Different strokes

Here's what Brent Simmons had to say about yesterday's item on news scanning and news reading:

1. NetNewsWire's Combined View works with channels, all-new-headlines -- and groups.

For instance, I have 145 subscriptions organized into 8 groups. When you view a group in the Combined View, you see all the unread headlines for that group. Like most people, I organize my groups by topic.

So I have a page for Macintosh news, a page for weblogs, a page for books, a page for baseball, etc.

2. Generalizing about NetNewsWire based on Steve Gillmor's use of it isn't fair.

For instance, I personally find Radio's batches-of-100-in-a-web-page to be awkward. With Radio I can't scan fast enough and I can't keep up. I developed NetNewsWire so I could keep up with more feeds with less effort. But everybody's different: different presentations work for different people. That doesn't mean that Radio's approach is better or worse than NetNewsWire's.
I agree. In trying to illustrate a point about scanning versus reading, I'm afraid I fanned the flames of the newsreader-style versus browser-style debate. In fact, the two modes can be complementary. I just bought the full version of NetNewsWire, which exploits that synergy as Brent describes. So does FeedDemon, which this posting prompted me to re-explore.

It's true that different folks will prefer different strategies for grouping and processing their feeds. But no matter which strategy you prefer, you need to harmonize two modes: scanning, and reading. And no matter which strategy you prefer, the same methods can be used to achieve that harmony. On the publishing side: untruncated feeds, containing HTML (ideally, but not necessarily, XHTML) markup, with a first element that can work standalone. This is often naturally the case, since a lead paragraph's job is to hook the reader.

On the consumption side: feedreaders that XHTML-ize content (in case it isn't already XHTML), use the first markup element to optimize scanning modes, and provide the full content for reading. Peter Eschenbrenner suggests this is already possible with FeedDemon:

You might want to check out FeedDemon by Nick Bradbury. While it comes with default style sheets, users are able to create their own XSL for efficiently processing the information. So, if you wanted to view just the first paragraph, you could create your own style or ask someone in the community to create one.

Radek, an active community member, has created styles that hint at what can be achieved with this combination, from rating your feeds in a database, to creating powerful MindMaps. [Peter Eschenbrenner: Note to Self]

Interesting! So far as I can tell, though, FeedDemon's XML data model corresponds -- as you'd expect -- to that of RSS. Which means the content is opaque. So while you can use XSLT to hack alternate presentations of channel-level and item-level dates and titles, XSLT can't see into the content. For example, the default stylesheet includes:

<div class="newsitemcontent">
  <xsl:value-of select="description"/>
But if it unescaped and XHTML-ized the markup within the description, FeedDemon (or any RSS reader) could expose the content of items to the same kinds of XML manipulation that we routinely apply to the RSS metadata wrapper. (For all I know, there may even be a way to do this now in FeedDemon, by breaking into the XML pipeline and inserting an HTML Tidy step.) Selecting initial elements, in order to normalize and improve summary views, is one reason to do that. The structured search technique I've been exploring is another. I mentioned yesterday that these opportunities have nothing to do with the RSS/Atom debate. As I should also have mentioned yesterday, they have nothing to do with the newsreader/browser debate either.

Bottom line: blog content needs to become a first class citizen in the XML world. And as it turns out, it's more feasible than I thought to make that so. Most people won't be producing well-formed content anytime soon. But tools that produce and consume RSS can compensate, with the help of things like HTML Tidy, and there are compelling reasons to do so.

Former URL: http://weblog.infoworld.com/udell/2004/02/22.html#a925