Tom MacWright


Another view from Angel Island

The only album I added to my collection this month was Outkast’s Aquemini, which has that singular sentimentality that only André 3000 seems to access. But discovering the COLORS YouTube channel was also a big change. See the EarthGang video first. It’s like the 2020 edition of La Blogothéque.

I’ve been practicing classical guitar a bit more, which means and, when I can find them, recordings of those arrangements. It’s always interesting to see different interpretations of the same song, showing that there’s so much room at the edges.


I finished a few books this month. The New Me felt so distinct from similar books because it communicated emotion without poetry. I was impressed with how well The Phoenix Project, a ‘business novel’ achieved its premise.

I’ve long been skeptical of blockchains for many reasons that are constantly being reaffirmed by the failures of the technology, people, and business associated with it. But smart people tend to bring up cross-border remittances as an example of a real problem that blockchain can solve. So it was nice to read JP Koning’s “Transferwise, why so fast?” as a counterargument for why blockchains aren’t necessary or superior for remittances either.

Another from Angel Island

Oxide computer company is one of the most interesting new startups, with a star-studded team, fascinating podcast, and interesting pitch. Jessie Frazelle’s “Power to the People”, a deep explanation of how servers get power, including AC/DC conversion efficiency, heat management, and UPS battery systems, was great. It pulls together physics, electrical engineering, software, and economics and gives you a real appreciation of the infrastructural advantages that “hyperscalers” like Google have created for their data centers.

Much of the reckoning of the past three-plus years can be traced to delegating policy operations to Sandberg, while the company’s all-powerful growth team — which reported to Zuckerberg – ran roughshod over everything else. Policy has always lagged behind the messes created by the growth team, and this was by organizational design.

I’m pretty full-up on Facebook corporate gossip, but the highlights from Steven Levy’s new book helped me understand some of the structural and personal dynamics behind it all.

“It’s just, when he has Fox, he has Obama to hate. If he doesn’t have that …” She kept looking over her shoulder. She was terrified of him. “I’m sorry,” she said. “I just need him to have Fox.”

Lauren Hough’s account of being a cable technician made me think about the nature of rage.

I’ve been reading through How to reassess your Chess, trying to level-up from my pretty weak ranking (1218, at the time of writing), as well as slowly but surely drafting an article about how chess has made me think harder about the nature of intelligence.


Since leaving Observable, I’m no longer working on “the future of programming”, but I’ve been following a few similar efforts. WhiteBox is a really interesting one - C++ instead of JavaScript, and it’s a Windows application. But it seems like a riff of the web developer tools in browsers, which I still think are some of the finest programming tools ever created. But most of the demos show working programs, and so much of my interest in that space is in crashes. High-qualty debugging - being able to step through programs, elegantly work with half-working programs, see good, clean tracebacks, is so incredibly valuable and underinvested by toolmakers.

I’ve also been thinking about frequency and depth in work. I think I was lucky to start my career at Development Seed, at a time the company was creating micro-websites for NGOs with project timelines in the sub-6-months range. It gave me such a valuable opportunity to get a clean slate on a schedule, making basic decisions again and again, getting to try out slightly different approaches each time. It was very similar to the workload of a newsroom, and several coworkers either came from one or went to one after that. On the other side of the spectrum, working for an API-as-a-service company means that you focus on one big project for years at a time. Your engineering decisions follow you around, and if you need to change a decision, it’s a dangerous “rewrite” that you’re embarking on.