Man, Sunday on Hacker News was pretty wild. I saw a few things that really caught my eye. Lemme give you the quick rundown.

LLMs and your job – friend or foe?

First up, there was this huge post, like, almost a thousand comments, about an engineer who feels like LLMs are totally eroding their software engineering career. Super relatable, right?

The comments were all over the place. Some people were like, "Chill out, AI is just a really good tool, a force-multiplier, not a replacement." But others brought up how it's kinda like when machine translation came out and people started accepting "slop" because it was free. One commenter even wondered if the whole blog post was just a setup for an Anthropic or OpenAI IPO – conspiracy theories alive and well!

From rock bottom to building from zero

Then there was this incredibly inspiring story: a guy building his life from scratch after addiction, prison, and a felony. Seriously powerful stuff.

In the comments, a bunch of people shared their own comeback stories, which was pretty cool. There was also a discussion from hiring managers about how tough it can be to find the right person when everyone's resume looks perfect, and how they appreciate honesty about past struggles.

Anthropic, where's the Linux love for Claude?

People are really keen on LLMs, because the next big one was a plea to Anthropic to ship an official Claude desktop app for Linux. You know how Linux users are – passionate!

The comments dug into the usual headaches for developers trying to support Linux: all the different distributions, packaging (Flatpak and Nix flakes came up), and the general quirkiness of various desktop environments. Sounds like a real pain for a company to deal with.

How Linear stays so zippy

For the tech nerds, there was a cool breakdown of how Linear, that project management app, manages to be so incredibly fast. It's a deep dive into their architecture.

One interesting tidbit from the comments was that they specifically *avoided* an "offline-first" approach because it adds a ton of complexity they didn't need. Goes against some common wisdom, but hey, it works for them!

Obfuscated C Code Contest winners!

Always a fun one: the winners of the 29th International Obfuscated C Code Contest (IOCCC) for 2025 were announced. You know, where people write the most unreadable, yet functional, C code.

Someone in the comments asked if AI could win this thing, but others pointed out that IOCCC is more about artistic flair and cleverness than just solving a problem, so maybe AI isn't quite there yet for that kind of creative obfuscation.

Scientists kicked out of a diabetes conference

This one was a bit of drama: scientists got ejected from a diabetes conference for handing out journal reprints. What even?

Apparently, it sparked a big debate about free speech versus conference rules. The conference organizers are a 501(c)(3) non-profit and have rules about maintaining a non-partisan environment. The reprints were from an editorial where the authors explicitly stated the opinions were *personal views* and didn't represent the association. Messy situation!

Using LLMs to *learn*, not just skip

And finally, a cool "Show HN" project called Lathe. It's all about using LLMs to learn a new domain, not just bypass it. This is exactly what I've been thinking about!

The comments were pretty positive, with people talking about how it's great for "interactive hand-holding" and generating small, educational code examples. Instead of just getting the answer, it helps you actually understand the material. Super useful for getting up to speed on new tech.

Anyway, that's the gist of it. Talk soon!