What happens to a fact on this site when the source that proved it goes offline?
The sentence stays. A dead link does not make a sentence wrong, so the page states that the original is unreachable and names an archived copy and its capture date where one exists.
In brief
A page cited here returns a 404. Nothing about the world changed: the article was published, it was read, the fact happened. What broke is the receipt.
The sentence is as true as yesterday and merely harder to check.
That is different from a true sentence going stale, where the world moves and the sentence stops being accurate.
What happens to the sentence
When only the evidence is lost
The claim stands and its support is downgraded in public: the sentence keeps its place, the page says the original is unreachable and names the archived copy where one exists.
When the claim itself is lost
Only when re-checking shows the fact was wrong or has expired. That is a correction, and it takes the route in how a wrong fact gets corrected.
Dropping the sentence because the link broke removes something true for a reason unconnected to its truth, which is a true fact left out by accident.
Four ways a link dies
The address moved
Still published, at a new path, after a redesign or migration. The cheapest failure and the commonest. The link is repointed, with a note of when it changed.
The publisher is gone
The outlet closed, the domain expired, nothing was migrated. A web archive may hold a capture, and if it does, that capture is now the source, a weaker one.
The page is intact and unreachable
Behind a login, a paywall, or a platform that blocks crawlers. It exists, and no reader can confirm it: when the proof lives on somebody else’s site.
It was taken down on purpose
Retracted, or removed at somebody’s request. The disappearance is itself information. Reaching for an archived copy of something a publisher deliberately withdrew needs a better reason than convenience, especially when the piece names other people: when a page names somebody else.
An archived copy carries two dates
An archived capture is a different object from the article it copies. It shows what a page said on the day it was taken, not that the publisher still stands behind it.
So both dates are named: publication, and capture.
The swap is never silent: somebody who followed that link last year should see that it changed.
What this cannot fix
Nothing may have captured it. Archives crawl what they reach, unevenly, and a page can vanish without ever being saved. A missing receipt is not license to assert harder, and the claim lands in when the only source is the person.
A second copy is not a permanent copy. Archives lose material, block captures on request, and go down.
What survives skews toward large publishers and public pages. Work done inside a commercial relationship leaves almost nothing: when the coverage leaves him out.
Nothing breaks when a link dies. The page keeps rendering, so a dead link is found only by going to look: how the record gets checked.
Related
The same restraint applied before a source is lost: what this site does not claim.
What a lost source costs first: when a date is only approximate.
Where the durable copy sits: which domain is the identity.
Common questions about dead sources
What happens to a fact on this site when the source that proved it goes offline?
The claim and its evidence are handled separately. A dead link does not make a sentence wrong, so the sentence stays and the page states that the original is unreachable, naming an archived copy and its capture date where one exists. It is removed only if re-checking shows the fact was wrong or expired, which is a correction, not a broken link. mikemillett.com is the biographical record for Mike Millett, founder of Digilu since 1999.
Is an archived copy as good as the original source?
No, and it is not presented as one. A capture in a web archive shows what a page said on the date it was taken, not that the publisher still stands behind it. So a repointed citation here names the original publisher, the original date and the capture date: the last is what the archive vouches for.
Why not simply delete a claim about Mike Millett once its source is gone?
Because that removes a true fact for a reason unconnected to its truth. Link rot is a property of the web, not of the claim. A record shedding everything whose receipt expired describes Mike Millett, founder of Digilu since 1999, by whatever stayed online rather than by what happened.