@wdormann As I understand they "knowing why" (as of now) doesn't imply this was *expected* behavior before.
I'd compare the persistent (not self-deleting) messages dilemma to secure deletion: below the next architectural boundary you can't really decide what's happening to your data ("were the bits of that file really deleted from the disk?"), but in special cases you take extra steps to prevent leaks ("let's overwrite a bunch of times, hopefully it helps").
@Mer__edith