They really need to think this through first
While I think such legislation is a fantastic idea, and in good keeping with Sir Tim Berners-Lee's views on data silos and walled gardens, as a programmer myself I can see a problem that needs to be addressed if this is to work.
That is, we need an W3C-style standard for exporting or importing this data. Obviously some kind of XML is the way to go, but we need some standardised labelling system within that XML. For example, how do I, as a web developer, know which part of a data block exported from Facebook is the user's first name, last name, phone number, email, post content and so on?
For example, if say Facebook do it like this:
<FirstName>Steven</FirstName>
<LastName>Roper</LastName>
and Google+ do it like this:
<person_Name>Steven Roper</person_Name>
then we're going to have the devil's own time trying to decipher cryptically named XML files from different social networks.
So there really needs to be a defined, global standard for describing social network data. The W3C of course is the best body to address this, but the issue does need to be addressed before any such laws are enacted.