Wolfram Alpha, Dalgarno
31 March 2009
For some reason somebody told me about this Wolfram Alpha man who wants to make … it’s hard to explain … he wants to make this web engine that is maybe in some sense like Google, in that it has access to a bunch of information, but, unlike Google, it understands the relationships between pieces of that information; so if you ask it a question (it also understands English), it doesn’t do what Google does, which is go find a document that answers your question, but instead just figures out the answer to your question and tells you.
It sounds a bit farfetched, but apparently he’s more or less done.
Leaving aside for the moment my suspicion that any interesting question will immediately break the thing, and assuming it works perfectly and comprehends the entire body of human knowledge, it still just makes me feel tired. I don’t care if some machine is going to answer all my questions.
I take that back: I care insofar as it then becomes a fun game to try to think up new questions that the machine can’t answer.
A while ago I was reading a book called Renaissance Curiousa, by Wayne Shumaker (a pretty great book, I might add, sort of a showcase of four of the craziest things people thought up during the Renaissance) and one of the chapters was about a guy named George Dalgarno, who had made a pretty fair start on inventing an artificial universal language — think something along the lines of Esperanto, except that in Dalgarno’s case, the language was built around what he understood as the essential categories of being. I’m speaking very loosely here, and the real examples are too weird, but the idea is, say, you decide that physical objects will start with the letter “s,” and objects that resemble other objects will follow the “s” with a “k,” and “u” will mean wood, and “tt” will be for electronic objects, and objects that make noise will end with with “ou,” and so a fake-wood-grain speaker is a “skuttou.” Except of course the categories that George Dalgarno chose were not the categories of my example, but were instead a bunch of bizarre and incomprehensible categories that probably seemed totally natural to people in the Renaissance.
If you’re making up a language like Dalgarno’s, you don’t actually have to write down every word in it; you just describe your various categories and the rules for combining and recombining them to make everything in existence, and then the reader can deduce the word for any particular object, including (always supposing your rules are sound) objects that no one has ever yet had occasion to refer to. That is, if Dalgarno’s system had been perfect, he would have been able, back in 1661, to offer words for “quark,” “courtesy wash,” “credit default swap,” and the like.
He never completed his system, but I think those words would have stumped him no matter how far he got.
I had fun thinking up those stumping words for a minute or two. Interestingly, most of the words on urban dictionary don’t work; they probably had a perfectly good word for “metrosexual” back in 1661.
The point is that the impulse of Wolfram Alpha (and Google, for that matter) to systematize all of human knowledge is not a new one. Also not a particularly rare one; I just happened to have these two examples on my mind, but you can probably think of more. Whether it’s ever been or ever will be “successful” may not be an answerable question (although it could be a good question to give the thing once it goes online: “Are you successful?”). It seems to me that the model of knowledge on which Dalgarno based his language is one that was on its way out by 1661; maybe Wolfram Alpha will be a hit, but I do think the most productive feature of it will be that it defines what it doesn’t know — I think the body of knowledge that is systematizable is precisely the body of knowledge that is no longer interesting.
Comments (1)
In 2005 I got stalked for 2 miles by Mormons on bikes while I was carrying Wolfram’s “A New Kind of Science” from the library. One of the Mormons was sad because he hadn’t seen his family for a long time. Occasionally I’d say, “Look at this,” and point them to a picture from the book. But mostly they wanted to talk about the Book of Mormon. The best part out of the whole exercise was that I later found out Wolfram also provides a website that will generate ringtones for your phone:
http://tones.wolfram.com/