A recent post on Jonathan Schwartz's Weblog : The Glamor in Mass Transit (?) basically talks about the efficiencies of scale. Sun computer, which Schwartz leads, has long argued the network is the computer, and has been trying to move intelligence back from the decentralized desktop into the highly centralized information technology control center.
The comparison between mass vs. private transportation and mass vs. private computing is worth noting. Everyone has a mental image of mass transit, though that image varies by individuals, some who see it is valuable, some as something they would not touch. The perception of course is shaped by their individual experiences, preferences, locations, and so on. Mass transit is efficient in certain specific contexts, but it requires users give up an element of freedom and (in general) spend more time traveling.
The personal computer revolution of the late 1970s and early 1980s enabled individuals to have control over their computing environment, without relying on a third party to provide that service. This provided freedom (I can write my own programs, and run them when I want in real-time, not having to go down to the computer center and load my program on to the Cyber at Rich Hall at Georgia Tech, and wait 20 minutes for the output to be printed (in below zero F temperatures, really, in Atlanta, January 1985, you can look it up .. Reagan's second innaugural was delayed by the same coldfront) so that I can do a homework problem for Professor Betamax's Fortran class (the course was videotaped, and was replayed every hour, so we could attend when we wanted). I hear horror stories of people who work in controlled environments like Lotus Notes, where they can't deal with email conveniently but require using a browser with a sluggish email program behind it.
So as much as we might curse personal computers, or cars, freedom of action is what they provide.
A good mass transit system, like the so-called web 2.0, can provide the same freedom through its ubiquity, and free the user from the need to manage complex systems (automobiles, computers), focusing only on the higher level decision (what I want to do, what I want to say, where I want to go). But it builds in additional dependencies (will the internet be up? will I have to pay to get access in a hotel room to my data? will the bus show up on time? does the bus really go there? what will google do with my data? do I want to see personalized ads based on my research paper on transportation?).
Freedom from and Freedom to are important distinctions. I would much rather have Freedom to act than freedom from cost or risk of acting.
What should be owned and what should be rented or provided as a service is one of those essentially tug-of-wars that shape every aspect of the modern economy. There must be some economies of scale, or we would not see scale, but there must also be diseconomies, and loss of freedom is one of them.