Retrieval first: notes you can find, links you can explain, sources you can recover.

Sortjunglecompassrn

Guide · References

Keep bookmarks and citations retrievable

Saving a link feels like keeping the source. Often it only keeps an address. This guide covers the few details that let you recover a source years later, even after the page has moved.

Start with one retrieval problem
Close-up of book spines with paper bookmark slips protruding from the pages
A bookmark points; a reference identifies.

Why bookmarks fail

Web pages are less stable than they appear. Sites are redesigned, articles move behind paywalls, organisations close, and URLs change without redirects. Researchers call this link rot. A related problem, content drift, is worse because it is silent: the link still works, but the page now says something different from what you saw.

The response is not to distrust the web but to record enough detail that a missing page is an inconvenience rather than a dead end.

Identifiers outlast addresses

Many sources carry a persistent identifier designed to survive changes of location:

DOI (Digital Object Identifier), standardised as ISO 26324, is assigned to most journal articles and many datasets and reports. A DOI resolves through doi.org to the current location of the item.

ISBN identifies a specific edition of a book. Record the edition, because page numbers differ between editions.

Archived copies. The Internet Archive's Wayback Machine lets you save a snapshot of a public page and records the capture date. If a page matters to your argument, saving a snapshot when you first cite it is a small, sensible habit.

Bordered reference rows: what to record

The table below is a minimum, not a maximum. Its fields loosely follow the Dublin Core idea that a few well-chosen descriptors (creator, title, date, identifier) go a long way.

SourceMinimum fieldsBackup
Journal articleAuthors, year, title, journal, DOI, page of key passagePDF in reference manager
BookAuthor, title, edition, publisher, year, ISBN, chapter and pagePhoto of quoted page
Web articleAuthor or organisation, title, URL, published date, accessed dateArchived snapshot URL
Video or podcastCreator, title, platform, date, timestamp of the momentTranscript excerpt
Internal documentOwner, title, version, date, storage locationExported copy, if permitted

Reference managers and citation keys

If you cite sources regularly, a reference manager saves real time. Zotero, a free and open-source tool originally developed at George Mason University's Roy Rosenzweig Center for History and New Media, can capture metadata from a browser, store PDFs and export to formats such as BibTeX. Whatever tool you use, give each source a citation key that is predictable: author surname plus year plus a short word, such as ahrens2017smart. Use the same key in your notes, and every note becomes traceable to its source in one search.

Structuring bookmarks without a folder maze

Deeply nested bookmark folders tend to collapse under their own weight. A flatter approach works better for retrieval:

Before
Bookmarks/Work/Research/2024/Misc/
  Useful/Read later/Article (3)
After
Bookmarks/
  Active project — onboarding study
  Reference — methods
  To read (cleared monthly)
Title edited: "Onboarding survey method — Example Org, 2025"

Edit the bookmark title so it contains the words you will search for. Browser bookmark search matches titles and URLs, so a descriptive title does most of the work a folder was trying to do.

Realistic limits

You will not archive everything, and you do not need to. Save the full detail for sources you quote, rely on for a decision or expect to revisit. For casual reading, a descriptive title and an access date are enough.

Reference reliability checklist
Next: sketch ideas into searchable structures