Readwise, Matter, and the Quiet Rise of the Read-It-Later Layer
On a Tuesday morning in February, a consultant named Raj Patel opens Readwise Reader on his phone during a train commute and finds 74 unread items waiting. A longform piece from The Atlantic he saved three weeks ago. Twelve newsletters. A PDF of a McKinsey report. Four research papers from Google Scholar. A Twitter thread someone sent him. By the time the train reaches his stop, he has skimmed eight items, highlighted four passages, and added a total of zero of those highlights to long-term memory. The highlights are now syncing to his notes app. He feels productive. He is not sure whether he is.
The read-it-later category has a longer history than most people realize. Marco Arment launched Instapaper in 2008 as a weekend project, sold it to Betaworks in 2013, bought it back in 2018 with partner David Eldeib, and still ships updates. Pocket, built by Nate Weiner as Read It Later in 2007, was renamed in 2012 and acquired by Mozilla in 2017. For more than a decade, the pitch was the same: save an article, strip the clutter, read it later on a device of your choosing. The apps were utilities, and they competed on small differences in typography, synchronization, and price.
Something changed around 2020. Matter launched in beta that year, built by a small team that included Sayan Sivanesan and Ben Springwater, and positioned itself as a reader for the kind of person who subscribes to nine newsletters and reads them on an iPad on Sunday mornings. Readwise, founded in 2017 by Daniel Doyon and Tristan Homsi as a highlight-syncing service for Kindle users, released Readwise Reader in 2023 and effectively declared war on the old category. The pitch was no longer “read articles later.” It was “build a personal knowledge base from everything you read.”
The distinction matters. Instapaper and Pocket treated the article as the unit. You saved it, you read it, you archived it, and the highlights mostly lived inside the app. Readwise and Matter treat the highlight as the unit. Every passage you mark flows through a syncing layer to Notion, Obsidian, Roam, or a plain Markdown folder. The article becomes raw material. The extraction is the product.
This is a meaningful shift in what the software thinks you’re doing. The old apps assumed reading was an experience to be completed. The new apps assume reading is an input to a long-running personal research project. For people who actually have such a project, academics, consultants, journalists, certain kinds of engineers, the shift is welcome. For everyone else, it risks encouraging the collection of highlights as a substitute for absorbing the ideas in them.
The substitution is the part worth scrutinizing. Readwise has a surface feature called the Daily Review that shows you a selection of old highlights each morning. The copy on the landing page suggests this will help you remember what you’ve read. The underlying mechanism is a kind of low-grade spaced repetition, though it lacks the active recall component that makes spaced repetition actually work. You see a highlight. You read it. You think, “yes, I remember that.” You move on. This is not retrieval practice. It’s re-exposure, which is closer to rereading than to testing yourself, and the cognitive-science literature has been consistent for years that re-exposure yields substantially weaker gains than active retrieval.
The cleaner version of the critique, made by the memory researcher Robert Bjork and echoed across the learning-science community, is that feeling of familiarity is not the same as knowing. Highlights produce familiarity efficiently. Skimming a daily review of highlights produces the comfortable sensation that your knowledge base is growing. The sensation does not map cleanly to retention. Anyone who has used these tools for more than a year can test this by trying to describe, without reference to notes, the argument of an article they highlighted eight months ago. The results are usually humbling.
None of which makes Readwise or Matter useless. It just reframes what they’re useful for. The serious case for the read-it-later layer is not memory. It’s retrieval. When you are writing something and you half-remember a study or a phrase from an essay, a searchable corpus of everything you’ve found worth marking is genuinely valuable. It’s a personal search engine over the subset of the web that has passed your own filter. For knowledge workers who write regularly, this is a real productivity gain, and the syncing layer to a note-taking system of your choice amplifies it.
The case gets stronger when the highlights are paired with actual thinking. Matter’s workflow gently encourages you to write a comment on a highlight before saving it. Readwise Reader supports inline notes that sync along with the passages. The people who get the most out of these tools treat the highlight as a prompt to articulate why the passage matters, what it connects to, what it contradicts. The articulation is what moves material from passive familiarity into something closer to understanding.
Patel, the consultant on the train, highlights aggressively and almost never writes the accompanying note. His Readwise library has 11,000 highlights collected over three years. He has searched it perhaps forty times. When he searches, the library is occasionally useful and more often overwhelming. The problem is not that Readwise is bad software. The software is very good. The problem is that collecting is not studying, and the feedback loop of highlighting rewards the collection habit more than the studying habit.
There is a version of this critique that applies to every tool in the personal knowledge management category. Notion vaults, Obsidian graphs, Anki decks: all of them can become archives of material that was encountered once and processed lightly. The difference with the read-it-later layer is that the encounter itself is already shallower than most, because the article is competing with seventy-three other articles in your queue. Attention per item is the scarce resource, and the tools have a structural incentive to grow the queue.
The honest advice, if you actually want to learn from what you read, is unfashionable and straightforward. Read fewer things. Read them more slowly. Close the tab and try to articulate the central claim from memory before you highlight anything. Treat the highlight as a bookmark to return to, not as an outcome. If you can’t explain the argument to a colleague a week later, the highlight did not do what you hoped. Pair reading with actual spaced repetition for the facts and frameworks you want to retain, rather than relying on the daily-review illusion.
Where does this leave Readwise and Matter? In the same place Instapaper and Pocket always were: as genuinely useful plumbing for people who have a clear sense of what they’re using them for. The new generation of apps is better plumbing. Reader’s PDF support is excellent. Matter’s newsletter handling is best-in-class. The syncing layer that connects highlights to your notes is a real improvement over what existed in 2015. These are not small wins.
The mistake is to let the tool substitute for the work. Highlights are not memory. Saving is not reading. A personal knowledge base is only as valuable as the thinking you put into it, and the thinking is not done by the software. It never was.
Photo via Unsplash.