Joy & Curiosity #11

Interesting & joyful things from the previous week

Not a lot of reading, listening, watching this past week. The pendulum has swung into the other direction again and I ended up putting quite a bit of time into this week’s post, How I use git. There’s also been a ton of code to write at work — not complaining, I’ve been enjoying it very much. (I’ve also done a nearly 4 hour stream this week, hacking on Zed, in case that’s your thing.)

And yet!

There have been some very interesting & joyful things I came across this week.

• This article, Always Measure One Level Deeper, by John Ousterhout (the John Ousterhout, yes) is exceptionally good. Mostly, I think, that’s because it’s based on practice, not theory. “Performance measurements should be considered guilty until proven innocent. When students come to me with measurements, I often challenge them by asking: ‘Suppose I said I don’t believe these measurements. What can you say to convince me that they are correct?’” It reminded me of so many situations in which I “just” wanted to record some benchmarks and numbers and ended up wondering what’s real and what isn’t, what’s a symptom and what’s a cause, what’s an improvement and what’s an accident. Highly recommended.

• These Reflections on Palantir are fascinating. If nothing else, Palantir is a very interesting company, and reading this was probably the closest I’ve ever gotten to figuring out what they do. There are some very real software-rubber-meets-the-customer-road moments described in there and some interesting observations of a company culture. “I’m not sure if they still do this, but at the time when you joined they sent you a copy of Impro, The Looming Tower (9/11 book), Interviewing Users, and Getting Things Done.” More companies should hand out books to new employees (I’ll never forget that Nick at Sourcegraph recommended I read Orbiting the Giant Hairball and now I’m looking at it, because it’s sitting on my shelf right here, cherished.)

• I spent Thursday morning reading through, whorl, “a single file, std only, async Rust executor” in ~600 lines. It nails what it set out to do and made me think of three things: Phil’s tweet that I came across a day earlier saying that “when writing pedagogical code, please please please put it all into a single file”, Patrick’s 200 and change project, and the 500 lines of less book.

• There are new Kindles coming out! I bought a new Paperwhite last year (after trying the Oasis and not liking it) so there’s really no need. Right? No need to get one with color, right? That’d be dumb, right? Who needs highlights in color, right?

• Very much enjoyed this Acquired.fm episode with David Senra, host of the Founders podcast (which I also enjoy very much). If I had to pick one word to describe this episode: meta. In the best possible sense. “The conclusion I've come to is, advice is an average, and reality is a distribution. Averages suck because they hide the distribution. You want to know the shape of the distribution.”

• “Jim Sinegal, the founder of Costco, says: ‘If you don’t spend 90% of your job teaching, you’re not doing your job’” — made me think of all the most shallow of all the trite phrases: “those who can, do; those who can’t, teach.”

• This is a good tweet.

• And this made me want to play Factorio again.

• “We must remember something important: life is not meant to be rushed through. It is not a race, nor is it a problem to be solved. It is an experience to be lived, and living well requires presence.”