Mandatory metadata retention forces corporations to collect data that is of little use on its own, the goal is that by combining all this useless information the government can make something useful out of it.
Its like a needle in a haystack, if they keep shovelling more hay on the haystack, maybe they will accidentally add another needle to the haystack, which they can then search for. They cant look for the needles before they put it on the haystack, because its someone else’s job to define what a needle is, and they might change their definition at a later date.
The problem for them is the sorters are expensive, and every time they change their definition of what a needle is, they have to start again. An extreme example of this is utah data center which sorts for needles as defined by the US government. It uses 65MW of electricity per year and millions of gallons of water per month. They sort a haystack of between 3 and 12 exabytes, which is between 416MB and 1.25GB for every person on the planet (before compression).
Government judges everything based on cost and this new surveillance system will be expensive, even if they push the collection and storage costs onto “service providers” they still need to do the sorting themselves (and thats more expensive than storage).
The costs of sorting can be influenced by us, the users, the more metadata we generate, the more infeasible the whole operation becomes.
Maybe we should be looking to start/promote projects that generate lots of metadata per user ?
One idea that comes to mind is a distributed search engine. So end users slowly crawling through the web in a coordinated way and pool the results, the person doing the crawler gains increased privacy by adding noise to the system, and there is a useful result in the end. An independent search engine also gives control back to users as big brother can force corporations to remove items from their index, but maybe not from a p2p system. (a p2p system could choose to remove items, but it would be more transparent)
EDIT: Trokam looks interesting
Maybe there are more practical ways to generate metadata, but just throwing that out there.