Just coming back from a few days of fieldwork (preparing ethnographic research in the field of semantic software) I could not help but share something I just learned. It fits quite nicely to what I have written before on the masses of non-human actors that populate the web today (crawlers, spiders, bots) and how the interdepencies between “them” and others (like us) change with the implementing of new web technologies.
Semantic technologies are build to process large numbers of unstructured documents and to automatically find (and tag) meaningful entities. And while these frameworks of crawlers, transforming tools and mining algorithms are actually quite good at finding structure in data, they are still (at least initially, they learn quickly) quite bad in assigning meaningful labels to it. They are quick and good in understanding that a text is about something, but they are bad ans slow at judging ambiguitive terms – they fail at understanding. But a recent trend called “gamification” (which is around for a while but was until recently used mainly for encouraging users to fill out boring forms) now is a good example how the configuration of agency changes on the web today. Human users are asked to play games that help annotating and matching ambiguitive patterns – for tagging pictures, texts, music, etc. So not machines are doing tasks for humans – humans are working for machines.