The eighth one went this afternoon. The newsletter was a perfectly competent weekly from a perfectly competent writer, formerly of Reforge, and the unsubscribe link was at the bottom of an email whose subject line I had to read twice before I understood it was, again, the loop diagram I had already seen four times this month from four different writers. The browser tab made the small confirmation sound. I closed the tab.
The thing that took me about three years to notice, and another year to act on, is that almost every product Substack I subscribed to between 2022 and 2025 is downstream of the same five or six writers, Marty Cagan and Lenny Rachitsky and Julie Zhuo and Shreyas Doshi, with Patrick McKenzie occasionally wandering in from the financial services flank. The next three hundred newsletters are a tighter remix of those five, the same diagrams redrawn in different colors, the same vocabulary compounding in a closed loop of writers reading each other and folding each other's coinages into their own next post. Read the canonical thirty essays once, and the next three hundred are statistical noise around the same mean.
René Girard wrote a whole book in 1961 about how desire is mimetic, by which he meant that humans almost never want a thing on their own; we want it because someone we admire wanted it first, and then we forget that we copied them. The genre of product writing is one of the cleaner mimetic loops I have lived inside. Each writer wants to be quoted by the next, and each new framework wants to compose neatly with the prior framework so it can be cited alongside it on a slide. The genre eats itself on a diet of fast carbohydrates, where every calorie comes from recognising a phrase you have already seen on someone else's quarterly board slide.
A phrase only survives that kind of circulation if it is safe, which is the less polite half of the problem and the half I took longest to say out loud. Most of the actual advice is unfalsifiable. Lines like talk to customers, have empathy for the user, and ship and learn are all true and all useless in the same way, which is that no PM has ever read one of them and known what to actually do on Tuesday morning. The essays that get the most claps are almost always the ones whose claims could not be wrong, because anything that could be wrong would lose half the room and the post would lose half its open rate. Patrick McKenzie made roughly this point in a long Bits About Money piece on why generic financial advice underperforms, in a different domain, but the shape is the same. Advice that survives at scale survives by being vague enough not to be wrong, which is exactly backwards from how knowledge in any actual craft is supposed to compound.
Which raises the question of who all this vagueness is actually being kept safe for, and the answer cuts against what most of us tell ourselves when we open the newsletter on a Sunday morning. Product Substacks are read mostly by other product people who would like to write product Substacks. The actual customers our products serve do not read Marty Cagan, and the buyer on a B2B sales call in Hamburg next Wednesday has never heard of intent versus action. The signal that should matter, whether your product solved a real problem for a real person who paid you for the solving, is not visible anywhere on the Monday morning open rate dashboard. The genre's incentive structure rewards the wrong loop, and the writers who optimise against it end up writing for the wrong audience without ever noticing the swap.
I want to be honest about an exception, because the honest version of this argument has one. I still occasionally read Lenny's interviews, not for Lenny but for the guests. When he sits across from Cat Wu or Stevie Case or some PM who actually shipped the thing the interview is about, the guest sometimes says the one sentence the host would not have known to write. The complaint I am making is not about any one writer. It is about the genre, and about what the genre's gravity does to the people inside it. Venkatesh Rao wrote in his Premium Mediocre essay in 2017 about a whole class of cultural products that look like the good version of the thing while being one rung below it on every axis that matters. The product Substack canon, after about 2022, has settled into being the premium mediocre version of product thinking, with the surface fluency of the real article and a quiet sag on every axis you actually care about by the time you have read your sixth post on growth loops in a quarter.
Naming the sag is easy enough. Saying what to read instead is the part I trusted last, because this is where I have to hand you a list and you have to decide whether to believe me that the trade was worth it. The thinking that fed my product instincts most in the year after I unsubscribed came from reading further from the canon rather than closer to it. In practice that meant Christopher Alexander on patterns and on how a building gets its room sizes right, and Stewart Brand on how buildings learn over the forty years after the architect has moved on. It also meant Naila Kabeer's work on the Bangladesh garment supply chain as an actually functioning industrial network, and Byrne Hobart's Capital Gains on the weeks when he is writing about industry structure rather than markets. These are the kind of books my own prior post on hiring for taste was pointing at without quite saying out loud. The Substack diet was the equivalent of an MBA's required reading list, where the curriculum has converged on a small set of high status texts and the graduate emerges fluent in the vocabulary of the field and confused about the field itself. Taste does not compound on that flat plateau of identical reading; it compounds on the longer, more inconvenient hike out of it that you mostly take alone.
The eighth unsubscribe is now sitting somewhere in the trash folder under three weeks of newsletter receipts. The inbox is at zero unread for the first time in roughly six months, and the active browser tab has moved across to the personal blog of a guest from last week's Lenny interview, where she has posted six paragraphs on the engineering culture problem the host did not think to ask her about. The sentence I am about to read next was written by a woman at her own desk in Brooklyn, in her own voice, with no Monday morning open rate dashboard pulling at the back of her sentences as she wrote them. That is roughly the trade I am quietly recommending. Eight Substacks gone, and the writer I am reading instead is one Marty Cagan has never quoted and Lenny has not yet booked.