Last week there was a story on NPR about a dancing parrot. A neuroscientist in San Diego discovered a sulfur-crested cockatoo named Snowball dancing to the Backstreet Boys on a YouTube video. There were a number of interesting aspects to the story, including the fact that YouTube is now being used as a research corpus for animal behavior. What caught my ear however was the mention of follow-on YouTube research by a graduate student in the psychology department at Harvard, which found that the only animals which exhibited dancing skills on YouTube were 14 species of parrot and an elephant. The graduate student notes that like humans — and unlike dogs or cats — parrots and elephants are both known to be vocal mimics. They can imitate sounds. The hypothesis, then, is that our ability to dance is a byproduct of our ability to vocally mimic others.
In the development of language, mimicry is crucial. We learn to speak by repeating what others our saying. We acquire vocabulary by hearing the words that others use. We almost never acquire vocabulary by looking for words in a dictionary.
Last week, I attended a "Semantic Web Meet-Up" in NYC. One of the speakers was describing Knoodl.com, which is described as facilitating "community-oriented development of OWL based ontologies and RDF knowledgebases." I think its fair to say that Knoodl would like to be a sort of dictionary for the semantic web. To my mind, the semantic web just hasn't happened yet because its been very hard to connect data from different knowledge silos. Vocabulary used in one silo tends not to get used in other silos. Ironically, the twitterers attending the meeting couldn't even arrive at a common hashtag for the meeting- at least three tags for the meeting were used. I'm not sure if that fact says much about Twitter hashtags or about the people attending the Meetup. One of the most intriguing things to me about Twitter has been to observe how hashtags are propagated. I find myself mimicking others as I slowly learn the vocabulary and grammar of the new environment. It struck me that it is this quality of Twitter that makes me want to anoint it as a substrate for semantic web actualization.
Maybe the semantic web needs more than just dictionaries and registries and authorities and linked data to become the next big thing. Maybe what it really needs is some dancing parrots. Software agents that have the capacity to mimic the semantics of the other software agents in a global environment.