Interesting idea, very like the core of the EPISTLE Templates implementation architecture. Micro-Formats are small re-usable pieces of semantic (intended to mean something to a human first, machine second) where each may conform to any appropriate existing standard, but which can be re-used and assembled into larger messages and applications. Via Monkey Magic, where Piers is still concentrating on the taxonomic / ontological aspects of the semantic web. Lots of good links if that’s your bag.
It’s still my bag in a way, despite focussing on the philosophical foundations of information, but it’s good to see that more and more of the implementation developments focus on more abstract aspects of architecture rather than explicit schemas and the like. Definitely more long term mileage in that.