Doug Merritt says:
Counterexamples to Joy's thesis are trivially found in cryptography, and less obviously in things like computer generated holography. The latter might easily become a GUI standard of the future, and will indeed require VAX-MIPS-millenia of computation to compute in real time.
They would also require similarly astronomical amounts of bandwidth to transmit. By today's standards, that's ridiculous to assume. But by the standards of 10 years hence, two dimensional video may well appear as primitive as 110 baud text transmission does to us today.
Judging the future by today's standards tends to leave one's predictions high and dry.
Three dimensional video would not require bandwidths that would make fiber optics wince -- even at the limits of human perceptional capacities. (Holography encapsulates the three dimensional image in a very high resolution piece of two-dimensional film. A digital analog would only require a large boost in resolution -- large by our standards, but not large by the standards of the bandwidth of fiber optic cable.) Given that even your best scenario for "expensive big application" comes up short, I don't see what the problem will be. I can easily envision what I would do with a computer ten thousand times more powerful than the one I have now. I can't see what I could do with a communication channel ten thousand times wider than what a bunch of fiber optics can in theory give me. Perry