You mentioned a few times in the paper the dotcoms with their hockey-stick projections, and how they frequently went bust. Most of the money going into the boom was chasing the idea -- promoted heavily by the geeks with the "business plans" -- that this "internet thing" was something new in the world of business. To the extent that they were relying on Metcalf's Law, they were overlooking something that any MBA will probably tell you he had figured out by junior high: popularity matters.

Oh sure, we can dress it up in all the academic language we want: social network, affinity, gravity law, information locality. But Zipf's Law is the one that really touches on the heart of it, popularity. Some connectoins are more popular than others.

Metcalf, in describing the value of a connection, did what any good geek would do. He used the term "value" as synonymous with "utility". A network with more connections is more useful than one with fewer connections. But "value" is a measure of desire. How much does someone want something. A connection that is valuable to you may be worthless to me.

Maybe we can quantify how important popularity is. But is it actually a useul predictor of anything?