No. “Baking in” is a term of art; I think maybe I invented it. I'll claim credit in any event, even if I didn't. So for a year there would be, as you say, two corpuses of data, the public data to which the blur is applied, and the data held internally. At the end of a year, post-publication, there is no longer any data that has unblurred imagery in it. The same data that Google has is the imagery that's public, which is including the blur. I understand that's been done for the Canadian imagery.
On November 25th, 2010. See this statement in context.