We took Rufus to our downtown YMCA for his first toddler swim class, which was perfectly scoped to half an hour. He made a valiant effort to blow bubbles in the water (ended up licking the surface), laughed when I flipped him from chest to back to chest again, and didn’t try to run off to the adult pool. The night prior we went bowling, also a first, and it was all we could do to keep him from dancing into the adjacent lanes. In the past month he’s had his first ice cream cone, rode his first choo-choo, and saw – and touched! – an in-service firetruck at our local fire station, where a firefighter nicknamed “Foxy” gave him his first cookie. He’s a happy, mischievous (we call him “Asian of chaos”), and deeply funny kid. I’m trying my best to savor these moments.

Work-wise I was out in Mountain View for a few days covering Google I/O (their annual developer conference) for Every. On the flight there I read Sebastian Mallaby’s thrilling The Infinity Machine, a profile of DeepMind CEO Demis Hassabis (which I recommend to the much-puffier documentary The Thinking Game).

Being there gave me the impression that there’s a growing chasm between what researchers at the frontier labs see in their most advanced AI models’ impacts on scientific research and robotics, and the error-prone results that the most people see in the AI summaries tacked onto the tops of search pages. I believe that Hassabis believes, based on the progress he’s seen working on AlphaGo and AlphaFold, that – as he so casually said during the keynote at I/O – we’ll in our lifetimes “cure all disease.” I also believe that of the frontier labs, Google is maybe the one most cognizant that they need to prove that people’s lives aren’t going to get worse on the way there.

That said: I also had a hand in editing this essay by Dan arguing why the fears of AI automating away human jobs underestimate our own intelligence. It’s a pragmatic counterpoint to both the hype and doom on either extreme, and Dan’s advice in this weird – maybe weirdest – moment in time is to “surf the models.” The piece is a brisk 8,000 words, and took a huge effort across the company.

My last night in California, I stopped at Books Inc. in downtown Mountain View and, as one does in Silicon Valley, picked up a copy of Iain M. Banks’s Consider Phlebas for the flight back—my first Culture novel.

But I’ll save my thoughts for when I finish.

Jack

p.s. June’s digital mending circle is next Tuesday the 9th at 7:30–9:00pm Eastern. I’ll be working on my website. Reply for the Zoom link if you want to join.

Unlock this post by becoming a patron

This month Rufus had his first overnight at my parents’ place without us (he did great!). In part, it was so Julia and I could go see poet and essayist Ross Gay, who was in town as part of Oakland University’s annual Maurice Brown Memorial Poetry Reading series.

The reading took place, of all places, on the basketball courts of OU’s campus athletics center, with a casual shootaround beforehand on the unbleachered half of the courts, the first time I’ve dribbled a basketball in years. The host later quipped about sweaty poets, Gay read his basketball-related work, of which there was more than I’d realized, including Be Holding’s appropriately drawn out opening ode to Dr. J’s legendary baseline layup in the 1980 NBA Finals. (Gay played basketball in school and has coached high school players, and is quiet tall.) A good way to spend an evening.

This On Being episode was my first introduction to Ross Gay’s unabashed gratitudes, and since the event I’ve been reading the poems in the so-titled book. A good way to start a morning.

Speaking of good ways to spend an evening: community theater. We saw, last night, a friend act in our local improv theatre’s musical parody of Jurassic Park, which was the first movie I ever saw as a kid in an American movie theater, and which struck me mid-show as a perfect parody target for this moment of nineties nostalgia. The props were perfectly DIY. The audience was mostly millennial. We hummed along, only half ironically, to the cast ironically humming the ending theme. An earlier musical number was sung by Samuel L. Jackson’s cigarette. It was excellent.

My latest piece for Every is on Living Software, and is in some ways a spiritual successor to The Slow Web. It starts:

Lately, I’ve been wishing that more software had a “freeze” button.

When pressed, the product would crystalize in its present state. The feature set would lock, and the interface would solidify, as if dipped in carbonite. There would be no more new updates. No changes whatsoever.

I want this button because companies are loading apps with more and more features, whether AI or the result of AI-accelerated development, making the tools unrecognizable. The additions are even more jarring for apps that I only use occasionally, like Figma. There, a chat box now beckons to describe my idea to make it come to life. A “Recents” toolbar above it has buttons for Figma Sites, Figma Buzz, and Figma Make—all launched last May. A sidebar module encourages me to try an AI image- and video-generation product called Figma Weave—and which I have to log into separately using my Figma account.

And here I am just trying to update the gradient on an app icon.

No shade on Figma (okay, maybe a little shade), but it’s more my gripes with The General State of Things and my distaste for the technical words “deterministic” and “non-deterministic” that the industry uses to talk about traditional software and AI software.

This makes the piece sound pessimistic—it’s not! I just think that better words (I propose, in the piece, the alternatives tool-like and living) can help makers make legible, for ourselves and others, what we’re trying to make.

To end: One of my new work colleagues also runs Flow State, which recommends two hours of ambient/instrumental music each weekday morning. A good way to start the workday.

Jack

I’m thinking about formats today. Though, it’s more accurate to say I’m often thinking about them for this newsletter. In the past, it’s been a book publishing dispatch, travelogue, link grab-bag, weekly pieces of first-draft poetry, and other, more shortly lived experiments. Since 2021, I’ve been pretty good about including a photo with every issue, though my cadence has slipped from weekly to monthly since Rufus was born (maybe not surprising to any of you parents reading).

All that’s to say that I’ve let go of the idea that this newsletter will ever have an entirely fixed format or rhythm. But I can feel, in this moment of time, a new one sorting itself out. I don’t know what the new format looks like quite yet, so I’ll do some thinking out loud here, in hopes that charting the ecosystem might help me better understand it.

If this doesn’t interest you, skip it! You’ll find more typical links in the next section. Here goes:

• My new position at Every seems the ideal outlet for my tech essays. Sometimes, these are the same ideas I’ve floated here as first drafts, just with more time for elaboration, background research, and revision (did I already say this last month?)
• My notetaking system has changed a bit recently, through the combination of new Bebop updates (coming soon!) and AI tools that help me review and resurface older notes and journal entries (which I’ve always struggled to do). I end up losing fewer ideas!
• I’d like to develop many of the non-tech ideas too, in ways short and long, and post or publish more elsewhere about books, architecture, photography, or plants and gardening.
• Including a photo in each newsletter has turned it into a bigger production than just sitting down Sunday nights and firing something off. I feel like I should take more photos during the week, and then want to take the extra step of editing them if I do. Another reason it’s become monthly instead of weekly.
• The monthly cadence, though, makes me feel obligated to deliver something substantial than I sometimes have energy for.
• For Dumpling Club members, I write a separate monthly letter that’s typically more personal or in-the-weeds, that I think might only interest the hardcorest of followers. But I haven’t clearly defined that for myself either (this email, for instance, is starting to feel more like something I’d send Dumplings).
• I have newsletter fatigue as a reader. I’m subscribed to too many, and went to the extent of vibecoding a personal RSS/read-later tool that makes me feel a little less guilty about not reading everything.
• I don’t love the term “vibecoding.” In practice it’s more like “painting with code.” But I digress.
• I’d started doing longer write-ups of projects here, but again, these are bigger productions.
• Maybe I should start blogging again? And just write shorter blog posts?
• Or longer social media posts?
• I’d rather not.
• I like the idea of POSSE in theory, but I don’t love the fragmentation. I’ve started experimenting with cross-posts to Patreon, but it all ends feeling like another platform to manage and check (and get sidetracked by).
• Then again, if I could more easily post and participate in those discussions without ever dealing with algorithmic feeds …
• The above doesn’t even get into how/what else I could be writing about my novel-in-progress.

So that’s on my mind. Reading this over, I don’t know if I’m any clearer on formats. It just feels more than ever like being a working writer in 2026 means running one-person media company in your spare time.

My March piece at Every was all about maintenance, Digital Mending Circle, and OpenClaw. One thing that didn’t make it into the piece is that after some of the initial technical maintenance around my weird little AI assistant, the act of maintaining it turned more into like the act of maintaining a relationship with a person, or pet, or plant. I’m hoping to explore this angle for the next essay.

I also found, looking back over my notes, a link to this piece on mesolomania, the obsession with intermediate scales. I don’t remember exactly what I thought it had to do with maintenance (the mention of bridges? a connection to Stewart Brand’s “maintenance layers” idea?) but it’s a fun piece.

Speaking of Digital Mending Circle: Our next one is Tuesday, April 14th at 7:30–9:00pm Eastern! Hit reply if you don’t already have the Zoom link.

Rufus isn’t yet old enough to fully appreciate Bluey, but that didn’t stop Julia and me from watching the televised Bluey stage play or, more accurately, stage puppet show. It’s easy to imagine a version of this show with actors wearing full body character costumes, but it would be nowhere close to as magical as this one. Nor did they have to release an orchestral album, but they did. Despite being massively (and deservedly) successful, the show keeps its idiosyncratic charm. Which continues its success. I think of it as playing to the highest common denominator.

Speaking of highest common denominator: The entity I most associate with this phrase is Apple. You’ve likely seen all the coverage making the rounds for the company’s 50th anniversary, and this oral history on Apple’s early days (I love oral histories!) is one of the standouts.

Speaking of early computing: this Version History episode about the vocoder is 👌.

Before starting my new job, the architecture/urbanism account @wrathofgnon was one of the few reasons I would still check Twitter. Their thread on good urban boundaries as illustrated by the movie Zootopia is a classic in the form.

It is not hard to compare Zootopia (2106) with Thomas Moran's approach to Venice (1887). pic.twitter.com/sMiTLs3Cnv

— Wrath Of Gnon (@wrathofgnon) July 24, 2017

Wrath of Gnon had, however, far more scathing words for Zootopia 2 (which Julia and I also watched without Rufus).

To end: I found myself driving, the other week, behind a van for a catering company called Edibles Rex. It reminded me of the year 2014, when, having just moved back to Detroit, I espied, next to me on the freeway, a service truck belonging to an HVAC company named Desert in Alaska.

Poetry is everywhere,
Jack

Unlock this post by becoming a patron

This week I joined Every as a senior editor. I’ve been freelancing part-time there for over a year, editing essays (and in a couple of cases, fiction), and editing will continue to make up a majority of my work. I’ll also be writing my own occasional pieces like the two I shared in issues #454 and #452, and (and!) doing some light product work (which I’ll get to in a sec).

This all started a few years back when a mutual friend (👋 Linda) connected me with Dan Shipper, Every’s co-founder, to coach him on a manuscript. From there I picked up some freelance editing assignments, under the direction of editor-in-chief Kate Lee (formerly of Stripe Press and Medium), and from there, the energy of Every’s global 20+ person team charmed me even further.

One of Every’s big bets is that writers will thrive as AI becomes more enmeshed with our lives – that writing will be a foundational skill if not the foundational skill. Many engineers I know aren’t writing code anymore but they are writing prompts. And good writing gets harder when words are cheap.

These tools are radically changing software. It’s hard to understate this; I totally share Craig’s enthusiasm. If you don’t work in the field, you might’ve only had AI show up in your daily-use software in its sparkle-emoji forms: summaries, autocompletions, and the like, that it the biggest platforms and software makers are trying to shove down our throats.

But I suspect everyone will soon start to feel of the effects of how software is now being made. I’m most excited about what it means for the smaller projects that would otherwise get deprioritized. I’m excited about home-cooked apps and barefoot developers working in local communities, about the potential of long-abandoned projects rising from the dead.

I built Bebop on the first wave of AI copilot tools, and the latest models and coding agents make it feasible for me to maintain and actively develop it in my spare time. I finally, after almost a year of inactivity, pushed several new updates because the models are good enough that I can dip in between work on bigger, higher-priority projects (ie. manuscripts) and watching a one-and-a-half-year-old. Sometimes while watching a one-and-a-half-year-old.

As I said at the top, part of my role at Every will include light product work, mainly on tools to improve the editorial workflow. Again, this would otherwise would be its own full-time job if the latest Opus and GPT models weren’t so damn good at code. Simply being able to ask a model, “What does this code do?” and get a near-immediate answer in plain english ... it’s hard to go back.

While sometimes I find myself in rooms where I’m the most optimistic about AI, other times I’m in rooms where I’m the most skeptical. I’m worried about the second- and third-order effects on computer jobs. I’m appalled (but also not surprised) at the recklessness at which the big players are skirting regulation and gobbling up copyrighted material without just compensation (See You in the Cosmos is one of the books in Anthropic’s pirated training set). I’m unsure about whether or not the short-term environmental costs, the staggering amounts of energy and water usage, will really be made up by hypothetical, AI-enabled future advances in clean energy and medicine.

But I also see benefit in staying abreast of these tools and understanding how they work – and how the work on us. Like everyone (even those who claim to know where this is all headed), I’m figuring it out as I go.

What this means, more practically, is that I don’t expect much else to change here (gestures vaguely at desk and website). I’ll continue working on children’s novels in my spare time (as OG Sunday readers might recall, I wrote my first novel on nights and weekends, and my subsequent novels have taken just as long as that first one). I’ll continue to maintain and update Bebop, and host digital mending circles and books clubs. I’ll continue teaching, visiting schools and libraries to talk with students, gardening and working on house projects and starting new hobbies and trying to improve my daily routines and processes.

So why now? Why, after more than a dozen years ticking off the “self-employed” box on forms have I taken a full-time job? My rational justification is that I’ve come to appreciate stability and predictability much more since Rufus was born, and come to realize how much I depend on external structure, whether someone else has set up that structure or I’ve set it up for myself.

But the real answer is that it just felt right.

One last thought (and maybe small prediction): These tools make it newly possible for anyone to venture out on their own, it’s true. Or at least becoming true. You can have AI do all the parts of the job that you yourself don’t enjoy doing or don’t have time to do. It’s never been easier to go solo. Which means it’s also never been easier to feel alone.

So: find your crew. It doesn’t have to look like a company. It could look like an artist collective, a coworking group, an eclectic research lab, a telegram thread with fellow alumni of an unconventional architecture program, a monthly podcast recording meet that persists even after you and your co-hosts have put the podcast on hiatus, or any other configuration of multiple human beings who share ideas and tools and knowledge, cheer each other on, and help each other make sense of what’s going on in wider world.

Jack

Unlock this post by becoming a patron

Happy Lunar New Year, friends. I have a new essay out this month for Every about taste. I kept seeing this word thrown around in online AI discourse, and it seemed to mean a different thing every time I saw it. So this was me trying to parse the different definitions.

In the essay, I only touch briefly on the role of status in cultural taste, and that’s because David Marx already wrote a whole book on the subject—that I read as part of my research. It was my first introduction to Marx’s writing and I highly, highly recommend it as a follow-up.

As I pitch ideas for future pieces, I’m noticing that my favorite ideas tend to start with a question I want to explore, for which I do not have a clear answer. I might have a hunch about the answer, or a hunch about where I might find it. But the actual answer? Always more complicated and surprising than I expect. Part of the fun.

Two: Some of you new Sunday readers found me through that same taste essay—welcome! Once a month, on the second Tuesday of that month, I host a Digital Mending Circle, in which we take on

the kinds of oft-neglected maintenance tasks that accrue around our digital lives. Instead of darning socks and patching jeans, we update personal websites, delete unused accounts, work on side projects, or even just catch up on email.

During last week’s mending circle, I installed OpenClaw (I fretted just now over what site to link to here, because I definitely do not recommend installing it without knowing the security risks).

What does a suspect AI agent that runs off a Mac Mini and a Claude Code account have to do with maintenance? Well, these AI tools are starting to automate a lot of those same exact activities we do at our mending circles. So what I want to understand is: What does maintenance look like when you have an AI assistant running 24/7?

… which may be the question for next month’s essay.

Our next mending circle is Tuesday, March 10 from 7:30–9:00PM Eastern. Reply to this email to get the Zoom link.

Third: I was finally able, this weekend, to push out a round of planned updates to my iOS quick notes app, Bebop. Updates that I simply wouldn’t have/didn’t have time for prior to the latest AI coding models.

My preferred development environment of choice these days is Conductor, which legitimately feels like it serves new different mode of making software. The sidebar, instead of holding a list of files and folders like its predecessors, holds a list of projects/repositories, because when you set coding agents to work on a problem and they run for minutes or longer, you can plug away at other projects while you wait.

I have never been more scattered and productive at the same time. I used to think those two words were antonyms. Now, not so much.

Jack

Current events have me thinking, oddly enough, about plant sociability.

Sociability describes how aggressively plants spread. Some stay where right they are, others creep a little, or stay in moderate but contained patches, while the most aggressive will, if left alone, dominate the landscape.

An example might be Canada Goldenrod, which spreads both by seed and rhizome (horizontal arms that shoot out from the main stem). I first saw three, four plants show up in the alley behind our garage a few years ago. This past fall, practically the whole alley was swathed in its eponymous color. Even though the plant’s native to our area, it crowds out less aggressive ones, both native and non.

You can try to remove these plants, or deadhead them before they seed. These strategies tend to work better when there are still relatively few plants in the area. What do you do if a highly social plant has already taken hold? You might introduce other highly social plants that can hold their own against solidago canadensis. You plant ironweed and bee balm, milkweed and Joe Pye.

You meet aggression with aggression. Because that’s what the situation demands.

Mark my words: Pluribus is going to be remembered as the series that best captures this mid-2020s moment. I’m talking not just our current flavor of sycophantic AI chatbots, but also the memory of pandemic alignment, of what we can do if we all work toward a common goal.

If you want to see a society’s subconscious, all you need do is look at their science fiction. What I see here is both a fear of and admiration for collectivism, a fear of and admiration for – intentional or not – an ascendant China. That tension is what makes the first season of this show so interesting. And so American.

Last month I mentioned I was playing the JRPG-inspired game Sea of Stars. I finished that (and enjoyed it!) and moved onto Final Fantasy X, which I finally played after reading Aidan Moher’s history of the genre, the brilliantly named Fight, Magic, Items.

From that book I went right into Matt Alt’s broader-scoped Pure Invention, which opens with a scene from Final Fantasy VII and makes a compelling case that a lot of what’s happening in American society right now happened in Japan decades ago.

Next up bookwise: W. David Marx’s Ametora. And gamewise, send me a JRPG rec, if you have one (preferably one that won’t take 40+ hours to finish).

Jack

Unlock this post by becoming a patron