Progressive summarization, coined by Tiago Forte, solves a specific problem: you saved a long article months ago, and now you have to reread the whole thing to remember why it mattered. The fix is to compress notes in layers, over time, so your future self can grasp them in seconds.
The layers
- Layer 1 — the source. Save the note in full.
- Layer 2 — bold the passages that carry the point.
- Layer 3 — highlight the best of the bold (a much smaller set).
- Layer 4 — a summary in your own words, only for the notes that earn it.
- Layer 5 — remix, when you actually use the idea in your own work.
Crucially, you don't do all layers at once. You add a layer only when a note resurfaces and proves useful again — so effort flows to the notes that keep earning it, and dead notes stay at layer 1, untouched.
Why it works
It's designed around retrieval, the step this wiki argues most systems neglect (the whole point of PKM). A layered note is scannable at a glance: bold and highlights let you grasp a page in seconds without rereading it — fighting the forgetting curve on the reference side rather than the memorising side. And writing the layer-4 summary in your own words is exactly the understanding-building move that copying skips.
The trap: over-highlighting
The famous failure mode is bolding everything. If half the note is bold, the bold means nothing — you've just recreated the wall of text you were trying to escape. The discipline is aggressive selection: each layer should be a small fraction of the one below it. Compression is the value; if you're not throwing most of it away, you're not summarizing.
When it pays off — and when to skip it
Progressive summarization shines for people who collect a lot to reuse later: writers, researchers, anyone building a second brain from articles and books. It's wasted effort on fleeting notes and quick captures — those want atomic notes or just a good inbox, not five layers.
It's also, frankly, manual. The whole technique is a workaround for tools that can't surface the relevant part of a note for you — which is precisely what AI-across-your-notes is starting to automate: ask, and the one useful sentence comes back without a single highlight.