all kinds of complicated rules about what food and clothing people were allowed to buy, as well as putting a prohibition on alcohol & cigarettes and imposing a curfew. Pandemic response immediately became a vehicle for imposing by fiat whatever pet policies ministers could vaguely link to it. Years later, they've all been rolled back, but the damage was done, and everyone still caught Covid.[1]
[1] https://www.groundup.org.za/article/nearly-everyone-in-south...
Later on the article, when he talks about not needing access to the receiving account to confirm payments, I kinda got it. Just the initial framing of "what if we used bank transfers for payments" threw me for a loop.
As a South African, it's weird seeing this presented as a strange/futuristic idea. I pay most of my bills through bank transfers and it's the standard way for two people with bank accounts to send money to each other. As long as you know someone's bank and account number, you can send them money.
All local online retailers, investment brokers, etc, support bank transfers, with some[1] even providing discounts for paying via bank transfer instead of credit card. I've even paid this way at a brick and mortar place that didn't have a card machine.
Jimmy Maher of the Digital Antiquarian wrote a couple of articles a few years back about the Eastgate School of hypertext this belongs to.
https://www.filfre.net/2017/02/the-eastgate-school-of-seriou...
Live chat windows that pop-up automatically and hassle you once you've spent X seconds looking at a website are annoying, much like email subscription modals. But when I've been actually looking for support and made the choice to click on the chat button, I've generally had a great experience, so long as it's connected me to a human rather than a menu of chatbot options. It's asynchronous enough that I can keep doing something else while I wait/chat, saves me from having to spell things out or repeat alphanumeric codes over the phone, and is way quicker than email.
I used a cellphone holster for years as a teen because one came with my first BlackBerry. I think I only stopped once I got my first touchscreen phone in a modern form factor that didn't fit my holster. Had some tense moments at a couple of metal detectors when reaching for the holster to put my phone in the tray.
My personal blog, with some other stuff lying around. I've been neglecting it over the past year or so. It's a Hugo-powered static site, though I started off using Ghost.
My design philosophy, such as it is:
Have a unique look without being too esoteric, and minimise cruft on content pages (https://davidyat.es/2019/05/05/site-redesign/)
Provide a pleasant reading experience
Use JavaScript sparingly and only for enhancements (example: https://davidyat.es/2020/12/31/footnote-previews/)
Gruescript is a similar system, released a couple of months ago (though it doesn't have official support for graphics):
I've seen a fair bit of this with Firebase apps, where devs don't write enough rules, or have collections that mix non-sensitive and sensitive fields. It's tricky, because the whole query-the-database-from-JavaScript model causes your app to fail open. I wrote a tool that acts as a generic Firebase datastore client to help find these sorts of flaws.[1]
To quote a since-deleted twitter thread written in response to this article at the time:
Golding is not trying to play this Rosseau-v-Hobbes game of bringing humanity back to the state of nature and see what they look like. His children are not living in a natural state: they are self conscious heirs to an entire civilization, and they realize quite early on that they live on the razor's edge of life and death. So they create rules and orders to keep their society running. A little democracy that honors natural goodness and leadership.
A little democracy that does not last.
And that is what people don't seem to get about this book. It isn't a parable about what happens when you taken men out of society--it is a parable about how men act in society.
The whole thread is worth reading, though it's not well preserved:
Markdown is pretty good for advanced blogging if you use it as a superset of HTML, as originally intended.[1] That way you get simple syntax for the stuff you're likely to use most often and can still implement more complex formatting when you need to.
[1] https://daringfireball.net/projects/markdown/syntax#html
Twitter is awful for longform, but it's where the eyeballs are.
It's called the Mid-Atlantic or Transatlantic accent and was designed as a way to blend the American accent with British Received Pronunciation. It was deliberately cultivated by the upper class and actors in the early 20th century; there was a book called Speak With Distinction that taught people how to speak like this.
It started dying after WWII and was basically gone from the media by the late 60s. Maybe because post-WWII America was less dependent on ties with Britain for its national identity, or maybe because people decided it sounded silly and fake. No-one spoke with the accent naturally, so it really only took a fashion change for it to disappear.
https://en.wikipedia.org/wiki/Mid-Atlantic_accent
https://www.atlasobscura.com/articles/how-a-fake-british-acc...
The best solution I've seen for making sidenotes responsive is to move them into the body text. Examples:
http://craftinginterpreters.com/a-map-of-the-territory.html
https://www.martigeon.com/textbook-themed-webdesign/
I'm not sure how well this would work for shorter notes though.
Mine is https://davidyat.es
I write about a lot of different things, often technical, sometimes not. Common subjects include technical tutorials and rants, reviews of books and games, and thoughts about different aspects of internet culture.
These are my three most popular posts:
How to set up GPU passthrough on Linux: https://davidyat.es/2016/09/08/gpu-passthrough/
Why RSS is better than its modern replacements: https://davidyat.es/2017/05/18/rss-nothing-better/
How to create a dual-headed personal wiki with Vimwiki and Gollum: https://davidyat.es/2017/09/01/vimwiki-plus-gollum/
The Digital Antiquarian has an article that goes quite in-depth on the history and context of this game:
https://www.filfre.net/2015/07/a-new-force-in-games-part-2-a...
This is so true. The other day I listened to a project manager rant about how Slack is "like a coding tool" because "you need to put stars for bold". She was really quite upset about it.
Another time a friend of mine, who I wouldn't call non-technical, but isn't a programmer, was collaborating with me on a Github wiki for a project we were working on. I was flabbergasted at the way he would copy-paste and edit the source of pages I'd written just to get bold text instead of typing the asterisks himself. He was just really used to WYSIWYG formatting buttons.
This HN comment[1] provides quite an interesting theory about why Markdown in particular has such a different reception by programmers and non-programmers.
Using LaTeX is like writing a program to automate something you do a lot:[1] lots of effort up front for payback down the line. Once you work out how to use it and iron out the kinks in your templates, it becomes much easier, permanently. Whereas with Word, while you'll get started much quicker, every sufficiently complex document will break in new and strange ways that require extreme dexterity with a mouse and a good deal of random fiddling to fix.
[1] https://xkcd.com/1319/ (ignore the bottom graph)