How to organise thousands of bookmarks

Bookmark archives don't fail gradually; they fail all at once. The system that worked at 200 links breaks at 2,000 and is unsalvageable at 20,000. A few habits applied early prevent that.

Don't use folders as the primary structure

Folders impose a strict tree. Information rarely fits one. A piece on AI policy is both "AI" and "policy"; a tutorial on Postgres is both "databases" and "Rust" if it covers an SDK. With folders you pick one, lose the others, and eventually forget where you put anything.

Use topics (sometimes called tags or labels) instead. A bookmark can belong to many topics. Topics can themselves be organised in a loose hierarchy where useful, but the primary axis is "this bookmark is about x and y."

One topic per bookmark, minimum

The single most useful habit at scale: tag on the way in. Future-you won't go back and retroactively organise 800 untagged links. Even a casual topic — "maybe relevant to my Rust project" — gives the entry an anchor in the graph.

Collections separate worlds

Once an archive grows past a thousand entries, it becomes useful to split it into collections — entirely separate stores for fundamentally different contexts. Work and personal. Research and hobbies. Active and archived. Cross-collection topics blur the distinction; per-collection topics keep each world clean.

Use search, not browsing

At thousands of entries, scrolling is a non-strategy. Invest the time to learn your tool's search:

• Title and description full-text search.
• URL substring ("everything on github.com/openai").
• Filter by topic (or topic intersection: "ML and ethics").
• Filter by date — "what did I save in the months before I started that project?"

If your tool only supports linear browsing, it doesn't scale. Pick one that does.

Periodic clean-ups

Three habits that take an hour each and pay off forever:

1. Duplicate sweep. Every six months, run a duplicate check. Different URLs, different sessions, same article — merge or delete.
2. Broken-link sweep. Same cadence. Anything 404 or domain-dead gets either replaced with an archive.org link or removed.
3. Topic prune. Look at your topics list once a quarter. Merge near-synonyms, retire dead topics, split overloaded ones (a topic with 500 bookmarks is doing too much work).

Visualise the shape of the archive

When numbers are abstract — "6,200 bookmarks across 180 topics" — it's hard to know whether the archive is healthy or rotting. Visualisations make the shape concrete:

• A topic sunburst shows what proportion of your archive each topic accounts for.
• A 3-year activity heatmap reveals saving rhythms and dry spells.
• A domain bubble pack exposes which sites dominate your reading (often surprising).
• A topic-share-over-time stacked area shows interests rising and falling — and helps you notice when you haven't saved anything on a topic you used to care about.

Own the data layer

Whatever tool you settle on, prefer one whose storage format is open and portable. A thousand-bookmark archive in a vendor-proprietary schema is tied to that vendor's product roadmap. The same archive in W3C linked data — in a Solid pod you own — survives any tool change.