I heard a fascinating presentation tonight by Benjamin Reeve about the impoverished nature of our understanding of information.  It was an important talk. 

The thesis (I'm waiting for the PDF of the manuscript so that I can post a link here) is that Shannon's understanding of information is not helpful -- that information is really differences that "make a difference" by causing a system to change its state.  Informational things have different qualities than physical things (you can't run an algorithm against a hill).  Most fundamentally, information is not conserved.  Information, instead, amplifies.  But amplification, just for its own sake, isn't an unmixed good.  Instead, what we should be interested in (and design for) is metainformational depth -- quality information about information.

The most important thing we can do is "make things deep" -- build for better information.  Maybe that means tagging data and having it report back to us about how it's doing.  At any rate, for any amount of amplification we create, we need to create MORE metainformational depth.

Stay tuned -- this was a terrific talk, and when I have some links I'll send them along.  If we get better at handling information, we can get better at building society.