[rfc-i] DOIs

Henning Schulzrinne hgs at cs.columbia.edu
Tue Jan 26 17:04:24 PST 2010

In theory, yes. In practice, how many people outside the IETF crowd (which does not need any URL, since they can presumably find RFCs using telnet if they had to) would know what to do with this, even assuming that there's a resolution mechanism in place for those URNs? With a DOI, every grad student can figure it out (and will have figured it out by the first semester, since IEEE Xplore and the ACM DL both use them, along with Springer and the other publishers). If they don't, the reference librarian will gladly show them.

All they get when they type in that URN is 

Safari can’t open “urn:ietf:rfc:1000” because Mac OS X doesn’t recognize Internet addresses starting with “urn:”.

The only real use for any RFC-related identifier is for people *outside* the normal IETF crowd. The reason we got an ISSN for the RFC series was to make it easier for "outsiders" to reference RFCs, except that they only identify the whole series. Same idea here.

I have no objection to using any URL (such as http://www.rfc-editor.org), but I'm a bit baffled by the not-invented-here syndrome in evidence. If we want to make our product accessible in the wider citation marketplace, we play by their customs, as opposed to having everyone else play by our idiosyncratic version.

One of the things happening in the digital library space is citation matching (for both bean counting purposes and to make it easy to jump to articles from the citations). Having a unique identifier that everyone agrees on is very helpful for making that work well. It is rather unlikely that the rest of the world will go for various URL formats, since the variability is just too large to be useful.

Exactly what harm would ensue if RFCs had DOIs?


On Jan 26, 2010, at 7:50 PM, Julian Reschke wrote:

> Henning Schulzrinne wrote:
>> As far as I understand the DOI model, the "publisher" (here, the RFC editor) allocates the numbers after the / locally. Presumably, this can just be the RFC number, so the extra work is essentially zero.
>> Compared to URLs, the DOI is not protocol-dependent (i.e., will work after HTTP goes away) and happens to be somewhat shorter than the URL. I find it peculiar that, among all technical publications, RFCs consider themselves so special that they have to re-invent what the rest of the technical publishing community has already done.
> We already have all that: "urn:ietf:rfc:...".
> BR, Julian

More information about the rfc-interest mailing list