Too many of the problems we've run into in the past 10 years in IT have been grounded in assumptions about what "sensible people" will do. That isn't what pushes the state of the art.
What he said. Nobody will do it and then think logarithmic growth from that point until everyone does it.