Earlier quoted context omitted.
> We should not be arguing that that nothing good comes out of surveillance. The problem is that we, the people, can never know what, if any, good is coming out of surveillance. Attorney General Barr admitted that in one of his speeches arguing for back doors in encryption. The government cannot reveal what is being discovered through surveillance without disclosing sources and methods that it (understandably) wants…
It's often not the the government "cannot" reveal those details (maybe not immediately and directly in some cases, sure, but certainly with the distance of time that tools such as FOIA requests require), but that they "won't" and have no interest to. It should be the public demand with each attempt to increase surveillance to increase oversight. Sousveillance (watching the watchers) is the best known defense we have…
Often that is way too much time--25 to 50 years in many cases, since those are the time frames for declassification of classified information--for such revelations to be useful for oversight, especially with the state of encryption as it is since computers and the Internet.
Before computers and the Internet, it was possible to have a reasonable tradeoff between strength of encryption and the ability of law enforcement to conduct surveillance, because perfect encryption was impossible and imperfect encryption got more expensive the closer you wanted it to be to perfect. So people were already making a cost-benefit tradeoff (difficulty of breaking the encryption and obtaining private data vs. cost), and it was reasonable for the government to ask that the potential benefits of surveillance be included in the tradeoff, since that would just adjust the balance of the tradeoff, and the adjustment could be periodically reviewed based on data on past surveillance that was revealed by things like FOIA requests.
But now, with computers and the Internet, perfect encryption is cheaper than imperfect encryption. Perfect encryption is just a mathematical algorithm, and it's straightforward to put that algorithm in computer code and verify that the code correctly executes the algorithm. Imperfect encryption requires adding code to that perfect algorithm, which adds cost, and also adds a risk that wasn't even there before, of whatever back doors are in the code being exploited. So now we users, to enable surveillance by law enforcement, would not be just making a small adjustment that could be periodically reviewed in a tradeoff we have to make anyway. We would be adding a new tradeoff that we have no other incentive to make, and thus taking on a new oversight burden, which is, if not impossible, at least extremely difficult to properly fulfill, that we have no other incentive to take on. That is simply not a bargain that free citizens of a free society should accept.
> embedding those checks/balances/required transparency in the surveillance processes in such a way that they cannot be circumvented by those in power.
The processes can't be transparent because, as I said, that would reveal sources and methods that should be concealed from adversaries. An application for a FISA warrant can't wait for the years it would take to allow a FOIA request to be fulfilled in the interest of transparency.