GavinAnderegg

Author here. I'm not complaining or trying to talk smack. I'm just pointing out something that seems to be coming: LLM price increases or the products being degraded to make more revenue. If I knew how to solve that, I'd be insanely wealthy.

Author here. The reason I wrote that local hardware is "sipping power most of the time" is because most of the time it's not doing LLM-related work. If you're just using your local machine (or eventually maybe even your phone) to do local LLM tasks, you're not doing that all day.

I agree that data centres will be set up to be more efficient, but we're also going to need fewer of them if local LLMs take off. If that's true, overbuilding data centres is more revenue pressure for AI companies.

Thanks for this! Seems like a bold stance… but the Online Safety Act also seems like a poor piece of legislation.

Is this something you're seeing personally? If so, how do you know it's because of the Online Safety Act? This is a personal blog and it doesn't seem to have any adult content that I can find. The homepage of the site isn't blocked when I check it here: https://www.blocked.org.uk/check

The book starts with generating fairly standard mazes, but transitions to making more interesting ones in later chapters. There are 12 algorithms explained in the book (listed in the link above), and the author does care about making pleasant mazes.

This is a great list! A while back I also enjoyed reading “Mazes for Programers” and playing around with different maze generation algorithms from that book over a holiday break. The book isn’t super deep, but it has a fun set of projects and further ideas/reading as well. https://pragprog.com/titles/jbmaze/mazes-for-programmers/

> AI is getting better/faster/cheaper at incredible rates, but regardless of when, unless you believe in magic, it's only a matter of time until we reach the point at which machine intelligence is indistinguishable from human intelligence. We call that point AGI.

I still don’t think this is certain. It’s telling that code generation is one of the few things these systems do extremely well. Translating between English and French isn’t that much different than translating between English and Python. These are both tasks where the most likely next token has a good shot of being correct. I’m still not sold that we should assume that LLM-based tech will be well-generalized beyond that. Maybe some new tech will come along to augment or replace LLMs and that will get us there, who knows. Just because the line is going up quickly at the moment doesn’t mean it always will.

From the FAQ on the homepage:

    What is Office EU?
    Office EU is a European productivity suite for files, email, calendars, documents and calls, built on Nextcloud Hub. It brings Files, Talk, Groupware and Office together in one platform.

Looking through the Office EU screenshots, they do look like Nextcloud Groupware/Files/Office with the logo changed.

Mostly adding this because I wasn't sure if it was a new product or not based on a first glance over the Office EU site. Nextcloud offers recommendations for providers on their site, most of which are in the EU [0]. The Office EU website seems to be new since around January of this year [1]. More managed hosts for Nextcloud is a good thing in my book, but I'd be a bit wary to host my stuff with a brand new provider.

[0]: https://nextcloud.com/providers/

[1]: https://web.archive.org/web/20260116234614/https://office.eu...

This is a fully local version of WordPress running in the browser. It's based on the WordPress Playground project (https://wordpress.org/playground/), and allows for editing the WordPress files, administrating the SQLite database, and doing things like installing plugins. A really handy way to play around with the platform.

I loved Christmas Lemmings so much back in the day! The snowfall visualization and the little Santa lemming clearing it. I made a much less impressive snowfall demo a while back based on that (minus the clearing lemming, because I always wanted to watch the snow pile up). https://anderegg.ca/projects/flake/

I'm not annoyed by it when people are trying to make XML compatible documents, but effectively no one is. Platforms like WordPress use self-closing image tags everywhere, but almost no one using WordPress cares about document validation. This ends up meaning that the `<img ... />` is just an empty gesture.

In the early 2000s I was 100% sold on the idea of strict XHTML documents and the semantic web. I loved the idea that all web pages could be XML documents which easily provided their data for other sources. If you marked your document with, an XHTML 1.0 Strict or XHTML 1.1 doctype, a web browser was supposed to show an error if the page contained an XML error. Problem was, it was a bit of a pain to get this right, so effectively no one cared about making compliant XHTML. It was a nice idea, but it didn't interact well with the real world.

Decades later, I'm still mildly annoyed when I see self-closing tags in HTML. When you're not trying to build a strict XML document, they're no longer required. Now I read them as a vestigial reminder of the strict XHTML dream.

EDIT: I just checked, and my site (at least the index page) still validates! https://validator.nu/?showsource=yes&doc=https%3A%2F%2Fander...

EDIT2: Hey, look, if you still want to use self-closing tags where they're not required: go nuts! I'm just explaining why I don't use them anymore.

Nice! You can check my site out here: https://anderegg.ca

About page here: https://anderegg.ca/about/

And the feed is here: https://anderegg.ca/feed.xml

I just learned about the Canon Cat a couple of days ago! I wrote up a blog post with some resources and tidbits: https://anderegg.ca/2025/10/13/the-canon-cat

I had a lot of fun with this emulator on Archive.org, and included some quick notes about using it with macOS: https://archive.org/details/canoncat

Author here. I was also surprised to see this getting a bunch of HN traffic suddenly. I guess the Liquid Glass hate is pretty strong when a dashed-off blog post about a real blog post can randomly do numbers! Heartened to see that others are annoyed by this design as well, though. Hopefully Apple will do something about it, but I'm not holding my breath.

I really need to give Deno another shot. Supply-chain attacks terrify me, and any Node project these days comes with dozens or hundreds of dependancies. It's basically impossible to vet them all, and I don't think this problem is going to go away. The Deno solution here seems like a nice way to handle this. Almost none of my dependancies need to touch my filesystem or directly access the network.

Over the last few years, I've found that most sites clients want can be built with static site generators and JavaScript. PHP is also great and easily hosted! But most times when there's a sprinkling of dynamism needed, it's OK if its happens at build/run time rather than when the page is rendered on a seever. This leads to faster page load and less to worry about security-wise. No shade! I've just been finding this has lead to good outcomes for me.

I wrote up notes here https://news.ycombinator.com/item?id=45180555 but this wasn't coordinated. As I mentioned in the first paragraph of the post, I was following a thread on Bluesky where Adrian and Jeff were chatting. I posted the article there. I didn't know Jeff had posted on Hacker News about this already, and I didn't know he would post my piece. I definitely would have done another editing pass if I had known!

Author here. I woke up to a surprising amount of traffic! Some notes based on the discussion.

This wasn't coordinated between Jeff Geerling and myself. However, I did mention the post in the Bluesky thread that Jeff was included in. [0]

I concluded the piece with “[t]his space is ripe for disruption”. That was a really poor choice of words. I've since updated the piece to better match what I was trying to say.

On YouTube: as I mention in the piece, I think the service is excellent as a consumer, and I pay for Premium.

This piece was mostly written because I've been frustrated that YouTube is effectively the only place for user submitted video on the internet. I wasn't going to write anything until I saw the video from RedLetterMedia that I mentioned in the post. They have a huge following and were blaming something that might be related? Or might not? It's really hard to tell! I'm not a YouTube creator, but I assume having metrics that determine your livelihood shift out from under you as a creator must feel awful.

[0] https://bsky.app/profile/gavin.anderegg.ca/post/3lyeayuckv22...

I ditched Google Analytics on my blog because it more than doubled the size of any page load (the “Universal” version loaded a lot of additional JS). This was fine for a while, as I wasn’t posting much. Later, I wanted to write more. This might be shallow, but I found having numbers associated with my writing helped me keep at it. I also found it fun and useful to see what got traction. I tried Plausible for a while, but it was overkill for my needs, and I grew out of the first tier quickly. I then moved to Tinylytics [0], and have been enjoying that.

[0] - https://tinylytics.app/