[upofadown]

So we have gotten to actual communication and no one has checked the "safety numbers" ... which is perfectly fine. What isn't fine is that the user is left unaware that they are implicitly trusting Signal with their privacy.

I think that we need a sort of "truth in advertising" principle for these encrypted messaging things that claim end to end encryption. It is perfectly OK to trust an entity like Signal somewhat, particularly when the alternative is trusting an entity like Facebook. Just be honest enough to indicate that a major feature is not effective.

[StavrosK]

How is this different from literally every single messaging service out there? Having to verify safety numbers out-of-band is an inescapable physical truth, and one which doesn't really matter unless you're worried about active attacks against you.

[upofadown]

Active attacks that are really trivial to perform in the case of something with a single server controlled by a single entity. Most of these things take that form.

>Having to verify safety numbers out-of-band is an inescapable physical truth...

The reasons for this are fairly opaque to most people. It is kind of dishonest to just assume that people will do the checks for no real reason. You have to at least allow the user to understand the tool.

Note that Signal made it easier to ignore the change of a safety number a few years back. So in a sense they are getting worse at addressing the problem.