From Tim Bray's weblog, ongoing:
Bob Metcalfe, that is, who regularly predicted that the Internet would collapse, and famously ate his words—literally, he ground ’em up in a blender and ate ’em, I was in the room—at one of the Web conferences. Well, at the moment I can’t connect to AIM and I can’t connect to Google, and I’m getting a couple dozen echospam per hour (bounces from spam with my address forged) and even though I just turned on the computer after a couple hours downtime, NetNewsWire says “nothing new” which I flatly don’t believe. Probably just something else flailing away at the Wintel monoculture. But I don’t like it.
I remember a friend of mine who used to rag on Metcalfe back when he was making such seemingly-dire predictions. The friend has a tendency to describe anyone who makes an assertion he strongly disagrees with as "wearing underwear on his head." I don't think Bob was really in that category, though--he was making the prediction based on how inefficient the protocols were for increasingly high-bandwidth uses, and based on how that demand seemed to be growing faster than the bandwidth was.
The internet was saved from his prediction chiefly by what that other great technology pundit, Alan Greenspan, termed "irrational exuberance." To wit, telecom companies overspent so ferociously at the end of the 1990s building infrastructure that the available bandwidth ultimately far outstripped demand. This puts this in a particularly ironic light: the internet has remained basically functional and happy thanks to all those companies driving themselves out of business or to the brink of bankruptcy. Ultimately, the underlying protocols on the internet aren't going to change due to bandwidth concerns, but to authentication and security concerns.