How a link spreads
Publication is the last moment anybody controls an address. After that it copies outward through channels that have no way to update themselves.
The copying is automatic
Very little of the spreading is people deciding to share something. Scrapers collect addresses from any page that publishes them and feed aggregate lists. Those lists get scraped in turn. Search engines index every copy. Archives capture pages on their own schedule. A forum post quoting a directory gets quoted by another forum post.
Within weeks an address exists in more places than anybody could enumerate, most of them created without human involvement, none of them holding a pointer back to where the address came from. There is no supply chain here, only sediment.
Corrections do not travel
This is the asymmetry that makes the whole thing hazardous, and it is worth stating precisely. Publication propagates outward through every copying mechanism at once. Retraction propagates through exactly none of them.
| Event | How far it travels |
|---|---|
| An address is published | Everywhere, within weeks, automatically |
| An address is retired | To the roster, and nowhere else |
| A directory corrects an entry | To that directory only |
| A wiki edit is reverted | To the wiki, never to the pages that cited it |
| A forum post is corrected in a reply | To people who read three days later, which is almost nobody |
How one bad address becomes many
A single wrong entry does not stay single for long, and this is why a wiki edit that gets reverted within a day can still be causing harm a year later.
- A wrong address appears in one place, whether by attack or by somebody transcribing carelessly.
- Scrapers collect it within days, and now it exists in several aggregate lists whose operators have never looked at it.
- Those lists get cited by forum posts and smaller directories, each citation making it look independently confirmed.
- The original is corrected. Nothing downstream notices, because nothing downstream is watching the original.
- Search engines now index a dozen pages carrying the same wrong string, and the agreement between them reads as corroboration.
This is exactly why agreement between directories is much weaker evidence than it appears. Three lists carrying the same address may be three independent checks or one error copied twice, and from outside they look identical.
What follows from this
- Age matters more than reputation. A recent entry on a mediocre site beats a stale entry on a good one.
- Agreement is not corroboration unless the sources are genuinely independent, and in this space they usually are not.
- The largest lists are the most derivative. Comprehensiveness comes from aggregation, and aggregation inherits every upstream error.
- Nothing you find can be assumed current. Not because sources are dishonest but because none of them have a mechanism for becoming less wrong over time.