I think I just got it. For the past few weeks I’ve been puzzling over what the OPML, RSS, AJAX alphabet soup will ultimately end up tasting like.

I’ve intuited for a long time that the whole gestalt is far more significant than most programmers or technology commentators realise; and of far more ultimate utility than as a succinct method of information categorisation. I now realise, OPML (or an OPML like outliner standard in XML) underlies the future of both the browser and the web.

Firefox 3, or its equivalent, won’t function primarily as a traditional link / url -> page display browser, rather, users will navigate through outline directory trees to reach their ultimate content destination – which may be any of a whole variety of open document types, inclusive of audio, video, and traditional text / graphic / interaction models.
Nodes will be linked dynamically, and updated at numerous trusted hubs (the del.icio.us’s of tomorrow). Such links will create sub webs, navigable and discoverable through reputation systems, tagging and recommendations.
Further, users will not merely navigate such OPML trees laterally, but through any of a whole variety of interface paradigms.

Where today each link on a site sits in relative isolation, the browser of tomorrow will aggregate all links on a given page in real time, construct and meaningfully ‘geographically’ categorise link feeds, which will provide both an additional outliner navigation layer, and a new means of scanning the content laid out within a document. This will be the hardest element to get right, as it departs most radically from out the web works today. My guess is that the ultimate solution will be something like newsvine, dynamically constructed, parsed through link, feed, and generator templates (e.g.: blogging engine, CMS) from any given page site or outliner – both in real time by the browser, and by next generation sitemaps (in reality linkmaps). Think google news, for every site on the web (and its linked sub pages and sites).
Todays feed grazers could be the templates for tomorrows browsers. Such browsing paradigms may finally provide an advantage for three dimentional interfaces – though my guess is two dimensions will remain more comprehensible and intuitive.

A few more interface ideas before I lay down the crystal ball. Pre-cached feed branches displayed as graphical document previews in a mouse over ‘mind map’. A home feed bucket which rises from the browser bottom to catch feeds, pages and documents dragged and dropped (think OS X’s dock, with icons representing not programmes, but outlines in your home OPML). Or how about a dynamically generated zooming interface like Jeff Raskins Archy project.

The best part is, such novel methods of navigation could be implemented today in AJAX as proof of concept, sitting on top of the web as a hotkeyed interface, which is arguable what the Flock guys are positioning themselves to do; but ultimately such technologies are unlikely to be fast enough to produce a robust solution.