So many layers of the semantic web stack, such as digital signatures, logic processing, are years away, but SIOC has the most groundbreaking implementation I have seen in 6 years of semweb watching.

This viral explosion is a Man (RSS 2.0 driven Web 2.0) vs Machine (Semantic / Data Web or Web 3.0) type affair. Man cannot work faster than machines, but Man can reason much better than machines. Ironically, the current Web 2.0 state of affairs has this reality the wrong way round, but the realization will be the very inflection upon which the unravelling of the Semantic Web occurs.

The Semantic Web brings Context to the broad mass of Content across the Blogosphere, Wikispehere, and Web in general. Shared ontologies such as SIOC and FOAF are examples of critical infrastructure missing in Semantic Web 1.0.

The Semantic MediaWiki approach takes it a step further, and it does it in a really nice way. I remember the Jot Wiki announcing something similiar a year or two ago (unfortunately, jot is neither free nor Free, as far as software goes).

For me, the most compelling thing about Semantic MediaWiki is that users are able to edit both the structure of information and the information itself (see the page for Laptop versus the page for Template:Laptop). However, I’m not sure how well this will play out in the long run. What I mean is: the templates (creating the structure of data) are editable–and this is a desirable thing–but what happens to the old data when the structure changes? It becomes a little less useful and a little harder to machine-process. Also, designing structure by committee can be a lot harder than defining content (it’s a lot easier to define “OBJECT is a CATEGORY” (e.g. a Honda Civic is a Car with x miles-per-gallon and y weight) than to define exactly what characteristics a car should have). Each user will have a different idea of what characteristics should and should not be included, and what they should be named. I suspect agreeing on stable templates will be a lot harder than creating instances of the templates. And this is dangerous, because it’s the templates, the definition of data structure, that make the system practically usable.

Are there any thoughts/solutions to this?