Ed Felten raises a very important point about many of the debates we have about data portability: we start off by making a poor assumption, and that closes off options.
An example is the Internet storm over Facebook canceling well-known blogger Robert Scoble's account. Scoble had amassed a vast amount of data in his account, and got caught using software tools to export it. Facebook has a vested interest in locking people into their service (more users = more advertising revenue), and the way they have chosen to do this is to give people free accounts, encourage them to invest a lot of time creating valuable (to the users, if not anyone else) data, but prohibit them from extracting that data elsewhere.
Hmmm... I must update my Blogger backup script. It hasn't worked well since Google made the upgrade from Blogger version 1 to version 2.
The poor assumption that we make is that data -- facts -- must be owned by somebody. As Felten says:
Where did we get this idea that facts about the world must be owned by somebody? Stop and consider that question for a minute, and you’ll see that ownership is a lousy way to think about this issue. In fact, much of the confusion we see stems from the unexamined assumption that the facts in question are owned.
Once we give up the idea that the fact of Robert Scoble’s friendship with (say) Lee Aase, or the fact that that friendship has been memorialized on Facebook, has to be somebody’s exclusive property, we can see things more clearly. Scoble and Aase both have an interest in the facts of their Facebook-friendship and their real friendship (if any). Facebook has an interest in how its computer systems are used, but Scoble and Aase also have an interest in being able to access Facebook’s systems. Even you and I have an interest here, though probably not so strong as the others, in knowing whether Scoble and Aase are Facebook-friends.
How can all of these interests best be balanced in principle? What rights do Scoble, Aase, and Facebook have under existing law? What should public policy says about data access? All of these are difficult questions whose answers we should debate. Declaring these facts to be property doesn’t resolve the debate — all it does is rule out solutions that might turn out to be the best.
UPDATE: Chris Finke has an innovative solution to the Facebook problem, one which could (in principle) be extended to all similar such websites. His Facebook Scavenger extension for Firefox lets you capture copies of the data once it's in your browser.