You no longer need to plan and manage large capital expenditures, so you do not need a large bureaucracy. What I expect the Internet to generate is an increase in the number of small businesses, and a reduction in the proportion of the population employed in large enterprises.
I have been looking through some of my early essays in order to try to trace my gradual evolution toward Austrian economics. I did not really learn any actual Austrian economics (other than the theory of capital as roundabout production) until a few years ago, and I am still mostly illiterate in the great works on the subject.
What I had going for me was an intuitive sense that trial-and-error learning and bureaucratic planning are two ways to handle innovation in a complex world. The large organizational planning process, which I termed the Dilbert sector, tries to filter out error in advance through analysis and group politics. The trial-and-error process is decentralized.
I also had a sense that people have different models of reality. I called this Halberstam's Law, because it is one of the lessons that I drew from his book The Best and the Brightest.
Somehow, instead of internalizing the strong assumptions about human cognition that are implicit in the mainstream economics that I was taught at MIT, I have consistently been intrigued by human capacity for error and misunderstanding. I believe that is part of the foundation for Hayek's thought as well, which may explain how I ended up moving in the Austrian direction.