The problem is people want to use 2026 tools to write their code but they want to be judged by 2016 standards.

In 2016, if I saw 10,000 lines of code, that carried a certain proof-of-work with it. They probably couldn't help but give the code some testing as they were working up to that point. We know there has to have been a certain amount of thought in it. They've been living with it for some months, guaranteed.

In 2026, 10,000 lines of code means they spent a minimum amount of money on tokens. 10,000 lines can be generated pretty quickly in a single task, if it's something like "turn this big OpenAPI spec into an API in my language". It's entirely possible 90%+ of the project hasn't actually been tested, except by the unit tests the AI wrote itself, which is a great start, but not more than that for code that hasn't ever actually run in any real scenario from the real world.

Nothing about any of that in intrinsically wrong. But the standards have to be shifted. While the bar for a "Show HN" should perhaps not be high, it should probably be higher than "I typed a few things into a text box". And that not because that's necessarily "bad" either, but because of the mismatch between valuable human attention and the cheapness of being able to make a draw on it.

It's kind of a bummer in some sense... but then again, honestly, the space of things that can be built with an idea and a few prompts to an AI was frankly fairly well covered even before AI coding tools. Already I had a list of "projects we've already seen a lot of so don't expect the community to shower you with adulation" for any language community I've spent any significant time in. AI has grown the list of "projects I've seen too many times" a bit, but a lot of what I've seen is that we're getting an even larger torrent of the same projects we already had too many of before.