the iCite net > news / blog > a permalink

news and thoughts on and around the development of the iCite net
by Jay Fienberg

Mark Pilgrim's Atom linkblog

posted: May 27, 2004 3:30:11 PM

Mark has posted an excellent tutorial, How to make a linkblog in Atom. I find Mark's tutorials to be admirably well done. A couple comments follow.

As I mentioned recently, I think Atom is important because so much effort is going into designing its usage protocols. Mark's tutorial is a good example: it's basically a protocol-by-example, and I think it reveals a clearer design than what one gets out of the format itself.

Definitely look at Mark's linkblog itself too—note that this page is actually the Atom XML file with an XSL stylesheet. Nice example of how well that can work.

Here is the entry Mark describes in the tutorial. Note the cross-reference to a related link. Conceptually, this is very interesting. I wonder if the way its visually presented could be refined (in other words, something seems off about it to me, but I don't have a good suggestion, so maybe I'm wrong).

Structurally though, both the main link and the related link are both expressed through the same data structure, a <link rel="related" . . . > tag. This is concerning, as the structure doesn't reflect the two different meanings (i.e., primary citation vs. related / cross-reference citations) that Mark is using.

This makes me question the use of the link tag altogether in this context: why not just put the links and summary together in a content element? I guess the easy answer is that it's easier to parse a separate summary and separate links—and since one of the links is being used as a linkable title, this seems required.

But, in general, this suggests to me that a linkblog and a content blog are not really the same interaction protocol, since a link blog always links out rather than being a destination (e.g., of content), and a content blog can either link out, be a destination, or be both (though content blogs, practically as a standard, are represented in RSS/Atom as destinations, i.e., the "main link" title is a link to the content, and not a link out).

The mixing of these protocols is evident in RSS, where the meaning of a link element is not strict (and, which I found to be a totally disorienting experience when clicking those main links in RSS readers). And, given that history, it looks like Atom is making a good effort to clarify and account for both protocols.

But, if I ran the zoo, I'd just have two, slightly different, formats for link and content blog feeds. (And this would assume that people would have feed readers that could just deal with processing the different formats, without requiring users to know that they are different formats). In particular, do I want to "read" a linkblog the way I want to read a content blog? Probably not.

But, the formats are less important than the protocols. And, as I said, the effort going into describing these protocols for Atom is an indicator of why it's more important than just a format—and even why some of these format constraints aren't worth being too concerned about.

permalink | comments {2} · trackbacks {1}

also available as: rss · rss2 · rdf · atom

Comments and Tracbacks

Comment by: Mark ·
posted: May 27, 2004 6:35:02 PM

I will have a separate tutorial on how to use link rel="related" to point to related posts/articles after a post with actual content (i.e. not a linkblog).

I initially resisted the format of the Atom linkblog as suggested on the wiki, but now I see the wisdom of it. rel="alternate" is *always* the permalink. rel="related" is *always* a link to something else. If all you have is a permalink and an outgoing link, then you have 2 link tags, one "alternate", one "related". If you have a cluster of links (my recent link to the Yahoo toolbar anti-spyware thing was a cluster of 5 links), then you have multiple "related" links. I think the order of the link tags is enough to distinguish them. (Oh yeah, I had a "via" link there in there too.)

This is actually what I''ve always wanted in a linkblog. Many links are really part of a cluster. Now I can represent that cluster of related links, *and* give credit, *and* unambiguously specify my permalink (so others can give credit to me, in turn). Linkblogging is so much more fun now. I always thought about links this way, but I never had any way to express it except using separate entries.

Maybe cluster linkblogging won't take off. That's OK. Laying down a "best practices" for different types of feeds benefits everyone, even if most people only use the bare minimum of functionality.

Comment by: Jay Fienberg ·
posted: May 27, 2004 7:40:10 PM

Thanks for your comment, Mark.

I think the clustering of links is quite significant, and am really glad to see you exploring it, both in terms of its presentation, and in terms of its data structure. (And, permalink + cluster + attribution = great.)

I guess it makes sense that "related" + link order indicates priority, e.g., the first link is the main one, and the others are "see also". In the scope of Atom, I don't imagine it actually makes sense to add another parameter to the links--it might as well be Xlink or something then.

But, my bibliographer background makes me imagine expressing more subtlety in these relations, e.g. a main citation and possible "see", "see also", and/or "compare" secondary citations.

I was imagining something like:

<link rel="related" . . .>
<link rel="related see" . . .>
<link rel="related seealso" . . .>
<link rel="related compare" . . .>

But, then this gets into the realm of VoteLinks and XFN, and that makes me think that this might be a can of worms better to leave closed.

trackback from: the iCite net development blog
posted: Oct 30, 2004 6:26:35 PM
title: Interaction protocol more important than format

I've already said it a number of times in passing ... but I thought this deserved it's own post: The protocol of interaction is more important than the format.

Note: All comments and trackbacks are moderated. Spam is deleted. Other comments are approved as promptly as possible.

Note: Older posts no longer accept new comments or trackbacks.

« prev post
What did you just say? On data markup

» next post
Web log entry type catalog

blog newsfeeds

brief content:

 XML  ·  RSS  ·  RDF  ·  Atom 

full content:

 XML  ·  RSS  ·  RDF  ·  Atom 

blog archive

jan · feb · mar · apr
may · jun · jul · aug 
sep · oct · nov · dec
jan · feb · mar · apr
may · jun · jul · aug
sep · oct · nov · dec

jan · feb · mar · apr
may · jun · jul · aug
sep · oct · nov · dec

may · jun · jul · aug
sep · oct · nov · dec

first post: 
April 30, 2003

highlight views:
Spammers' Choice

Jay elsewhere online
Jay Fienberg - the official home page

Wrong Notes - the music blog of the Ear Reverends

Fine & Full, aka, a fine and full burger

Sociomobilepoetextologia (moblog, currently inactive due to lack of proper mobile)

to enjoy roll
sites I like to read when I start from here

· Anastasia Fuller
· Andy Baio
· Biz Stone
· Boris Mann
· Bre Pettis
· Chris Dent
· Danny Ayers
· Dare Obasanjo
· David Czarnecki
· David Weinberger
· Don Park
· Evan Williams
· Greg Narain
· Jason Kottke
· Jim Benson
· Lucas Gonze
· Marc Canter
· Matt May
· Matt Mullenweg
· Michal Migurski
· Nancy White
· Rebecca Blood
· Reg Cheramy
· Richard MacManus
· Sam Ruby
· Shelley Powers
· Tim Bray
· danah boyd

powered by blojsom

Entries by blojsim