March 25, 2003

Real Standards and RSS

Bill de hӲa makes some interesting points in response to my previous comments about the standardization of RSS - "When do we get a real RSS Standard?" In particular, he highlights the distinction between de-facto and de-jura standards.

I think this gives us part of the answer, but not the whole picture. The main distinction between Bill's de jura and de facto list seems to revolve around certification. Bill makes the point that W3C are not in his list of de-jura standards development organizations (presumably because W3C have not really had any focus on certification testing in it's specifications to date), but OASIS "may be".

This all raises the interesting example of "Oracle SQL" - should we regard that as a "de facto standard" or a "proprietary standard"? I would classify it as the latter, even though it is very widely used. But I can't help thinking we need another term here, because I am sure many people would regard Oracle SQL as a good example of a de-facto standard.

So, there is definitely another useful angle to consider here, and I think it is succinctly described by Patrick Gannon from OASIS with his "traction" and "sanction" dimensions for effective standardization. The idea being that "real standards" require BOTH traction and sanction

Standards Classification Matrix

This is the distinction I was drawing in my original post - along the "sanction" axis, the line between "recognized standards development organizations" like W3C and OASIS [right of center], and "proprietary" organizations like Oracle [left of center].

The migration of WS-Security from left to right on slide 7 of Patrick's presentation is precisely what I am saying needs to happen to both RSS and most of the WS-* specifications currently doing the rounds.

However, I certainly need to add a "traction" angle to my checklist for "real standards", to avoid the problem that Danny Ayers points out:

And what of 'standards' that meet all of Jorgen's tests, yet nobody uses?
[ Danny Ayers - Would the real RSS stand up ]

Yes, "real standards" are the ones people use!

Of course, actually "real standards" are those ones that live in the top right corner - having both sanction AND widespread traction / use.

So getting back to where we started from, what about RSS?

Well, I would classify RSS into the top-left quadrant - widespread usage/traction, but still a largely proprietary format / specification. The situation is even more confused then you plot the positions of the different versions of RSS!

RSS Version Standardization

So, what needs to happen to RSS to make it a "real" / ubiquitous standard ?

  • We (the industry) need to decide on ONE single version for the format, and stick to it. RSS v2.0 would seem the obvious candidate here.
  • All attempts to "fork" RSS need to be resisted - for example, proposals for RSS v3.0 have possibly more similarity with SMTP as they do with any previous version of RSS! RSS 3.0 is not even XML based (!) and is clearly an attempt to hitch a free ride on the previous good name of RSS, IMHO.
  • We need to get the RSS specification some official "sanction" - probably through OASIS I would imagine, but maybe other places would be more suitable? If it goes to W3C, we will just end up with RSS 1.1, which would not be sensible IMHO!

BTW, none of this should be taken as a downer on the good work Dave Winer did to nurture and create RSS, but I believe it is now time for the format to grow up and leave the nest, and take its rightful place alongside the other ubiquitous Internet standards the industry now relies on.

Entry categories: RSS Standards
Posted by Jorgen Thelin at March 25, 2003 03:00 PM - [PermaLink]
There's an awful lot of ancient and modern historical baggage around RSS (for ancient history see, more recent try Burningbird and elsewhere). When Dave Winer announced his version 0.94 would overnight become 2.0, a lot of people hoped he would take the opportunity to unify or at least narrow the gap between the specs. Unfortunately he had his own ideas, and damaged the situation further. Of significance to your comments are the fact that RSS 1.0 is considerably more versatile than 2.0 thanks to its RDF base and well-defined modular nature. For every new namespace used with 2.0, the whole of the new language has to be designed from scratch, whereas RSS 1.0 can reuse the RDF framework. RSS 1.0 also passes 4/5 of your tests, where RSS 2.0 only manages at most 2 (the final or definitive status is difficult to ensure when the thing is in the hands of one man, but I'll give it the benefit of the doubt). RSS 3.0 was released largely in reaction to Dave Winers 2.0, to demonstrate the weaknesses of that spec (and the childishness of his numbering) - you note that RSS 3.0 is not even XML based. This was following DW's reasoning about simplicity. Posted by: Danny on March 25, 2003 05:00 PM
I'd love to see everyone settle on a single spec. I've read all the related entries on this site as well as some of the linked commentary from other blogs, and I follow your logic regarding the need for a single standard. But near the end of this entry, you made an assertion which you failed to explain... the second sentence of this quote, "We (the industry) need to decide on ONE single version for the format, and stick to it. RSS v2.0 would seem the obvious candidate here." Why do you consider RSS 2.0 to be 'obvious'? I've read all the specs (yes, all, starting with 0.90 from NS). I also follow the debate (and occaisional whining) between the factions on RSS-DEV. I personally like RSS1.0. It appears to do everything RSS2.0 can do, only more extensibly, with better use of existing standards (read: namespaces), and with more potential interoperability (RDF, as RDF gains more ubiquity). What am I missing that makes 2.0 'obvious' as a basis for a standard? While I think that 2.0 is more commonly available than 1.0 (although many sites offer both), I don't think ubiquity implies suitability. Am I incorrect? Posted by: Jason Clark on March 30, 2003 02:59 PM
I was viewing 2.0 as the most "obvious" version mainly due to the traction I have seen for it for my weblog, and similar pictures elsewhere. However the current stats from the Syndic8 website seem to paint a slightly different picture - v0.91 is way out in front, followed by 1.0, and with 2.0 lagging in third place. I guess RSS 2.0 is probably not such an "obvious" choice after all! Posted by: Jorgen Thelin on March 30, 2003 04:52 PM
Update 18-April-2003: An Internet-Draft of RSS 2.0 has been produced - more details here. Posted by: Jorgen Thelin on April 18, 2003 08:28 PM