I am a huge advocate of metadata. My only problem with your distributed network is having too much metadata. You have to ask yourself what’s important and what’s not. Lets say I saw a comic, but I don’t remember what strip it was or who the creator was. All I remember is that it had a cat. The cat is pretty damn funny. Well if I searched for a web comic that has cats and a result came back for every comic that had a cat in the background (or any little thing that might resemble a feline) I’d probably short myself. Yeah, you could have relevence rankings, but (as google shows us) sometimes somebody elses ranking is much different than out own.

Metadata is very cool, though. I really hope someone creates a standard for it and incorporates it into all OSes.

It’s the same with your “cat” search. Maybe the character speaking was a cat, like in Garfield. Maybe it was just a background prop, like the cat in Penny-Arcade. Maybe it’s a desk lamp shaped like a prop. Maybe somebody said “cat.”

Useful metadata is more than just a text file with keywords. You need ontologies (categorical breakdowns) as well.