Re: [syndication] Thoughts, questions, and issues.

[Stephen Downes <stephen.downes@ualberta.ca>]

Dave Winer wrote:

> Stephen, with all due respect, we have two schools of thought here.
> The extensibility school thinks they're right. They've moved aggressively,
> and call their spec RSS 1.0.
> Those of us who believe in simplicity and working together, think RSS should
> go in a different direction. I gather you don't agree.
> So what's the solution? How do we avoid a total meltdown? How do we avoid
> having two specs with the same name?

OK, the biggest problem I've encountered so far reading RSS and other
site summary files is the proliferation of formats which already exists. For
me, what this means is writing code for each different type. It doesn't have
to be a lot of code (and it isn't), but I don't like where this is heading. If
I have to deal with six or seven different flavours to handle even the 6000+
feeds that are out there, how am I going to be able to write code when
there are 6 million feeds?

What I see happening is that each developer adapts the standards to his
or her own needs. I just viewed iSyndicate's RSS feeds, for example,
and saw yet another variation on the theme. It's recognizably RSS, sure,
but I have to parse the file slightly differently in order to extract the
channels. And I don't blame iSyndicate - I think they're doing what they
should do. But the fact is iSyndicate - and Scripting News, and Userland,
and Moreover, and more... each adapt the standard slightly to fit their
own needs. I don't see this ending, and as a result I think that any simple
and rigid spec will be left behind.

This because provider sites know that aggregators will adapt their
scripts - with reason - to collect the content.

This is why I am in favour of extensibility. It allows a site to define a
flavour or variation, to code it in a schema or DTD, and to allow
aggregators to parse content based on that schema. No code rewrites.
Happiness all around.

This is why I am also in favour of a distributed approach. Not all
aggregators want everything. Some just want headlines. Some just
want staff listings. Some just want keywords. A distributed approach
lets an aggregator scan a site summary and then to drill down to
more precise summaries to get at what they want.

I don't think a 'one size fits all' approach will ever work on the web
ever again (as if it ever worked in the first place). I want simple -
I really want simple - but not at the cost of flexibility. And yes, I
do believe in working together, and I think that also requires
flexibility.

--
Stephen Downes - Information Architect - University of Alberta
stephen.downes@ualberta.ca  http://www.atl.ualberta.ca/downes