I am in support of the census package that Canonical is trying out. We could do with more public access to the data set the package will create, but that’s not really a policy of the package. That’s a Canonical policy thing.
Having a fully public data set would of course prove beyond doubt about the anonymity of the data being saved.