nxc18

I’m wondering if electrician is a good option given the author requires 100% remote.

I’m not sure why people are talking about web components seeing as they aren’t mentioned in the article a single time.

Anyway, my complaint is about the troglodytes who refuse to learn CSS, and who promote troglodism in the form of tailwind and similar antipatterns.

You can also just buy an iPod in 2026, they are still well supported on Mac and PC. You can get them with new batteries quite cheaply. You can even get 30-pin Bluetooth adapters for them.

Extremely easy on iPhone with a Mac. Just plug it in. Supposedly same with iTunes on PC but I haven’t tried it.

I am comparing to GPT-5.3 and 5.2, and I perceive that things have not been noticeably better since then. I also know that I can predict new model releases with high accuracy when my coding agent suddenly becomes regard-level at following instructions and completing simple tasks. This is how I knew 6.0 was about to be released - 5.6 suddenly got unusably bad.

I could point out that I said 6.0 seemed good only in comparison to nerfed 5.6 - people would say I’m just a RSI denialist - but now it is in vogue to accept that 6.0 sucked now that 6.1 is out.

It’s 50/50 on whether it will fuck up implementing an integration test suite when given a list of tests to write and examples of existing tests. It still adds needless abstractions (the reference count codelens in VS Code is good for detecting this sort of thing).

On these metrics it is much better than it was in March of 2025 but no better than it was in March of 2026.

5.6 Sol in the last two weeks became much dumber such that what used to be one correction turned into endless rounds of corrections before just giving up and coding it manually. I’m mostly having it do the “chore” part of coding so it is disappointing that it isn’t better at that.

Was it a poorly-authored PDF without fields? You’d be SOL entirely on Windows in that situation. I find Preview’s PDF editing pretty intuitive generally.

I also like that Preview gets page numbers right in complex documents like textbooks. If a problem set is on page 89 (which might be the 115th actual page), I can just go to it. I’ve yet to see any other free or built in PDF reader do that.

How does this jive with the exponential growth claims? Theoretically sol models are better than the 4 series models I was using at the beginning of the year, but in practice the results don’t seem to be much better. They always nerf the models over the course of the release so it _looks_ like the next version is better but I haven’t seen actual capability growth since ~January, and I’m pretty sure that was all tooling/harness improvements.

Leaning into brainrot is not going to unrot your brain, even - especially - if you have ADHD.

I watched the videos and confirmed they are worse than brainrot.

I don’t think the rote memorization step per se is important, but the knowledge is. You can pick it up just from doing a lot of math work with those basics. Same thing in chemistry - you will learn most of the periodic table automatically just from doing so many lookups.

But IME, not having the knowledge is a major handicap. I did eventually learn essentially the full multiplication table with speedy recall and it _is_ useful in day-to-day life. I think its basic numeracy on a par with literacy if you ever have to do basics like compare prices for consumer goods, sanity check claimed facts/statistics, etc. I don’t think those skills go away when you have AI anymore than they do when you can “trust” Amazon to do the price math for you.

> What more does supporting it entail than being sympathetic to the sentiment?

It is a bad, counterproductive impulse that will hurt their society long term, just as I think xenophobia in the US has been counterproductive.

You should also recognize that there’s a difference between the people who visit, and the people who completely restructure their lives to immigrate. And there’s a difference between immigrants who do their best to learn and integrate into the culture, and immigrants who try to impose their culture on the host nation.

Xenophobia doesn’t really jive with recognizing people as individuals and recognizing that a foreigner can learn Japanese well and follow the cultural norms (aside from the norm of having been born in Japan to Japanese parents, but only because it is impossible for a gaijin to do that).

In other words, I don’t think it’s right to broadly group immigrants into the same bucket as disrespectful tourists, but I also see on a human level how that impulse arises. The greater majority of well-behaved tourists don’t stand out like the problem-causers do.

> Are the Japanese really terrified of foreigners or just annoyed at misbehavior

The “phobia” part of the word is probably confusing you, in actual usage, xenophobia encompasses dislike, “fear” is not necessary. Similar to homophobia or Islamophobia, the term includes fear _or_ dislike/hatred.

I don’t support xenophobia, but after visiting Japan, I can see how it happens. Foreigners are disruptive and disrespectful, often carelessly, but also often when they are trying not to be. I was really trying to be polite and respectful, and I learned some Japanese (not nearly enough), and I could tell I was bumbling at times.

It is one thing when it is one person, it is another when it is every tenth person.

I came away with more empathy for Japanese xenophobic sentiment. I was getting frustrated by rude gaijin and I was one (well I was trying really hard not to be rude).

> Why wouldn't Linux or even FreeBSD be even better?

Because Linux doesn’t offer a rich set of system services built around the needs of desktop users and exposed through curated APIs.

Linux can’t be better at providing a shared calendar API or a shared photos library API, or one of dozens of other APIs because Linux just doesn’t have those features.

Some particular DE could offer those features, but then who is going to build the app ecosystem around that? And isn’t it going to suck when you want to make a Linux native photos app but it doesn’t work on Ubuntu because your LLM was trained on KDE?

My photos are not text files.