RSS, RDF and syndication again
Prentiss Riddle
of RSS, RDF, and syndication on her weblog (you'll want to click on the "more"
and "N mice roared" links):
http://www.burningbird.net/weblog/2002_03_01_burningbird_archive.php#10368033
She's somewhat critical of RSS as too intimately tied to Userland
Radio and insufficiently flexible compared to full-blown RDF.
I'm going to pastebomb my reply from her discussion to here:
Here's another vote for Bill's point: RSS was designed for
syndicating headlines, not full content. While syndication of more
complex content is also a goal worth working on, I personally am
quite attracted to the headline model and RSS. I see Shelley's
point about "control" and RSS, but am I correct in gathering
that this criticism applies mostly to RSS 0.92, not 0.91? If so,
this together with Bill's argument should rehabilitate RSS 0.91:
as I understand it, 0.91 does a good job of doing headlines,
while 0.92 reflects Dave Winer's and Userland's desire to remove
certain "limitations" in 0.91 in order to cram in more content.
To which I say, here's to knowing when to stick with a good set
of limitations!
I've got my own experimental mini-aggregator going as a sidebar on
my own weblog. Unlike aggregators that work on the "portal" model
(e.g., Fyuze), I use individual items or headlines as the unit
of granularity I'm aggregating, not whole RSS files or channels.
Seen that way, the issue of context quickly becomes important.
Part of context that's missing in an RSS feed is the "community"
and "style" of its original environment, as Shelley points out.
But since you can presumably click on an item of interest and
see it in its original context, that's not a major problem.
More serious is whether you can understand the headlines in order
to determine whether they are of interest in the first place.
Jakob Nielsen has long been berating us to write appropriately for
the web; Jon Udell reminds us of the standby journalistic concepts
of heads, decks and leads. As Jon says, these stop being matters
of netiquette and start being engineering principles when you're
aggregating RSS headlines and find yourself trying to make sense
of cutesy-pie items like "A Tail of Treachery" or "Driving Mr.
Frodo". (Yes, I'm as guilty as the next person of being too cute
for my own good.)
Which simply means that content syndication is as subject to
human failings and the garbage-in-garbage-out problem as any
other decentralized system on the web. High-quality metadata
can only come from third parties, preferably well-trained and
disinterested ones. That's why libraries don't rely on the
publisher-supplied cataloging information inside the front covers
of books and Google doesn't trust Dublin Core descriptions embedded
in HTML headers. I think maybe that RSS headlines, ambiguous as
they are, are about the most we can expect webloggers to produce
with even minimal reliability (and that only with tools which
make doing so a no-brainer). RSS as a lingua franca would give
some of our familiar tools (Top 40 systems a la Daypop, voting
and discussion communities a la Slash or Fyuze, hopefully a decent
if-you-liked-that-then-you'll-love-this recommendation engine like
Amazon, keyword-based stream-sniffers like Google with a memory,
etc.) a handle with which to grapple with Blog (and other newsy)
content. At least I'd like to think so.
For links and discussion, see either Shelley's site or mine:
http://www.burningbird.net/weblog/2002_03_01_burningbird_archive.php#10368033
http://www.io.com/~riddle/toys/?item=20020306
-- Prentiss Riddle ("aprendiz de todo, maestro de nada")
-- rid...@io.com / http://www.io.com/~riddle/