∮project

Margin

An experiment in marginalia as a primary writing surface — what happens when the side-note is more important than the main text.

active since September 2025 tools for thoughtinterface aesthetics

Margin started from a small irritation: the most useful sentences I write about a paper rarely end up in the paper. They end up in the gutter, on a sticky note, or in a Telegram message to myself. Mostly the last one.※Or, embarrassingly often, in a comment-only Google Doc that no one but me ever reopens.

The project explores what happens when an interface treats those marginal sentences as the primary surface — and pushes the “main text” into the role of seed. Not a new note-taking app; closer to a reading prosthesis.

The working hypothesis, in one sentence.

A reader who annotates aggressively at the moment of reading remembers considerably more, and synthesizes considerably better, than a reader who makes neat notes afterwards — and the existing tools punish the former.

I’m pulling on three threads in parallel: how to capture margin notes without context-switching, how to retrieve them weeks later, and what to do when two notes about different texts turn out to be the same idea.

For the broader thinking that motivates this, see Tools for Thought and the running notes in Reading on paper vs. screen.

River

1. Oct 30, 2025 Gesture set v2 note Second draft of the gesture vocabulary for Margin, after the first prototype.
2. Oct 29, 2025 First prototype reads a paper end-to-end timeline-entry Used the prototype to read Karpicke & Blunt 2011 cover-to-cover. It survived. The gesture set didn't.
3. Oct 20, 2025 Notes vs. marks note Two different annotation primitives, often confused, with very different design implications.
4. Oct 12, 2025 A reader's stance post On reading as an active, aggressive thing — and why the tools we have mostly assume otherwise.
5. Oct 8, 2025 October planning meeting meeting A self-meeting to scope the next two weeks of Margin.
6. Oct 8, 2025 Are you the user? question When you build a tool for yourself, how do you tell whether the design generalizes — and does it matter?

Pointed at by

A reader's stance — “…the reader sit up straight. This is part of what Margin is trying to figure out. <Q>If aggressive readin…” read

Notes vs. marks — “…t, and network-connectable to other notes. For Margin I'm leaning toward two distinct primitives, and a…” read

October planning meeting — “…end</Task> <Task>Write a short retrospective in Margin</Task> </Tasks> ## Open A nagging worry, record…” read

Why I paused Reading Room — “…otel; you sleep well, but you don't think. Hence Margin. The next iteration of this work belongs there.…” read

Andy Matuschak — “…way, conserve it. ## How it shapes my projects Margin is downstream of the evergreen-note idea, but wit…” read

Reading on paper vs. screen — “…n-reading practice. ## What I take from this for Margin - Treat each "page" as a fixed unit, not a slice…” read

A reading practice that survives the week — “…*first* pass. The digital follow-up happens in Margin (or, before Margin ships, in a plain text f…” read

Reading Room — “…rn from it, and the next iteration belongs inside Margin rather than as its own thing. The earlier essays…” read