What I found interesting this time

by: Artur Dziedziczak

November 23, 2025

“Introducing Beads: A Coding Agent Memory System,” n.d. https://steve-yegge.medium.com/introducing-beads-a-coding-agent-memory-system-637d7d92514a

LLM issue tracker made as replacement for "markdown plans". All to push vibe coding to a limit. My limit was crossed when I finished this article XD

“Beads Blows Up,” n.d. https://steve-yegge.medium.com/beads-blows-up-a0a61bb889b4

Omg he continues hahahahhahaha
I like that he is paying for this. Like he’s putting money into his gambling like adiction without actually thinking about core issues of vibe coding which is halucination and lack of reasoning XD

“When Did People Favor Composition over Inheritance?,” n.d. https://www.sicpers.info/2025/11/when-did-people-favor-composition-over-inheritance/

Article explains when people started to favour composition over inheritance and when to use both of them. Also the author mention that there are other ways to compose programs like lambda functions.

“Self-Hosting DNS for No Fun, but a Little Profit!,” n.d. https://linderud.dev/blog/self-hosting-dns-for-no-fun-but-a-little-profit/

How to self host small DNS server.
After recent Cloudflare fuckup it wqs a refreshing read. The tool used is called knot-dns.

“Mass Claim: Uber’s AI-Driven Pay System Cut Drivers’ Income While Boosting Profits,” n.d. https://nltimes.nl/2025/11/19/mass-claim-ubers-ai-driven-pay-system-cut-drivers-income-boosting-profits

“It’s Your Fault My Laptop Knows Where I Am,” n.d. https://www.amoses.dev/blog/wifi-location/

Detailed blog post about WPS (WiFI Positioning System).
It mentions history of this solution just as it’s applications now. It also shares the issues with it like how companies abuse it to gather more data than necessary.

“Graphics API Is Irrelevant,” n.d. https://youtu.be/xNX9H_ZkfNE

Short video about PPM format and how to use C to render shader videos on CPU.

“Olmo 3: Charting a Path through the Model Flow to Lead Open-Source AI,” n.d. https://allenai.org/blog/olmo3

Another OpenSource model which seem to be trained on date which is not stolen. I really hope this will be the future.
Omg I really need to go back to learning about LLMs architrctures. Half year ago I could say that not much changed but now I see some new ones pop up.

“There’s Always Going to Be a Way to Not Code Error Handling,” n.d. https://utcc.utoronto.ca/ cks/space/blog/programming/AlwaysUncodedErrorHandling

“The Algorithm That Detected a $610 Billion Fraud: How Machine Intelligence Exposed the AI Industry’s Circular Financing Scheme,” n.d. https://substack.com/inbox/post/179453867

Looks like OpenAI and Nvidia commited a freud.