Look, if a Princeton professor has trouble setting up network printing, then perhaps academia and industry can spare a little time to consider the problem more thoroughly?
What is odd is that at least part of the problem in this specific case should be solved already. We have zeroconf, right, that is supposed to enable the kind of auto discovery Felten expected? Apple, at least, has a reasonable implementation. I seem to recall they even effected a guerilla port to Windows, something those users have to either download manually or gets installed by Safari or more likely iTunes. Even Linux and BSD have an implementation available though I have no idea how common packages are in the more popular distros and flavors.
Actually, glancing through the Wikipedia entry it seems like everything is not as rosy as my OS X tinted glasses make out. There does appear to be some fragmentation and disagreement in how to implement this idea. There even appears to be some disagreement in the actual formulation of the specification itself. While searching for the spec’s web site, I even came across some tantalizing link titles that suggest that there may be enough overlap with other specs and technologies so as to sow some further confusion and/or competition in the space.
This leads me to believe that network printing is not inherently hard, though, unlike factoring primes on a classical binary computer. Rather because the space for network standards and printers is so competitive, consumer convenience is all too quickly sacrificed in the name of maintaining an edge. Printing would seem to be such a basic complement to computing that at least a certain core functionality would seem to supersede such petty squabbling. I guess not.