Footnotes Pointing at Air
The most troublesome thing about link rot is not that it leaves us one page less to read, but that the link is still asked to testify on behalf of a sentence. The footnote remains at the bottom of the page; the number, the URL, and that bit of blue are all still there. It looks like a reliable gesture, pointing the body text toward somewhere else. What has broken, though, is not a string of characters, but a sentence’s way of accounting for where it came from.
In a report from ten years ago, a certain number carries a footnote. At the bottom of the page, it says this was a government notice, or a statement someone once made in public. The body text is composed, as if by following the number downward, the factual support behind it would reveal itself. The reader moves the cursor over it, the pointer turns into a hand, and after the click the browser returns only a brief answer: not found.
That brief answer omits too much. The page may have moved, the server may be misconfigured, the content may have been withdrawn, the owner may no longer wish to make it public; what the reader sees is often the same kind of absence. But these absences are not the same. A move means a new address may still be found. A misconfiguration may still be fixed. Withdrawal may suggest the original words can no longer be trusted as before. Refusal to disclose is more troublesome still: it may involve privacy, security, or some pressure that makes public exposure inconvenient.
So I do not quite trust the impulse to turn every webpage into amber. Some pages should be let go; some old names are better off leaving search results. If a small site stops running, it need not arrange its own decline into an announcement. Preservation can also become detention, pinning a provisional remark or a photograph later regretted to a wall harder than the one it first appeared on.
The problem lies somewhere else: a piece of material is still being cited, corrected, and asked to explain its source, yet only silence remains. Later readers may only have wanted to confirm where a number or a quotation came from, but what they receive is not an answer, nor a refusal, but an unsigned piece of absence. Correction does not require old words to remain valid; it only requires those who come later to know that they have changed. Withdrawal need not lay out the content either, but it can at least leave a boundary: this sentence was here; it can no longer be used in the same way.
A link looks like an address, but in truth it exists temporarily through renewals, migrations, and redirects. Only when someone connects the old path to the new one does it appear never to have moved. We are used to treating openability as stability and failure to open as breakage; but stability is often only maintenance that has not yet shown itself. A page’s remaining in place does not mean it is inherently reliable. It is merely still being held up by certain people, certain systems, and certain bills that have not been cut off.
What is even more troublesome is not necessarily the 404. The fact that a page can open does not mean it can still testify. Some links preserve the appearance of a route while replacing what lies at the route’s end; the reader arrives smoothly, which makes it even harder to notice that they are no longer in the original place. Seen this way, a 404 can sometimes be more honest: at least it does not pretend that the relationship still holds.
The footnote remains at the bottom of the page, the blue still bright. What it points to is not blankness, but a change that has not been explained.