Bratha

Part 4.1 — Progressive Summarisation

7 min read

This is Part 4 of 7 in the Productivity Enhancement Series

Table of Contents

The problem with a saved note

PARA told you where to put a note. This article is about making the note worth keeping. Because here’s the trap: you highlight a whole article, save it, feel productive, and then six months later you reopen a wall of yellow text that’s as much work to read as the original. A note you have to re-read in full has saved you nothing. Progressive Summarisation1 is the technique that turns a raw capture into something you can grasp in seconds. And for this blog, it isn’t a separate practice; it’s the maintenance routine for the maps you already build in the Learning series.

Progressive Summarisation, the original technique

The original technique is layers of emphasis, added over time, only to the parts that prove valuable:

  1. Layer 1 — the captured note. The raw thing you saved: the article, the highlights, the meeting notes.
  2. Layer 2 — bold the best. On a later pass (only when you actually open it for a reason), bold the few sentences that carry the point.
  3. Layer 3 — highlight the best of the bold. On a still-later pass, highlight the handful of bolded phrases that are the real core.
  4. Layer 4 — a one-line summary in your own words, at the top, if the note earns it.

The crucial discipline: you only add a layer when you're already in the note for a real reason. You don’t summarise everything up front; you let use decide what deserves the effort. Notes you never reopen never get summarised, and that’s correct, because they didn’t matter. The valuable notes get compressed precisely because you keep returning to them.

The reframe: your map is the summary

Here’s where this blog’s version diverges from the original, and it’s the useful part.

In the Learning series you don’t take linear notes; you build maps: a structured picture of a domain, its key ideas, and how they connect. A map is already a summary. It’s Layer 4 by construction: the act of mapping a topic is the act of distilling it into its core concepts and relationships. So you don’t run Progressive Summarisation on a pile of highlights and then build a map. The map is where the summary lives.

Which changes what “progressive” means here. It’s not “add bold, then highlight, then a summary line.” It’s: every time you learn something new about a topic, you go back and make the map sharper. Progressive Summarisation becomes progressive map-refinement:

  • You read something new on the topic. Instead of saving it as a separate orphan note, you ask: what does this change about my map?
  • You update the map: add the new node, correct an old one, sharpen a definition, redraw a connection.
  • You link the source back to the map as supporting context, so the map stays the centre and the raw material hangs off it, retrievable but not in the way.

The map is the living summary; new inputs don’t pile up beside it, they get metabolised into it. The summary doesn't grow longer as you learn more; it grows sharper.

This is the antidote to the orphan-note swamp

The reason most knowledge bases rot is orphan notes: hundreds of disconnected captures, none linked, none summarised, none findable. Map-refinement structurally prevents it, because every new input has one job: improve an existing map or start a new one. There is no “save it somewhere and hope” path. Every input either sharpens the map or it isn't kept.

The loop: each pass sharpens the map

In practice it’s a small, repeatable loop, and it rhymes with the self-improving workflow from Part 3:

  1. Encounter new material on a topic you have a map for.
  2. Integrate: update the map with what’s genuinely new, in your own words.
  3. Link: connect the source to the map node it supports.
  4. Prune: if the new understanding makes an old node wrong or redundant, fix or cut it.

Notice the parallel. In Part 3.1, corrections get written back into your instructions so the workflow sharpens. Here, learnings get written back into your map so your understanding sharpens. Same shape, different artifact: write the improvement back into the durable thing, don't leave it loose.

The AI multiplier on map-refinement

Hand an assistant a new article and your existing map and ask: what here is genuinely new versus what I already have, and where would it attach? It does the tedious diff between fresh material and your current understanding, proposing the specific update. You still make the call on what’s true and what matters (it proposes, you decide), but the work of comparing the new against the known is drafted for you. Combined with RAG (Part 5.0), you can also ask your whole map collection a question and get an answer synthesised across them.

Why this beats re-reading

The deep reason to do any of this is in the Learning series: you don’t understand something by re-reading it, you understand it by retrieving and reconstructing it. Every time you update a map in your own words, you’re doing exactly that: pulling the idea up, restating it, and connecting it. Map-refinement isn't just storage hygiene; it's a learning rep disguised as note-keeping.

So the payoff is double. You get a knowledge base that’s actually retrievable (summaries, not walls of highlight), and the act of maintaining it deepens the understanding itself. Re-reading gives you neither: it feels like learning, but it’s the Mindlessness enemy wearing a study costume. With the map sharp and the material organised, the last move in the knowledge layer is to make it act, not just inform, which is Part 4.2.

Part 4 Takeaways

  • A note you must re-read in full has saved you nothing. The knowledge layer needs summaries, not hoards.
  • Progressive Summarisation adds layers of emphasis only to notes you actually reopen; use decides what’s worth compressing.
  • This blog’s reframe: ==your learning map is the summary.== You don’t summarise then map; the map is the distillation.
  • “Progressive” here means progressive map-refinement: every new input either sharpens an existing map or starts one, and links back as context.
  • The summary grows sharper, not longer, as you learn more. This structurally prevents the orphan-note swamp.
  • Same shape as the self-improving workflow: write the improvement back into the durable artifact. And it doubles as a learning rep, unlike re-reading.

Your Summarisation Task List

This week

  • Pick one topic you keep saving articles about. If it has no map yet, build a rough one (see the Learning series).
  • Take your three most recent saves on that topic and metabolise them into the map instead of leaving them as separate notes.
  • Link each of those sources back to the map node it supports.
  • Resist summarising notes you never reopen. Let use decide.
  • Next time you read something good, make it earn its keep: what does it change about a map you already have?

Sources & references

Footnotes

  1. Tiago Forte, “Progressive Summarization: A Practical Technique for Designing Discoverable Notes” (Forte Labs). The layered-emphasis method (capture → bold → highlight → summary, added only on revisit) is kept and adapted here. The departure is treating the Learning-series map as the summary artifact, so the technique becomes progressive map-refinement rather than note-highlighting. The learning rationale (retrieval and reconstruction beat re-reading) is developed in the Learning series. ↩

Graph View

Backlinks