September. My god this has felt like the quickest summer ever. Our hope for a summer-of-love style return to normal (you remember, from five newsletters ago) was thwarted by the rise of delta and what, to me, seems like undue anxiety among the vaccinated class. I maintain that, if you’re otherwise healthy, it’s our civic responsibility to responsibly hold and go to events to start rebuilding the social fabric. As the season changes and pumpkin spice reappears, we must all do our part (if you're in San Diego message me).
In June I participated in an AI Vignette workshop and wrote a short story called Jackpot! which I've finally gotten around to posting. It's a bit of a silly piece, but it captures my fear that new technology will very effectively satisfy our base impulses while neglecting or diminishing our better natures (see also: a brave new world).
This has been an evolution in my thinking over the past decade; I used to hold strictly to a libertarian perspective that people's behavior reveals their preferences more honestly than anything else, and that if they want to indulge in vices that only hurt themselves we should let them. I now think it's more complicated than that. We're made up of parts that exist in a complex ecology internally and externally, and our surrounding environment reinforces certain parts, weakens others, and can act in a predatory manner to strengthen internal parts that are misaligned with our aspirational selves. In particular, our current environment is relatively better at satisficing base parts than it is at promoting higher, more nuanced parts, and our technological trajectory is increasing this differential. All this to say I have more sympathy for the small c-conservatives who warn about the pernicious effects of new tech.
Last year I also wrote two other AI Vignettes, Skin Deep and Terms of Service. Skin Deep in particular is my favorite likely-hyperbolic prediction for the next decade - persuasion technology will rapidly improve and won't, by default, be aligned with our values.
Writing short fiction pieces is fun and unexpectedly helpful for thinking through one's actual model of the world. It's not a replacement for more rigorous forecasting, and in particular narrative bias will and does rear its head, but trying to make a story hang together, and discussing the plot holes with others helps to make the scenario planning concrete. Numbers in a spreadsheet don't seem to cue the brain to think about the details in quite the same way as a good story.
Related: Metaculus is hiring analytical storytellers to combine quantitative predictions with narrative storytelling.
Browser Interviews Applied Divinity Studies: ADS has been one of my favorite new bloggers over the past year. This is a good interview with him covering lots of topics like how to write well, cultural vibes, and how to be weird.
Amazing Marvin: I've started experimenting with Amazing Marvin as a todo manager and I like it - it has modular strategies for experimenting with different productivity techniques. For instance, I can easily add an Eisenhower matrix to my workflow and see if it helps me prioritize. I'm betting in the long run this flexibility will lead to me discovering better approaches to getting things done. If I keep using it over the next month, it'll make its way to my recommended things list.
Tweet thread about the Afghanistan War Docs: Things that particularly caught my attention:
In the later years there was basically no plan, with the top General unable to articulate a meaningful high level strategy.
1/12th of the Afghan GDP was stolen by corrupt government officials. One audit revealed 40% of contract funds were stolen. Opinions differ if the corruption is worse in Afghanistan than DC or if it’s just a different type.
Twenty years, thousands of lives and trillions of dollars later, all of the people who ran the war seem to have done just fine because their career success was based on looking forward not back.
Long Covid is not necessarily your biggest problem: One of my worries about COVID was the specter of long term covid symptoms. Elizabeth Van Nostrand makes a compelling argument that, for people in my age bracket, it’s less of a concern than other potential health risks I’m already taking.
Neurocracy: A new alternate reality game/murder mystery set on a 2049 version of Wikipedia. This Verge article contains a lot of interesting details - they used GPT language models to create text to populate filler and background material on the world. I’ve started my investigation.
When does the devil keep his deals: Good game theory analysis to keep in mind when dealing with the devil or other less demons.
Someone Dead Ruined My Life... Again: I really enjoyed this dive into how difficult it is to do accurate scholarship. CGP Grey does deeply researched interesting videos - I particularly liked his pirates video - and in this video he details how much work it took to investigate a single historical claim about the origin of the name Tiffany, and how much false or generally unsupported info is floating around. I haven't heard of a general 'replication crisis' in historical research, but given videos like this, the degree to which historians re-interpret events through present day ideological lenses, and the lack of robust error correction mechanisms in academia, all makes me think that there's a tremendous amount of bullshit in the field.
Watership Down: Reread the classic and it’s a great piece of fiction.
Jefferson Airplane's White Rabbit: Speaking of the Matrix, I heard this song in the new Matrix trailer and have been feeling it. It's great vibing music. And in another effortless segue, I’d say my go-to ‘vibe’ music over the past two years is Glass Animals. It’s a blend of hi and lofi that really does it for me.