The "most dangerous" part is doing a lot of work there. Just like I think law enforcement needs to admit what they can and cannot do (e.g. they cannot protect a golden key), I think we need to admit some things too. A lot of dangerous criminals are stupid. Maybe not the most dangerous ones, sure. But if law enforcement has a tactic that lets them catch, say, the stupidest 30% of terrorists, that's an extremely valuable tactic that probably saves a lot of lives in practice. It would be wrong to claim that society loses nothing by engineering away that tactic.
I think this sort of thing leads to a lot of frustration on both sides. As a programmer, I find it very frustrating that law enforcement and the media consistently get some of the most basic details wrong about how communication and encryption work, and about the negative side effects of the new laws they're proposing. But I assume that law enforcement folks also feel frustrated about how people like me have no idea how they actually get their jobs done day-to-day, or the negative side effects of the technologies we're building.