Paying rent on my own highlights

Six years ago I sat in a quarterly review at a logistics company where the head of strategy opened his remarks by quoting a business book from memory. Not the gist; the sentence, whole, with the clause structure the author had chosen. It was a good sentence and it landed, and I remember thinking the man had done the work most executives only fake. Ten minutes later he recommended the decision the same chapter had spent eleven pages dismantling, with identical confidence, and the room approved it, and so did I, because the quotation had spent my scepticism before I counted it. In the elevator afterwards he showed me the app on his phone: four thousand highlights, sorted by book, synced from his Kindle overnight. He paid for a service that emailed him a handful of them each morning. He had opened that email every day for two years. He told me this the way a man tells you about his pension.

I've thought about the elevator more than the meeting. Reading a book well and being changed by it are different activities, and only one of them has an app. What he had built was a facility for contact with ideas, which is real and worth something. What the meeting needed was the unbillable work of checking whether the idea still held, which the quotation had let him skip. Readwise is the best product on the market for the first activity. The pitch claims it serves the second. The product is a memory tool. The promise is a wisdom tool. Those are not the same business, and the gap between them is where the interesting part lives.

The email is a memory product wearing a reading product's clothes

What the service actually does is unglamorous. It connects to your Kindle, your phone's reader, whatever else you mark text in, and it pulls out the passages you underlined. Then it sends a rotating sample back to you on a schedule that stretches as a passage survives review, the technique psychologists call spaced repetition. The lineage here is honest. Hermann Ebbinghaus ran the original forgetting experiments on himself in the 1880s, self-funded, a sample of one, and his curve has replicated often enough that nobody serious disputes it. The modern version is retrieval practice: Roediger and Karpicke, working in a publicly funded lab at Washington University, showed in 2006 that testing yourself beats rereading for retention (their paper is here). The effect in the lab is large. The effect on whether you can rebuild an argument, rather than recognize its best sentence, is smaller and mostly unstudied, and the companies selling the technique have no reason to fund that particular study.

There's a deeper problem with the raw material. A highlight is a friction mark made at reading speed, which is not the same as a judgment about importance. You underline what startles you, and you're startled by what flatters the person you already are. The daily email rehearses your reading taste, not your reading. It produces the feeling of an educated mind at a cost of ninety seconds, and feelings of that kind are competitively priced, which is why they sell so well.

The archive is somebody's balance sheet

Consider the incentive. Readwise is bootstrapped, two founders, no venture money, paid entirely by subscribers, which is the cleanest funding structure in this category and I want to be fair about it. The pricing is two-tier: a full plan at $9.99 a month billed annually, $12.99 month to month, and a cut-down Lite tier at $5.59 billed annually. The checkout page tells you your card will be charged $119.88 for twelve months, which is the honest sentence. Early subscribers are promised their price locked for life, a retention mechanism wearing the costume of a favour, and it works on me, because I have one and I know what it is.

Now the structure of the thing. The Kindle sync runs on Amazon's sufferance. Amazon caps how much of a book you can pull, roughly ten percent of the text, a limit publishers control, and it could withdraw the pipe tomorrow. On top of that sits Reader, the read-later app bundled into the full plan, which is a good product and a second queue. When Mozilla shut Pocket down in July 2025, Reader was the loudest beneficiary, and a decade of read-later lists moved house into the same account that holds the memory of your books. That's the interesting part: nothing here is a moat until the archive grows, and then every year of highlights becomes a cancellation tax you pay by leaving. Reader even ships an AI assistant that will summarize what you haven't read, so the machine offers to complete the process of not-reading for you, in both directions. The export tools are real and freely advertised, which looks like generosity and is actually good sense. An export nobody trusts is a refund nobody has to give.

The case for paying to forget less

The counterargument deserves better than polite gestures, so here it is at full strength. A serious reader puts a hundred hours a year into books, and loses a meaningful share of them to time. The full plan costs about the price of two paperbacks a month; if it returns one usable sentence a week, an essay, a deposition, a lecture, it has beaten the alternative, which is buying the book again or shrugging. The founders' incentives are as clean as this industry gets. No advertisers to please, no board demanding engagement curves, subscribers only. And the lock-in is soft: I've pulled my own copy of the archive out through their export, and it works. Against a category full of companies monetizing your attention, here is one monetizing your memory, which is a better deal than most of the alternatives I've reviewed.

I also can't pretend to be outside this. I use the thing, and my incentives aren't clean. The daily email feeds me quotable material for essays like the one you're reading. I'm the man in the elevator, with a different app and better vocabulary.

The streak is the product

It depends, and the dependency is whether the email functions as rehearsal or as wallpaper. For people whose work product is a sentence, writers, lawyers, teachers, it's cheap, honestly built, and does what the retrieval research says retrieval practice does. For everyone else, watch one variable: whether you open the daily email or archive it. The day you archive it unread, cancel, because the archive has stopped working and started decorating. The man in the elevator still opens his, every morning, four thousand highlights deep. I've never decided whether that's discipline or tribute.