1) In 1983, Ken Thompson, one of the pioneers and giants of computer science, received the Turing award. His acceptance speech, “Reflections on Trusting Trust”, presents a fascinating attack vector. You can read about it here .

2) Russ Cox, over at swtch, has a great explainer going into further detail on how this would work, implementing it, and how the Go compiler avoids it: Running the “Reflections on Trusting Trust” Compiler

3) Finally, there is also a short story Coding Machines based on that idea. The CoRecursive podcast performed the audio version if you prefer listening.

4) I’m very interested in the idea of data ownership. It’s one thing to be forced into someone else’s platform, but why can’t I at least keep my data? Most —but definitely not all— platforms support an extract feature, but can we go further? Martin Kleppmann (of Designing Data-Intensive Applications fame) and a few collaborators seem to think that, and consider CRDTs the basic foundational tool for that. Read up more on inkandswitch: Local-first software: you own your data, in spite of the cloud

5) One of the advantages of being a software engineer is you have the ability to build bespoke solutions for yourself and your family. And yet I feel not that many people take advantage of it. Part of the reason is the conditioning we have over building things: they must be scalable, they must be professional, you need to think about how to monetize them. That’s not true, says Robin Sloan: An app can be a home-cooked meal

6) The README of this repo reads:

The goal of this project is to collect software, numbers, and techniques to quickly estimate the expected performance of systems from first-principles.

And that’s exactly what napkin-math provides.

7) Old news at this point, but I thoroughly enjoyed Anthropic’s Mapping the Mind of a Large Language Model on understanding LLMs and how they work. And I had fun playing with their Golden Gate-obsessed AI back in May.

8) We close this roundup with a classic: Google employee explains internal process of the company, the world is fascinated. In this case, the concept of Readability: Google’s Temple to Engineering Excellence . As much as I dread the overhead of having to ask for more reviews, I have to admit when I have to open PRs in languages I’m not so proficient in I would quite appreciate one of these. And reviewing PRs by others where I am more comfortable in the language is also a good quality control. So maybe they’re onto something here!