File
SMF·PINBOARD
Received on
31.07.2026
Reviewed on
28.09.2026
Exhibits annexed
3
Questions
4

SMF·PINBOARD

Can a prompt replace Pinboard?

Read it later & RSS — read-later, bookmarks and RSS

Yes Verdict recorded on 28.09.2026 · Verified on 31.07.2026
Price
$1.83/moSource: pinboard.in · Checked on July 31, 2026
Per year
$21.96
Build time
One sitting
Votes
0 votes
Yes (checked)AlmostNot yet

Exhibit tracking slip

Exhibit A The prompt
Exhibit B What you lose
Exhibit C Why people still pay: capture polish, sync, and content partnerships
Exhibit Q Questions

Verdict

Pinboard is deliberately the smallest useful thing: a URL, a title, some tags, a note, and a search box. Rebuilding it is genuinely one sitting, and the archive option — fetching and storing a copy of each page so a dead link still has content behind it — is another hour. The one thing worth noticing is that Pinboard's appeal was never the features; it was that it kept working, unchanged, for over a decade while every competitor pivoted or shut down.

Exhibit B — What you lose

Exhibit A — The prompt

Received on31.07.2026
Build a self-hosted bookmarking service optimised for speed of saving and speed of finding. No feeds, no reading queue, no unread counts.

Saving: a bookmarklet and a small extension-free save page that takes a URL, pre-fills the title from the page, and accepts space-separated tags plus an optional note. The save form must be usable entirely from the keyboard and submit in one keystroke. Tag input offers completions from tags already used, ordered by frequency.

Archiving: on save, fetch the page with a descriptive user agent, honour robots.txt, run Readability over it, and store the extracted text plus the original HTML. If the fetch fails, save the bookmark anyway and mark the archive as unavailable — never block a save on a network call. Re-archive on demand. Never attempt to get past a paywall or a login wall.

Finding: a single search box that queries titles, notes, tags and archived body text through PostgreSQL full-text search, with results ranked by recency within relevance. Tag pages, tag intersection (two or more tags at once), and an untagged view. Support a private flag per bookmark and a to-read flag, but no separate reading interface — a to-read bookmark is just a filter.

Bulk operations, which is where a bookmark archive actually gets maintained: select many bookmarks and rename a tag across all of them, merge two tags, or delete. A tag-rename that touches 500 bookmarks must be one transaction.

Link rot: a scheduled job that HEAD-requests every bookmark on a slow rotation and flags the dead ones, with the archived copy still readable. Show a dead-link count on the dashboard.

Import and export: Netscape bookmarks HTML in both directions, plus a JSON export that includes the archived text, because an export that loses the archive is not a backup.

Out of scope: RSS subscriptions, a social or public feed, mobile apps, and any recommendation layer.

Opening prefills the prompt — press enter to run it.

Exhibit B — What you lose

  • B.1 the public bookmark stream and the social layer around it
  • B.2 the API that a decade of third-party clients were built against
  • B.3 somebody else guaranteeing the archive outlives your own server
  • B.4 automatic import of everything you favourite on other services

Prior art

Exhibit C — Why people still pay: capture polish, sync, and content partnerships

Because a bookmark archive is only worth something if it is still there in ten years, and paying a small annual fee to a service that has already survived that long is a reasonable bet against your own future attention.

Questions

Can I import my Pinboard bookmarks?

Yes, completely. Pinboard exports JSON including tags, notes, timestamps and the private and to-read flags, and all of it maps directly. Archived page copies are not in the export, so the importer re-fetches them, which means anything already dead comes across as a bare link.

How much disk does the archive need?

The extracted text is trivial, a few kilobytes per page. Keeping the original HTML as well pushes it to a few hundred kilobytes each, so ten thousand bookmarks is on the order of a couple of gigabytes. Storing images too would multiply that, which is why the prompt does not.

What does it cost to run?

A small VPS with PostgreSQL, five to ten dollars a month, against Pinboard's annual fee of roughly the same total. The economics only favour self-hosting if the server is already there for something else.

What is the one thing that does not survive the rebuild?

Longevity you do not have to think about. Pinboard's pitch was that it would still be running in ten years, and it was right. Your own server is only as durable as your own attention to it, and a bookmark archive is exactly the kind of thing you stop maintaining and then miss.

Receipt

Already built this yourself?