token body energy anxiety

Three things rattling around in my brain this morning. First, Sam Altman is living rent free in my head, comparing the energy use of a transformer to the energy use of a human. From Matteo Wong’s piece on this in The Atlantic:

Last Friday, onstage at a major AI summit in India, Sam Altman wanted to address what he called an “unfair” criticism. The OpenAI CEO was asked by a reporter from The Indian Express about the natural resources required to train and run generative-AI models. Altman immediately pushed back. Chatbots do require a lot of power, yes, but have you thought about all of the resources demanded by human beings across our evolutionary history?

“It also takes a lot of energy to train a human,” Altman told a packed pavilion. “It takes, like, 20 years of life and all of the food you eat during that time before you get smart. And not only that, it took, like, the very widespread evolution of the hundred billion people that have ever lived and learned not to get eaten by predators and learned how to, like, figure out science and whatever to produce you, and then you took whatever, you know, you took.”

He continued: “The fair comparison is, if you ask ChatGPT a question, how much energy does it take once its model is trained to answer that question, versus a human? And probably, AI has already caught up on an energy-efficiency basis, measured that way.”

Second, Nikunj Kothari’s piece on Token Anxiety, making overly broad statements about life in San Francisco now, but I can sort of relate to this:

Dinner conversations used to start with “what are you building?” That’s over. Now it’s “how many agents do you have running?” People drop the number the way they used to drop their follower count. Quietly competitive. The flex isn’t what you’ve accomplished anymore. It’s what’s working while you’re sitting here not working.

… Waking up and checking what your agents produced overnight is the first thing now. Before coffee. Before texts. You open your laptop and grade homework you assigned in your sleep. Some of it is good. Most needs rework. But you start shipping a plan before you sleep just so you can wake up to more code written overnight. Saturdays became uninterrupted build windows. No meetings, no Slack, twelve hours of you and your agents. Sunday morning X is all terminal screenshots and shipping receipts. “What’d you ship this weekend?” replaced “what’d you do this weekend?”

Finally, Toby Shorin’s Body Futurism. Terrifying.

“Attention is all you need”—the name of the whitepaper that inagurated the current era of transformer models—is a perfect catchphrase for the transition to a pure attention economy where bodies matter most. AI researchers in San Francisco have all started GLP-1s and weightlifting routines, having convinced each other that physique will be the final competitive edge after AI takes all the white collar jobs. If their conclusion is excessive, the premise is correct. We have already entered an economy of charisma which grants status to various kinds of extreme physicality or virtuosity.

The combination of severely curtailed career mobility and the mainstreaming of gambling have created an economic environment characterized by grift. The biggest winners are naturally those who exude charisma and prowess. Charisma has long been the dominant mode in social media, but athleticism is now rewarded more than ever. Jake and Logan Paul pivoted into boxing; IShowSpeed does exhibition races; philosophy YouTuber Jonathan Bi lectures from exotic locations wearing open linen shirts that reveal how fit he is. Less and less, influencers are talking heads. More and more it is the bodies doing the talking. Only from this perspective can the preoccupation with testosterone levels, jaw angles, and height maximizing be understood.

Send help – in the form of books to read, music to listen to, recipes to make, art to stare at. Thanks in advance.

even if it's fake it's real, the michael shannon edition

On a last minute invite from a friend, last night I saw Michael Shannon channel Michael Stipe in a wild mix of nostalgia, fan service, performance art…and true love. Shannon, with guitarist Jason Narducy and a killer backing band (“holy shit, that’s John Stirratt from Wilco!”) has been covering REM and going on tour with it for the past few years. Last night they played Life’s Rich Pageant from start to finish in front of a sold out crowd of Gen Xers at The Fillmore; many of us knew every single lyric from every single song.

This project is wild. I’ve never wanted an 80 minute Netflix documentary more in my life, because I have so many questions. How did this start? How did Shannon and Narducy meet? When did Shannon realize he could pull this off? How did they rope in Stirratt? How did Michael Stipe feel about this when he first heard about it? What are fans getting out of this when they come to shows? And what was it like in the room last year when this happened?

Stash App

Here’s a slightly edited text from a friend this morning:

Using Claude Code feels like taking drugs. I get high when I use it and then the come down is a bitch.

Truth.

I haven’t blogged much about how I’ve been using Claude Code, because everyone is blogging about how they’re using Claude Code, and I don’t want to become just another one of those people who blog about how they’re using Claude Code. Instead, I thought I’d just ship something that is the result of all of this terminal-powered pseudo drug usage.

Background: I am an inveterate note taker. I’ve tried all the apps, all the methods, and nothing quite works the way my brain does. You can blame all the time working in and living with social media, but my brain loves short little bursts of things in reverse chron, with #hashtags and @people mentions, and expects quick search. I don’t like folders, graphs are cool until they’re not, and speed is the most important thing.

So I built Stash App, and it’s currently in beta. If you want to play with it, you can get the TestFlight for macOS and iOS. You can read more at the little site I built while enjoying the high of Claude Code, but here are the highlights:

  • Native macOS and iOS apps, and they sync via iCloud.
  • No folders, just a timeline of notes. Tag them with #hashtags for projects, and @people for mentions. They autocomplete after the first time you use them. #todos get a special visual treatment.
  • Fast keyword search, or tap on a #hashtag or @mention to filter your view.
  • Gmail-style keyboard shortcuts in the macOS app (j/k, e, #, c – iykyk).
  • Archive or trash your notes. Empty your trash to permanently delete them.
  • Just the right amount of Markdown rendering.
  • Export to Markdown, import via Markdown, both in the macOS app.

You can learn more here; there’s a draft FAQ which has more details. I have been using it while building it for the past few weeks and it’s now part of my workflow because it works like my brain works. It may not work for you, and that’s fine! But if you think it might, grab the TestFlight and check it out. All feedback welcome.

Oh, and there are more drug-induced things coming soon. Stay tuned.

three books dot net

As a moderate, semi-dangerous San Francisco liberal, The Ezra Klein Show is required listening. The Vox co-founder and New York Times opinion contributor brings on smart guests to talk about smart things. At the end of every episode Ezra asks his guests this question:

“What are three books you’d recommend to the audience?”

Every time I listen to the show I think to myself: “I should write these down! I’m always looking for good non-fiction books to read, and these seem interesting!” Now, the Times does the Right Thing by including all of those book recommendations in the RSS feed, and linking to them on their site. But I’m lazy, and I wanted them all in one place.

So I put them all in one place: 3books.net.

  • Built with Claude Code, hosted on Vercel & Neon, with book data from ISBNdb.
  • The system parses the RSS feed from the Times, and uses GPT 3.5 to pull out the recommended books, and write little bios of the guests. Stuffs all that into the database.
  • It looks up the books in ISBNdb, grabs some metadata about the books, stuffs all that into the database.
  • The system does the best it can to associate a single book across multiple episodes (The Origins of Totalitarianism has been recommended in seven episodes!).
  • The processing isn’t perfect! Sometimes it will include a book written by the guest! I’m OK with that.
  • The home page shows the books recommended in recent episodes (and skips any episodes that don’t have book recommendations), detail pages for books and episodes show, well, details.
  • I’ve started doing some basic “recommended with:” pivots on books, so you can see what other books have been recommended alongside the one you’re currently viewing.
  • Search sort of works! It’s not fancy.
  • I like the “random book” and “random episode” features.
  • The system currently has about 1300 books across nearly 500 episodes. I want to explore more ways to browse this corpus; it’s a tidy little dataset.

I like projects, I like books, I like podcasts, I like RSS feeds. I think I would like Ezra Klein! Seems like a nice guy. (Hmmm, is all of this just an extreme case of parasocial fan behavior? Yikes.)

And I love making software. Making software with Claude Code has been a very interesting experience. I’ve gone through all the usual ups and downs – the “holy shit it worked” moments, the “holy shit the robot is a f’ing idiot” moments, the “wow, you really are a stateless machine without any memory, aren’t you” moments. But it’s super fun and incredibly empowering to have what Josh Brake calls “an e-bike for the mind” at your beck and call.

The site isn’t perfect, and there’s still more that I want / need to do. But if it’s good enough to buy a domain name for, it’s good enough to share.

So, go. Browse. Find a good book to read! Let me know what you pick.