> At the very least, this would need some quantification of risk
Yes, of course.
And in the case of autoupdates:
1. the risk of finding a RCE in any random PC within the next few years is close to 100%
2. an unpatched RCE is strictly worse than a backdoor
3. how many computers won't be patched without forced autoupdates?
In the case of e2ee, I'm afraid it's much harder to quantify, though.