AI Counter-Trends Have Receipts Now
Hey Everyone - Hope you had a great weekend. The "analog revival" story has been bubbling for two years, and most of the takes treat it as a Gen X / boomer nostalgia trip. The numbers that came in last week make that read look wrong. The revival is real, it is measurable, and it is young. The hedge inside it is not the one most readers would guess. Let me walk you through it.
This week:
The Signal - The analog revival gets hard numbers, and the opportunity inside it
What I'm building - The up-front investment thesis, and why the value of AI lives in the system, not the model
Resources - The AI layoff trap, the HBR "brain fry" study, the adaptive-capacity gap, LeCun on xAI, and Goodhart eats AI benchmarks
Skills to Develop - Host a Recurring Thing, the durable role most people opt out of
Let's dive in.
This week’s Signal
🌎 AI Counter-Trends Have Receipts Now
For two years the "analog revival" has been an aesthetic, not a market. Vinyl on Spotify subscribers' shelves. The occasional vintage camera in a coffee shop. Easy to dismiss as a boomer nostalgia trip or a TikTok aesthetic. The numbers from the last two weeks make that read look wrong.
Wholesale film order volume is up 127 percent from 2020 to 2026, and more than 300 new film photography labs opened globally in 2025 alone. 65 percent of surveyed film photographers are under 45, with the sharpest spike in the 25 to 34 bracket. Eventbrite's first-ever "Reset to Real" Social Study found over 600,000 people attended block parties in the US, local art walks grew 49 percent year over year, and 89 percent of younger adults specifically want events that connect them to their local community. None of this is nostalgia. The people building it grew up online.
The reframe the reporting lands on is what I want you to carry. The revival is a "renegotiation of technology's terms." It is not a rejection of the digital, and it is not a return to the past. It is a deliberate choice, by a young cohort, to put restraint, scarcity, and in-person ritual back into the value equation. The signal of quality is no longer "more capability." The signal of quality is "human fit."
Here is the part most coverage misses. The opportunity inside the analog revival is not selling film, vinyl, or board games. The margin on the object is thin and getting thinner. The opportunity is hosting the ritual that the object is the excuse for. The film walk that brings 40 people to a Saturday morning. The listening night at someone's living room. The board game evening that the same six people show up to every other week. The art walk that 1,500 people attend because someone organizes it.
Hosting is the durable role. The object goes through Amazon. The ritual does not. AI can write a newsletter inviting people to the event. It cannot be in the room when the event happens. It cannot remember who brought the snacks last time. It cannot make eye contact with the new person at the door. Every part of the ritual that gives it value lives in the body of the person hosting it, which is exactly why this market is going to keep growing even when the underlying product gets cheaper.
The hedge translation for the reader is simple. Stop trying to sell the new generation a product. Start hosting the small recurring thing that the product is an excuse for. I will spell out the practice in the Survival Skill section below. The shorthand is that the analog revival is going to make a lot of people money, but the durable position is the one in the middle of the room, not the one selling at the table.
What I’m Building
the up-front investment thesis
The trade I keep coming back to in my own work is to do a lot of unglamorous up-front investment now, in exchange for compounding value later. Two specific bets are eating my time and capital this quarter, and the logic is the same on both.
The first is audience. I am spending more on growth than the spreadsheet would defend in any single-month payback model, because the value of an engaged audience compounds faster than the value of almost any other asset I can buy. The cost is real. The payoff lives in the years, not the quarters.
The second is AI systems. I have been spending a lot of time building actual systems with AI, not just prompting it. The realization underneath that work is that the value of AI in 2026 does not come from the raw capability of the model. It comes from how well I design the systems and planning around it. The reader who is fastest with a model is going to lose to the reader who is best at thinking about where the model fits. It is a creativity problem, not a capability problem, and the only way to get good at it is to do the up-front work of building, breaking, and rebuilding the workflows.
The meta-lesson is that the cost of being early in something usually looks worse than it pays. The work I am doing now is expensive. It is also the only path to the version of my business that I want to be running in 2027.
What I’m Learning
lot’s of stuff
arXiv: "The AI Layoff Trap" - Tech layoffs are running at 1,115 workers per working day in 2026 with AI as the headline justification almost everywhere. The uncomfortable finding is that the cuts are not improving returns. Many firms are cutting to signal "AI-forward" to the market, not because automation replaced the work. The hedge translation is to read your own employer carefully. Cutting because AI works and cutting because cutting says AI works require very different survival strategies.
HBR: "When Using AI Leads to Brain Fry" - 1,500-worker study finds heavy AI use produces acute cognitive fatigue, worst when juggling several AI systems at once. Workers describe mental fog, headaches, slower decision-making. This is a faster failure mode than the long-term offloading data, and it argues for bounded, single-purpose use over the autonomous-everything dashboard.
NBER: adaptive capacity to AI displacement - The roles AI automates most aggressively (around 6.1 million US clerical and administrative workers, disproportionately women) also have the lowest capacity to move into something else. The career-wall story has a second part nobody talks about: the exits are narrow too. "Learn to adapt" is empty advice unless you name what adaptation actually requires.
Tech Startups: Yann LeCun calls xAI "kind of a failure," warns the industry is overspending - When one of the field's most credentialed researchers publicly questions whether the spending is sustainable, that disagreement is itself the signal. The hedge is not picking a side. The hedge is being the kind of person whose income, skills, and community hold up either way.
InfoFina: Goodhart's Law eats AI benchmarks - A growing cluster of analysts argue benchmark scores have decoupled from real-world reliability because every lab trains against the same public tests. Coding assistants ace HumanEval and ship code that fails catastrophically in production. The deeper read for the reader is that the same dynamic breaks the KPIs your job is measured on. Anchor to outcomes you can verify yourself.
Survival Skill
Host a Recurring Thing
Pick one in-person ritual and commit to hosting it on a schedule. Movie night every other Tuesday. Monthly dinner on the first Sunday. A walk every Saturday at 8 a.m. A board game night every two weeks. The form does not matter. The cadence does.
The reason this is the skill that maps to the Signal is that the analog revival is going to be captured by the people who host, not the people who attend. Attendance is a consumer behavior, and it will compete with every other consumer behavior in your life for time. Hosting is a producer behavior, and it puts you on the durable side of the ritual.
Three things to keep in mind.
One. Keep the format dumb. Same time, same place, same handful of people. The whole value of a recurring thing is the recurrence. Optimize for showing up, not for scaling.
Two. Be the one who reminds. The host's actual job is the text on Friday that says "still on for Sunday." 80 percent of small gatherings die because nobody confirmed. Be the one who confirms.
Three. Eight weeks before you judge it. Most gatherings feel awkward for the first three or four. By the fifth, the people who are going to be in the room have surfaced. By the eighth, the ritual is load-bearing in everybody's calendar.
The chatbot economy can produce an infinite supply of content about how to connect with people. It cannot show up at your door on a Tuesday night. Hosting is the move that almost everybody opts out of and almost everybody benefits from. Be the one who opts in.
Three reflective questions
Which recurring in-person ritual in your life has somebody else hosting it, and what would change if you started one of your own?
Where in your work are you optimizing the prompt when you should be optimizing the system around the prompt?
If your employer cut headcount in the next quarter and cited AI, would the cut be because AI replaced the work, or because cutting signals AI-forward to the market?
Weekly AI Prompt
Act as a community-building coach. I am going to describe my current
social calendar, the people I see regularly, and the rituals (or lack
of them) that anchor my week.
Your job is to help me design one new recurring in-person ritual I
could host, starting in the next two weeks.
Walk me through:
1. Given the people in my life and my schedule, what is the cadence
that would actually work? Weekly, every-other-week, monthly?
2. What is the format with the lowest activation energy that still
produces real connection? (Be specific: meal, walk, game, screening.)
3. Who are the three to six people I should invite for the first one,
and why those?
4. What is the single sentence I should send to invite them this
week?
5. What is the most likely reason the ritual dies by week four, and
what would prevent that?
Be direct. Do not give me a list of options. Give me one
recommendation and the message I should send today.
Here is my situation:
[paste]Until next week,
Ken