Wednesday, 10 June 2026
What I found interesting is the point that “labels are necessary and useful, but when you depend on the label alone, the value you can draw out gets limited.”
- On Apollo 13, they solved the problem with the CO2 scrubber by using a sock as “padding/a plug to fill a hole.” In other words, they stopped depending on the label “a sock is something you wear” and re-framed it as “cloth.”
- Applying the same logic to AI: if you carve out parts of an existing job (coding, debugging, refactoring) and hand them to an agent while leaving the rest unchanged, it works but it’s “low leverage.” There’s a better way. Aren’t we labeling “development with AI” as just that?
Developer Tea: Principles Oriented Thinking as a Durable Skill in an AI First World https://podcasts.apple.com/us/podcast/principles-oriented-thinking-as-a-durable-skill-in/id955596067?i=1000771990792
Maybe there’s also the issue that once you summarize or conclude a certain amount of something, that itself becomes a label, and the value you can draw out from it gets fixed in place.
I want to put together my BoldVoice notes so I can memorize them through repetition. What’s the best way to do it…
I prepared the documents to submit to the Legal Affairs Bureau. For the company’s official seal there were three sizes — square, round 1 (1.65mm), and round 2 (1.8mm) — and I went with round 2.
Because I ate a bit too much yesterday, sure enough my weight had gone up. I’d forgotten I’d bought some prepared side dishes and ended up eating those, plus I cooked one go of rice and ate it all straight up. My body felt heavy when I ran 11km too. Gotta watch the overeating.