While the other replies here have some very good ideas, there might be another reason for this (perceived? real?) difference that is far simpler: running a weba

While the other replies here have some very good ideas, there might be another reason for this (perceived? real?) difference that is far simpler: running a webapp on the scale of Obamacare/NHS is probably a much harder job than tapping the internet backbone. For the NSA/GCHQ/etc, the actual engineering task of setting up some DPI boxes, beam-splitters, and other wiretap equipment is not particularly complicated on the engineering side[1]. The difficulty with that kind of project is access and legality. Even their massive DB in Utah, while (very) large, is probably a fairly simple schema compared to a many business DBs. Compare that to a healthcare website, that has to implement many random features mandated by law in a way that is compatible with a giant list of "standards", medical jargons, insurance requirements of all sorts, financial regulations, to name just a few of the details. And it has to work, because lives are going to rely on it. While this kind of website might be simple in concept, the random, contradictory, poorly-specified, "design by committee" project requirements are going to be nasty. [1] crypto may be different, but very little of what the NSA is currently doing has anything to do with crypto. They don't even need side-channel attacks when it's all in plaintext... sigh