A Brief Observation on URL Structure and Citation Stability

Conversations about citation stability frequently focus on the form of the URL — its length, its readability, the absence of query strings — to a degree that is, in my view, mildly out of proportion to the underlying problem.

The form of a URL is a presentation question. The persistence of the underlying resource is a separate, more substantive question. A beautifully composed URL pointing to a document that has moved is no better than an ugly one pointing to the same document. The two are equivalent in their failure mode.

What stability requires

What makes a URL stable, in practice, is not its form but its custodianship: the willingness of the document's host to maintain the mapping from URL to content over time, including across the various operational changes (server migrations, software upgrades, scheme adjustments) that tend to interrupt that mapping by accident.

A small archive in the custody of a single attentive author has, in this respect, an advantage over institutional archives, which suffer from the diffusion of responsibility that accompanies institutional operation. The single author is the custodian, and the cost of maintaining the mapping is paid out of the same attention that produced the material in the first place.

Implications for URL form

If the form of the URL is essentially a presentation question, the choice between conventions can be made on grounds of legibility and operational simplicity. My own practice — opaque identifiers under a single uniform path — was discussed at greater length in an earlier note. The choice is not, I think, of much consequence. What matters is that the chosen scheme is followed without exception, and that the custodial discipline behind it is sustained.

A practical note

One small practice I have found useful: when an entry must be substantively revised, the original URL is preserved and the prior version remains accessible at a versioned suffix. This requires a small amount of additional discipline at the moment of revision, but it preserves the citation stability of the prior reference. The value exceeds the cost.