Was George Gilder Wrong?

I was reading a paper by Jim Gray (the same dude I was talking about in my Information Overload posting) called Distributed Computing Economics. He said, "Over the last 40 years, telecom prices have fallen much more slowly than any other information technology. If this situation changed, it could completely alter the arguments here. But there is no obvious sign [as of March 2003] of that occuring." This is a direct contradiction with Gilder's Law, which, as you'll recall from reading my handy-dandy List of Laws, claims that network bandwidth doubles every 9 months.

It now appears to me that storage doubles the fastest, followed by computation, with communication doubling coming in last. This is a purely subjective judgement. I say that because it seems like I have more storage than I could ever use (I'm using about 10% of my hard disk), and with my newest PC I have enough processing to handle video playback (one of the most CPU intensive tasks that I do), and because Jim Gray's arguments about where computation should be done (client vs server) fits the way I actually develop "real-world" software (I've written both kinds of software -- software that people download and install and software that runs centralized on a website). Looks like Gilder was wrong about more than just his investment advice. Perhaps it's time for a "Gray's Law"?

But here's the real mystery: what determines the doubling times? Why would storage, computation, and bandwidth double at different rates? Keep in mind, the doubling rates are averages over long periods of time -- in the short term, there is no smooth doubling curve, instead the "curve" is extremely choppy. Inventions and innovations that advance technology simply do not arrive in a predictable fashion. Also keep in mind that bandwidth cost is harder to measure (you have to account for distance, so metric is bytes per second per mile (or kilometer for those who use the logical measuring system) per dollar. And Jim Gray does not, in his paper, explain where he gets all the data for his conclusions either.