How I was taught to write an essay in high school:. - intro and thesis.- 3 paragraphs of examples and supporting ideas.- conclusion. How we professionals write computer programs:

LLMs: they're weird computers. And, maybe slot machines. > This is a weird form of metaprogramming: we write “code” in the form of prompts that execute on the LLM to produce the actual code that runs on real CPUs. > – Mario Zechner.

If I were a betting man, I’d wager that human supervision enters the AI lexicon as context engineering and vibe coding fade in the hype cycle.

Reading is always better outside, if you have the means.

That output was great, and it’s nice to know I can do it! But, it took the air out of most of my other projects. So, I’m emerging from a newsletter-writing sabbatical, trying to find a more sustainable pace. More soon!.

"Focus on 2 or 3 things, let everything else slide" is also pretty useful advice. See also, _Four Thousand Weeks_.

"It ships because it's time:thirty" is a pretty good starter constraint. Scheduling (a queue) of drafts makes this more sustainable and generates a little margin for one review before a newsletter or post goes out to the world.

You may ask yourself, how could I keep an 8-week streak of writing and publishing a newsletter going? Reader, surprise, I recently did!. A couple “hacks” helped, but I came away wondering feeling like newsletter’ing was crowding out the other things I want to do.

Lars Ulrich’s kick drums seem like they were recorded to sound best on the stereo of a convertible Mustang with a hole in the roof. (Also, I don’t think he cares.).

RT @PepLovesFriday: THANK GOD IT'S FRIDAY!

Fast (looking) machines.

From the archives:.

This has worked out pretty well over the years. Lots of good ideas started out this way and worked their way in to the library or framework. In short: we can go further if we can agree to cooperate a little and give each other margin to experiment in.

Ruby and Rails developers agree to allow deeply fiddling about inside language and framework internals, in the name of experimentation and pragmatism, for the most part. (Granted, a few ideas caused a ruckus, temporarily.).

BMW and Mercedes agree to limit their cars to 155mph, in the name of safety and avoiding zero-sum games, for the most part. (It's a protocol, if that's your jam.).

Don't rely on working harder or catching better luck.

When your project is running late: dig out, not up. Attempt to explain how you got here in the first place. Share the results of retrospectives. Update the information on your project and scope based on what you've learned.

See also: agile software projects accidentally reinvent waterfall when they seek to add more predictability and remove risk/surprises from the process.

Particularly in software development, if we attempted to prevent or foresee every possible illness, the project would barely get off the ground before it was cancelled.

Sadly, you won’t have a complete model of how long your project will take until you're done with the project. This doesn't mean you shouldn't estimate, it means you should treat them as unreliable narrators.

“Eat trash, be free”, ain’t that the (situationally correct) wisdom of the times? Courtesy of @indirect:.