We need to centralize around protocols/standards and decentralize ownership of data. If somehow each piece of data on a user were exponentially more expensive to keep, that might be possible. But that would be a very strange law…
Dave and I agree with your diagnosis, and we would like to propose a solution. Who is Dave? Well you know him, and I know that you know him and I know that when we both talk about Dave we must be talking about the same person.
(mLuby I probably don't really know you, this is just a hypothetical).
A casual observer could not approach our common understanding by adding more data. There's so many Daves after all. In fact they'd need to subtract all but the right data.
So the solution is a protocol in which increased data adds noise faster than it adds signal. Such a protocol requires 1000s times more plausible yet incorrect noise for every signal. Digital chaff.
Are you saying that nobody can know which Dave we're talking about unless we identify him as Dave-with-cell-5551234567? I don't think that's true, since human social circles are pretty easy to figure out; see the humorous hypothetical collection of data about US Founding Fathers. http://www.newenglandhistoricalsociety.com/phone-spying-paul...
there are a few other things too like only being charged to upload files once vs paying a subscription for the rest of your life, or only needing one password for all sites.