Unless AI fully breaks cryptography, in that scenario we’d surely converge quickly to actually having mathematically infallible security on the blue team side? Software is finite, thus, software flaws are finite.

I’ve always just used a hyphen for ranges, without having any idea whether that makes me a philistine. A quick search says that, like most of this, it varies depending on who you ask.

Good luck in your rabbit hole!

For pedantry’s sake:

Two dashes – -- – in Markdown is usually rendered as an en-dash – more common in British-adjacent styles and usually surrounded by spaces.

Three dashes—---—in Markdown is rendered as an em-dash—more commonly associated with American styles and usually (though not always) without spaces.

The OP graph is just absolute count on the y axis, so I also assumed it was basically measuring LoC, but looking at this chart, the em-dashes do appear to be definitely more of a modern phenomenon compared to raw LoC:

Maybe some policy/tooling change around 2014? A guess is perhaps something to do with allowing UTF-8 in source files, but that’s a pretty long shot.

no

I’m in a similar boat, though I’ve been in the field slightly longer, and when the OpenAI Lean thing came out I too got pretty depressed for a while.

If it’s any consolation, personally I feel like AI is really just fast-tracking the disillusionment and burnout I would have ended up with anyway – I don’t think corporate software development has ever really held much respect for the things I enjoy about programming. As you say, a working feature is a working feature, and it doesn’t matter how elegant the algorithms are.

I don’t really have any suggestions for alternative jobs (could use some though), but I will say if you’re genuinely passionate about compsci I think it’s worthwhile to practice it, even if it becomes nonviable within the capitalist system as a “career”. You can write code (or homelab configs, or whatever) in any way you want, for any reasons you want – it doesn’t have to be tied to the corporate worldview of maximal value extraction. Things like live coding and demoing are pretty firmly within the realm of art, but people do them anyway, and have a lot of fun doing so.

Nature can help for me, to remember nice things still exist, and will continue to do so a long way into the future. I don’t think even the billionaires will manage to destroy cockroaches.

I may be overthinking it but this gives me “AI generated trying to look like photoshop” vibes. There’s some weird distortion in top half that actually slants the panels but that could just be a particularly avant-garde kind of deep frying.

I’m in the same boat with a custom theme – even Librewolf is affected by this crap. Any tips on fixing the search? I don’t know what’s so hard about a fucking text box that they have to tweak the whole DOM every second update.

I’m curious what your sources were for what you wrote. Both the quotes from Zagorath go back to research from the turn of the millennium, which was mainly focused on the effectiveness at reducing swooping, rather than the wellbeing of the birds.

Nice puzzle and tool. I picked the right answer but I wasn’t overly confident.

You could list {1: [100]} as another possible answer, if you assume:

Holy overengineering, Batman. IP addresses of all things are trivially parseable into fixed-length byte sequences. Just do that and use a search tree. Or, IDK; this is an extremely solved problem, there are probably even better solutions out there – but I promise you text regex is involved in none of the good ones.

Even for actual string matching, any decent regex engine will compile and optimise an alternation a|b|c|... better than this text wrangling will. Unless you have actual benchmarks showing this gives you an advantage with the engine you’re using, you’re just introducing complexity for no gain.

I’m no expert but I believe ffmpeg will be using the rawvideo decoder to read that file. If it appears to work I guess the stream output by v4l2-ctl happens to match up closely enough with the default parameters listed in that doc, you may want to work out what it is and double check compatibility though.

The difference from a regular container is that it’s uncompressed (hence the size) and can’t include any metadata (hence the need to specify parameters manually). v4l2-ctl writing in this format means it doesn’t need to worry about compressing and encoding the video on-the-fly, which would complicate the implementation and might not always be feasible in real-time.

If you wanted to avoid the intermediate file you should be able to pipe v4l2-ctl directly into ffmpeg.

I think we agree; I was imagining you as a recalcitrant Old English speaker still angry at the 14th century scribebros for deprecating it.

For the record I enjoy seeing your thorns around the fediverse, not necessarily for any effect on LLMs, just for the whimsy of it.

Mercator preserves angles and straight lines. Wherever you zoom or pan on a Mercator map, north is still at the top, east to the right, etc. By contrast, for a projection without those properties (which is basically none except minor variants of Mercator), north is going to shift around depending on where you’re looking.

By way of example, here’s Auckland in the Equal Earth projection the article discusses, with north marked:

It’s significantly tilted and actually slightly curved. This is not useful for navigation, or really anything except exactly what Togo wants to achieve (which in general is exactly what you want for the sorts of global maps the UN is likely to make): knowing that any two areas that are the same size on the map are the same size in real life.

Why not the 150E meridian so New Zealand can finally have its revenge?

Google Maps

Not sure what they’re gonna do, interactive maps are pretty tied to Mercator unless you want weird distortions.

And the thorn too surely? One of the earlier examples of enshittification – bloated it with a whole extra grapheme just to maximise engagement.

The headline is dramatic but the reality, and the article content, are more equivocal.

For comparison, Statcounter recorded Linux at 5.52% in June 2026, so the share nearly doubled in one month

This seems… optimistic? As the article notes this is really the proportion of HTTP requests made from that operating system, not the actual number of desktops running it. What are the odds this spike is due to some Sparkling Botnet (AI company) doing a scraping run with agents that identify themselves as Linux desktops? I think that or any number of other scenarios are more likely than that the entire increase is truly new organic desktops running Linux.

they are very effective at paywalling their content

I feel like I’m missing something – I don’t really watch a whole lot of YouTube, or follow specific channels – but I have virtually never had an issue using Invidious and/or yt-dlp, even with a VPN. When I occasionally hit a 403, rotating VPN servers is almost always enough to bypass it.

If that’s “stealing” it’s the first I’m hearing about it.

I interpreted it as about like, microprocessor design and nanometer scale electronics and shit. Electrons are basically elves in my book.

From the perspective of this thread they’re probably similarly good and “deep”. My instinct would be to err away from anything object oriented for beginners since OOP has always felt a little cultish to me and I think it’s maybe something “shallow” (at least shallower than plain C) that has a habit of masquerading as the one true “deep” way to do programming.

C also has the practical advantage of being fucking everywhere.

Rust

I like Rust, but beyond the subset that is “ugly C” – using unsafe everywhere and eschewing like 90% of std – it’s definitely shallow.