I was having a good conversation last night with folks who know, and there is a reason I say "unsupported by any real data." No one has any reliable, replicatable data on the nature of internet traffic. The companies transporting the data keep it quiet for a variety of reasons -- concerns over proprietary data, concerns over privacy, fear that an ability to monitor for purposes of data gathering will triggering responsibilities to police content, and other reasons ranging from good to bad.
But the upshot is that estimates about the quantities of data moved across networks and the nature of what applications are used are at best SWAGs (scientific wild ass guess) and at worst outright deceptions.
From a public policy and economics of the free market perceptive, that is a disaster waiting to happen. Let me illustrate with one example from the market. A good deal of the internet bubble valuation of carriers, and subsequent collapse, devaluation of carriers, loss of billions of dollars, overbuilding of capacity (now eliminated by increased volume) derived from false statements made by carriers that were unconfirmable. UUNET used to boast that traffic was doubling every 60 days or something like that. Everyone in the industry wondered how it was possible, but they (and Worldcom) were saying it and they were one of the major backbone players. Build outs and other economic decisions were made on the basis of these statements that were unconfirmable, and subsequent proved to be false. As in, they were known at the time to be lies, and were made for the purpose of boosting the value of Worldcom's stock (this is one of the reasons Mr. Ebbers, former head of Worldcom, is spending an extended vacation at the Greybar Hotel).
Not having data sucks rocks. As always, the question is what to do with imperfect data. Suggests folks study game theory before suggesting what appears to be obvious answers, btw.