Are URIs really names?

Henry S. Thompson
School of Informatics, University of Edinburgh
Markup Systems
16 Oct 2010
Creative CommonsAttributionShare Alike

1. Introduction

URIs are at the heart of the Web

Like much of the computational universe

Their official definitions have never quite matched actual practice

Three examples:

  1. The 'httpRange-14' issue
  2. The semantics of the HTTP response codes
  3. The composition of redirection and fragment identifiers

2. Why ask "Are URIs really names"?

One of the primary roles of "the philosophy of ..."

Within the web community discussions about URIs use words such as 'identify' and 'denote', as well as 'name' itself

These are terms of art within the Philosophy of Language

3. A previously unsolved problem?

The status of names within the Philosophy of Language is by no means settled

We can identify at least three positions:

A name is short for some collection of descriptive properties which pick out its referent(s)
A name is attached to its referent by a (typical ostensive) act of baptism
  • The result is then a so-called rigid designator
The referent of a name is determined by its (evolving) use by a community of speakers

4. The official story about URIs

There is a moderately clear official consensus about URIs

Here's a summary of the WebArch version:

(Here, as in WebArch, http: URIs are the kind of URIs most clearly in focus)

5. Parallel number 1

'Resource' names a role in a story, not an intrinsically distinguishable subset of things

Things are resources because someone created a URI to identify them

Things are referents because some word or phrase refers to them

6. Representations?

'Representation' names a pair: a character sequence and a media type.

Just as, in order to interpret utterances or enscriptions, we need to know the language they are expressed in

7. Russell or Kripke?

Compare this illustration from WebArch:

[no description, sorry]

With a Fregean view

[no description, sorry]

Or Kripkean view

[no description, sorry]

On the face of it, the AWWW view seems to be very close to the Kripkean one.

8. What about meaning-as-use?

This certainly looks attractive from the perspective of "computation in the wild"

Does the Wittgensteinian position distinguish names from any other kind of referring expression?

Consider the XHTML namespace URI

9. Resources and identity

Resources are not immutable

The idea of a 'home page', whether for persons or institutions, has been an important part of the Web since very near its beginning

And home pages not only may change

Some mutability is clearly part of e.g. the Kripkean story for names

But, is the time-varying nature of the resource in some way too flexible for the rigid designator view?

10. The paradox

The thing which most distinguishes URIs from other referring symbols

Is also the thing you can rely on the least

11. Two Webs, two kinds of URIs?

Maybe we should treat OFW URIs as distinct from SemWeb URIs

SemWeb URIs are (Kripkean moving towards Wittgensteinian) names

Namespace URIs and other "not accessed" URIs similarly

But most OFW URIs are not names

Perhaps our manifest inability to define "information resource" is trying to tell us something

12. More, different, history

The thing we're interested in on the Web has a complex history

It's worth noting that name-URIs (URIs which refer, which don't respond to access attempts) are quite old

13. What are names for?

Names function as names within propositions

Even on non-strictly-analytic accounts

The primary propositions OFW URIs occur in are HTTP requests

SemWeb URIs do occur in sentences

14. Names are discoverable

We know what to call things

There is nothing corresponding to this for URIs

15. notes

Some (syntactic) names are indexicals: e.g. 'Air France 5050'

History (My email of Date: Mon, 24 Nov 2008 12:54:51 +0000 to Kunze, Fielding and Masinter, Replies to history query), Three phases: pre-theoretic; OFW (2616); SW (3986).

Apparent differences: retrieval of representations

Apparent similarities: Kripkean parallel: baptism, indefeasability (maybe)

Indexicality---useful or marginal? It is hard to think of using a file: URI for other than an information resource---what about file:///

16. References and further reading