This is really interesting. I know the old "Attacks always get better; they never get worse." but I've never thought about "Tools and technology always get cheaper and easier to use" before as it applies to security.
The central dynamic internal to government is, and always
has been, that the only way for either the Executive or the Legislature
to control the many sub-units of government is by way of how much
money they can hand out.
...
Suppose, however, that surveillance becomes too cheap to meter,
that is to say too cheap to limit through budgetary processes. Does
that lessen the power of the Legislature more, or the power of the
Executive more? I think that ever-cheaper surveillance substantially
changes the balance of power in favor of the Executive and away
from the Legislature. While President Obama was referring to
something else when he said "I've Got A Pen And I've Got A Phone,"
he was speaking to exactly this idea -- things that need no
appropriations are outside the system of checks and balances.
[1] https://www.youtube.com/watch?v=nT-TGvYOBpI#t=625 http://geer.tinho.net/geer.blackhat.6viii14.txt
Now however there is at least one cop in every home, or at the very least a snitch. This thanks to internet connected computing devices.
Privacy is dead also cuts both ways. Say goodbye to undercover policing.
Technology reduces costs. Especially attack costs.
Extrapolating out to a population of 300M that gives you a yearly cost of about $8.5M, which appears to be the same cost as building 2.5 miles of freeway in California.
Basically there's no longer an economic barrier to storing almost unlimited amounts of metadata for individuals. The fact is we just don't generate that much. And the costs will only decline over time.
(Would love it if somebody could run the same numbers and see if the result is replicated.)
So, you mean cost per year?
Note that S3 is extremely expensive. Pure storage media cost with hard disks is around 0.001 USD per GB-month unreplicated--and replication might not necessarily be a priority, depending on the goals one has with that data collection.
Also, that would give you about 1 m precision at one second intervals, which might be excessive (or not).
Really, there is no longer an economic barrier to storing "real" (as in "non-meta") data. 24/7 audio recording in telephone quality with ten year retention would cost about 4 USD/year in pure storage cost at current prices per person. Or in other words: storing all phone calls forever doesn't really cost anything. Actually, it's so cheap one can expect that the NSA is already doing it.