Files are probably the most widely-recognized metaphor for persisting digital information: an opaque container with whatever inside, to which an arbitrary label is attached. The URI, taking cues from the naming schemes of hierarchical file systems, extends this metaphor — with all its benefits and shortcomings — to the Web.

The first problem in this space is that people are really bad at naming things, a condition to which anyone with an Untitled-1 on their system can attest. Moreover, a computer's latent threat of discarding your hard work unless you choose a name right now does little to inspire creativity in this arena.

This problem is exacerbated by a second: you have no control over who or what links to you. Once you mint a URI, advertise its existence and reply to a request for it, that's it: it's out there for good. Of course, the courteous thing to do is to keep your URIs the same — forever. This promise isn't so easy to keep when you can rename or delete resources on a whim.

This lack of control over inbound links is one of the central criticisms of the Web held by Professor Ted Nelson, the progenitor of hypertext. His alternative, in essence, is to tether documents together with permanent, bidirectional links that enforce citation, copyright and compensation for the author. It is also worth noting that implementing this model would almost certainly require a common carrier to control these links — a fact which essentially describes Nelson's original business plan.

However, our problem goes beyond mere courtesy and into the realm of serious business. In the mid-1990s, some guys made a bet that every link to a resource is a vote for its fitness. By proxy, this bet gave birth to an entire industry known as SEO. The net effect is this: if your resource got hastily named, giving it a better one without compensating for the original will likely do more harm than good. A final complication is that more pithy names — aside from being more helpful to people — are also found to contribute to this fitness rating, driving incentive to do so. Therefore: