Large organizations are terrible at keeping secrets. Even clever people make stupid mistakes from time to time. Even patriots defect from time to time.
I would say that defection at this point is basically part and parcel to being defined as a 'patriot'. That's a personal view of course, and based on my understanding of the nature of our country (the US.)
Relying on secrecy in order to stay ahead in security is a bad strategy. The more experts you recruit into your research team, the higher the risk of leaks. All it takes is one rogue employee in 100. The fewer experts you recruit, the less competitive you become compared to the worldwide open research community in academia. Either way you lose.
Yes. The only silver lining to the massive dragnet and billion dollar datacenters is that casting a wide net yields lot so fish. If anyone is eventually called to account for their past malfeasance it will likely be due to the existence of this resource. But it is far from a certainty that this happy outcome will ever be realized.
I looked around once a few years ago to see if there was some sort of open-source intelligence platform. Something like just interested participants inputting data about observations, and other participants writing code to analyze the data. Something distantly related to Wikileaks in some ways. Didn't find anything, and I 'm not sure it is terribly viable since the cost of attacking it (with bogus or crafted input) is probably orders of magnitude less than that of defending. Asymmetric warfare going the wrong way (by my point of view.)