the semantic web and the property knowledge graph

(WIP)

Once upon a time, our forebears (no, I do mean our direct ancestors, who probably don't appreciate being called that quite just yet) dreamed a dream of the mechanization – the machinization – of language. The Sapir-Whorf hypothesis wasn't quite dead in the ground yet, the dream of cybernetics was onyl a century young, and the good books had Memexes in them. It really did seem as though, if we could but harness the greatest interface that humanity has ever had with itself – language – if we could but just bridle it, gather its complexities into pockets of – partial! we must remain humble. approximate, for even approximations might suffice; are not imprecise things spoken every day, to great effect? – where was I? – of instantiated meaning, set carefully like rubies in their correct relations to each other, strung with expressive netting – why then, each and every man and woman would take one step closer to being more. To knowing what the machine knew, to speaking with the machine's voice, to finding truffles in the dirt with the machine's snorting snout. I'm mixing my metaphors It's a piece about the difficulty of rendering natural language machine-legible. It's on purpose, leave me alone.

The semantic web was a simple but overambitious idea: to organize the web in terms of a general ontology. You would have a Thing – like, say, Apple – and it would have an Address, and live in relation to, say, the Quality called Red. To find out the nature of the relationship, you would walk up and down a system of categories. It was meant to be asympotically exhaustive, definitive, and complete. It was the classic Arsitotelean revenant, born of the death of the Platonic ideals. Everybody knew about Godel, but Godel was a mathematical monster. Everyone knew about the modernists, but come on, their only real problem was that they hadn't had the internet yet. We repeat these cycles once a century probably, (faster now), and hope it's a spiral. And so indeed it might have been: at least according to me, the arbiter of taste in this little corner of the internet, I like that they thought in terms of protocols; that they designed for distributed writing, or at least federated writing. Their authorities were more in the flavor of administrators than monarchs. (todo attempt a better history here than blowharding)

The semantic web did not work.

See, top-down models always have this problem: reality is messy. People are reality.

You can recover much of what there is to like about the smeantic web from property knowledge graphs.