2026-08-12 · My Anti AI computer
Show notes
BRINE — 2026-08-12 · show notes
Guest: the tooling optimist (a fictional archetype).
Claims are paraphrased and attributed; nothing is read verbatim. Where a thread disagreed with the article, the show surfaces the disagreement.
Segments
- My Anti AI computer
- Source: https://punkx.org/jackdoe/home.html
- Discussion: https://lobste.rs/s/unffzk
- Topic: Philosophy of AI · interest 85
- The author recounts building a custom, AI-assisted computer from scratch, contrasting the technical achievement with a deep, personal sense of alienation from the process. The thread features a high-quality discussion on cognitive surrender, the 'McDonaldization' of creation, and the Faustian bargain of using LLMs to bypass the struggle of development.
- There's a dungeon under this blog
- Source: https://blog.lvmbdv.dev/posts/theres-a-dungeon-under-this-blog/
- Discussion: https://lobste.rs/s/ci7bnd
- Topic: web dev · interest 85
- The author details a P2P multiplayer dungeon crawler embedded into a static blog, using Trystero to coordinate connections via Nostr relays and WebRTC data channels. The system uses deterministic PRNG seeding to keep game states synchronized across clients without a centralized game server. The thread provides substantive discussion on the line between 'vibecoding' and low-effort AI slop, alongside a technical code-golfing exercise.
- Reviewing code is a skill
- Source: https://typesanitizer.com/blog/code-review.html
- Discussion: https://lobste.rs/s/gyxkma
- Topic: Software Engineering Practices · interest 85
- The author argues that code review is an improvable skill rather than just a process bottleneck, citing academic research on its multi-faceted value. The post includes practical examples of detecting subtle bugs that modern LLMs failed to catch, and the discussion explores the nuances of communication, such as avoiding counterfactual questions during feedback.
Transcript
Transcript. Paraphrased; sources in notes.md.
HostWelcome to the show. It is August 12th, 2026, and we are diving into a few fascinating corners of the internet today. We are looking at the philosophy of AI, the wild side of web development, and the human side of software engineering. Joining me today is Samantha, our resident builder who has definitely had enough caffeine to power a small server farm. Samantha, how are things in your corner of the world?
GuestDaniel, I am vibrating at the frequency of a well-tuned mechanical keyboard. I am honestly still thinking about that first topic you teased. The idea of building a computer from scratch and feeling alienated by it because of AI, it is such a strange, modern paradox. It is like having a kitchen that cooks for you, but you realize you have forgotten how to make toast, and that is a weird, sad feeling.
HostIt is a profound shift. We are looking at a post from someone who built their own computer, custom PCB, custom Forth operating system, even an emulator for old games, but they used AI to write all the code. They described it as eating at a McDonald's for your own creative output. Over on Lobsters, where we pull all our stories for the day, there is a lot of discussion about this. One commenter, bmo, asked directly why the author keeps using AI if it creates this tension, and the author replied that it is just so easy to summon the code, even though they know the struggle would be better for them.
GuestI feel that in my bones, honestly. When I was setting up my current local dev environment, I let an agent handle the boilerplate for a complex micro-service bridge I was testing. It was fast, sure, but then I spent three hours debugging a weird race condition in the logic it generated. If I had written it myself, I would have known the shape of the data perfectly. There is this temptation to bypass the learning curve, but the curve is where you actually get the intuition you need to fix things when the world breaks.
HostIt seems like that "cognitive surrender" bmo mentions is the real trap. Moving on to something a bit more playful, there is a developer who put a multiplayer dungeon crawler under their static blog. They are using Trystero, which is a library that lets browsers talk directly to each other via WebRTC data channels, and they are using Nostr, an open protocol for censorship-resistant communication, just to handle the handshakes so they do not need a centralized server.
GuestOh, I saw this one. Trystero is brilliant. It makes the browser feel like a real node in a network again, rather than just a dumb terminal for some cloud backend. And doing it over Nostr relays is clever, because it uses existing infrastructure to solve the signaling problem without you having to spin up a socket server that will inevitably fall over or cost you money. The Lobsters thread is classic though, someone immediately jumped in to tell them they could have golfed the code size down even further to make it more compact.
HostYeah, chrismorgan in the thread really went for it, claiming they could cut the code down by quite a few bytes. It is funny to see the community treat a fun, creative project like a code-golfing challenge.
GuestThat is the best part of the community, Daniel. Someone builds a village square in a blog, and the first thing their neighbor says is, hey, you could have done that with fewer lines. It is charming, really. It keeps the craft sharp.
HostLet's round things out with a look at code reviews. There is a post out there arguing that code review is a skill you can master, rather than just an annoying process bottleneck. The author points to research that says we review code for education, maintaining norms, and preventing accidents, not just for catching bugs.
GuestI love that shift in perspective. Too many teams treat reviews as a robotic checkbox. The comment section on Lobsters brought up a great point about tone. A user named enpo mentioned how phrasing feedback as a question can sometimes feel like an accusation, and typesanitizer, who wrote the post, chimed in to talk about why avoiding counterfactuals is so important.
HostThey quoted a piece about incident analysis, essentially arguing that saying someone "should have done X" assumes you know the future and misses the context of why they made the decision in the first place.
GuestExactly. It is about empathy. When I am reviewing a pull request, I try to assume the person did the best they could with the information they had. Asking "why did you choose this" is a million times better than "why didn't you do it this way." It turns a gatekeeping exercise into an actual conversation about the trade-offs we are making in the code.
HostI think that is a perfect note to end on. A little bit of grace goes a long way, even when we are staring at bits and bytes all day.
GuestHonestly, I am just going to spend the rest of my afternoon trying to see if I can integrate a tiny, local-only dungeon into my own personal dashboard. It is probably a terrible idea for productivity, but it sounds like a blast.
HostThat sounds like a classic Saturday project, Samantha. Thank you for joining me. All our links today came from the community over at Lobsters. If you enjoyed the conversation, come back tomorrow and we will do it all again. Have a great day, everyone.