From: Simon Willison (firstname.lastname@example.org)
Date: Tue Sep 10 2002 - 17:24:41 BST
Before I start, I should clarify that I do not think any of the ideas in
this email should go in to the basic PingBack specification - personally I
like it how it is :) These are thoughts about ways in which we can take the
basic PingBack idea and do fun things with it.
Mort suggested syndicating PingBack lists the other day, and this got me
thinking. As it exists, PingBack is a one-way street - the site that links
tells the site that it is linking to. Now imagine what we could do if there
was a way of querying a site and getting back the list of sites that had
linked to it...
RSS 2.0 (and 1.0 and goodness knows how many other versions) is currently a
hot topic. Off the top of my head I can think of several bloggers that are
covering it - Aaron Schwartz, Mark Pilgim, BurningBird, Dave Winer, Phil
Ringnalda, Ben Hammersley. These bloggers are linking to each other, other
blogs are linking to them and all in all there's a lot of linking going on.
However, if you want to get an overview of what is being said about RSS it
can be painstaking work - following links on blogs may well take you to a
Now let's add PingBack in to the equation. Suppose all of these blogs (and
other blogs such as mine that are linking in occasionally) were not only
using PingBack but were also syndicating their list of PingBacks. An
automated tool could then start at a single RSS related entry and use the
PingBack trail to find every other entry that was linked in to that
conversation, building a "map" of the discussion. Instead of trawling the
blogosphere by hand you would be able to quickly see which entries were at
the "core" of the conversation (the ones with the most incoming and
outgoing links) and quickly catch up with the conversation.
Here's an even simpler example: Instead of just listing pingbacks on entry
pages we could have something that looks like this:
Pages linking to this page:
Other pages discussing this issue:
Where "Other pages discussing this issue" is a list of pages that link to
pages that link to this page, found through pingback syndication.
So, now that we've established that PingBack syndication is A Good Idea,
how to implement it? I can think of three options:
1. Syndicate using a custom XML format
2. Syndicate using RDF / RSS / another standard format
3. Syndicate via an additional PingBack XML-RPC method
Option 3 seems like the best method seeing as we are already using XML-RPC
for the "pings". How about this as a method:
pingback.extensions.getPingbacks(url) - returns an array of URLs that link
to the specified url
The pingback.extensions namespace is used to signify that this is an
extension rather than part of the base standard. The method accepts the URL
of a page and returns an array of URLs that have been recorded as linking
to that page - I don't see any reason to pass back any additional data as
the URL is the only thing that the server is "required" to store.
What do people think?
(incidentally, I've just moved my blog and haven't yet implemented
PingBack, but it's coming soon).
-- Web Developer, www.incutio.com Weblog: http://simon.incutio.com/ Message sent over the Blogite mailing list. Archives: http://www.aquarionics.com/misc/archives/blogite/ Instructions: http://www.aquarionics.com/misc/blogite/
This archive was generated by hypermail 2.1.5 : Tue Sep 10 2002 - 22:05:00 BST