At some poncy London winebar last night, Jason made an interesting point about weblogs, saying that they could be considered as an enormous database, a single weblog a slice through this, this page being an Interconnected cross-section of all that content. Interesting, because I'd never thought before of weblogs as a mass, a whole, a single thing.

But it's not [an individual weblog isn't] really a slice, it's more of a trail, a chronological record of a passage through the web (a trail in the memex sense). Every so often, the trails overlap, and there would be value in seeing who else has crossed this point (a point being a node, a post, often a link, or a piece of commentary), and maybe following their trail for a while instead. But a trail is 1-dimensional, and a slice should be more. So how to do this?

Firstly, all posts should be categorised (possibly into multiple categories). This would mean that from each step on the trail, you could step sideways in the same category onto a different path (or rather, the same path but a different point in the time-stream. But you'd be crossing different other people's paths). Categories very important.

Secondly, for commentary-only posts we need transpublishing so it's easy to make references, comment, object and so on, and that such references are implicitly joined to their origin.

And as long as I'm dreaming... At each node you should be able to see what else is nearby, based on a web-of-trust-like system. And the entire thing should run through a universal map, so that the interface can be anything from a text adventure to a Nokia phone.