If I were arguing by analogy I would ask why we allow door locks. Door locks let people do bad things inside their houses. Yet we are all safer from criminals because of them, overall. And it also helps keep police honest about getting a warrant first before they disturb your house.
The counter argument I've heard to that is that there is no such thing as an unbreakable lock. If someone is doing something illegal in a building, the government can eventually get in. Strong encryption is effectively unbreakable, and basically free for anyone to use. Trying to make a faciliy even remotely as secure is going to take millions of dollars because you need to not only secure the lock, but the entire facility has to be hardened.
Strong encryption may be unbreakable in a mathematical sense but encrypted communications are not unbreakable in practice. Key extraction through side channels or compromise of communications by physical surveillance or evil maid attacks is practical. It just can't be done to a billion people at the same time. Read here https://github.com/maqp/tfc/wiki/Threat-model for an overview of all the ways an all-out, unusually secure encryption system (encryption and decryption done on separate computers with data diodes, 100% reproducible build) can be broken.
True. I would say that the government can still compel individuals to surrender access to encrypted files with physical force.
I would also say that, similarly to how you can’t do all your criminal work inside a self-destructing safe, you can’t do all your criminal work in one e2e chat. Data will still be entering and exiting it somewhere, and if multiple people are involved, the relationships between them are weak points. But that’s getting too into the weeds. I think talking about something as simple as a lock would reframe the discussion for all but the most sycophantic.
