"So politicians and intelligence services calling for encryption want, institutionally, to keep people safe. How can tech companies do that without breaking or backdooring encryption? That's the real problem to solve, and the first person to figure out how to do that will be way ahead."

That's the problem: they can't. The tech just moves information around in ways that are visible and can be invisible. Even the subtle timing of the signal can be used to communicate. What you're asking is that tech companies do two things:

(a) Ensure every tech with communication mechanisms can only work in a way known perfectly ahead of time and controlled to prevent all leaks.

(b) Ensure that every message, regardless of its words or properties, isn't aiding evil somewhere.

(c) Ensure that it's impossible for a user to put obstacles between their identity and a communication tech so we know exactly who the evil-doer was.

That's simply impossible. It can't be done at all. Talking about lay terms, North Korea has the most totally controlled Internet and country locked down by both technology and terror. Possession of a cell phone can get you locked up and leaving can get them after your family. Yet, there's regularly communication from small numbers of courageous people coming out of that country. If that ideal surveillance setup won't work, how on Earth could any tech accomplish the same thing with the diversity of sites, language, interactions, and tech that exist in Western countries like America?

It just can't because it can see bits and read the minds/hearts of those that create them. That's what the government wants. It's not possible. Otherwise, they'd probably have eliminated most crime already looking at people's homework in school or text messages divining their character. ;)

Also, the problem's been known so long there's even a meme for it:

https://en.wikipedia.org/wiki/Evil_bit