The number of leaks and breaches shows there's a serious problem, but keeping keys secret isn't necessarily implied to be near impossible by consequence of this.
"Trust us, we'll keep it secret" has been empirically proven to be not as true as they want us to believe.
Taken from the "other side", it does not seem universally true that generally deployed strong, unbreakable encryption built into "secure" general-purpose commodity hardware is in the best interests of humanity going forward. It seems to be an open question. It was nice to see rational/objective/neutral discourse on HN in the past that considered all sides. But, such a universal perspective seems to be missing of late, and the more recent parochial attitude seems a natural form of pushback, given the current chaos. Hopefully good comes of this.
"Snow Dawg" is currently partaking in thoughtful discussion arguing against NSA's policies on his twitter, if anyone is interested.
Can you prove that there is no such thing as a "perfect" backdoor? Can you show that the existence of a skeleton key introduces risk beyond losing the key? Has this been formally proven?
That might be a good starting point, and I apologize if my understanding is wrong, but can't one build a skeleton key into encryption that cannot be broken with any greater likelihood than otherwise would be possible by compromising the encryption itself? If the surface area of attack is doubled at most, that seems a viable trade-off. Yes, it's potentially a huge SPOF if designed sub-optimally (I'd suspect that there is a way to build something akin to a one-time use set of segregated skeleton keys), but that risk needs management like all risks.
(redact)