The problem with opt-in telemetry is that it messes with the scope of the research.
If you want to understand something about most users (and not just the ones that are active enough in the project to participate in opt-in) you need this, otherwise your results only tell the needs of this subset of your userbase and this sometimes can go completely against the needs of the majority of users.
The problem with telemetry isn’t the telemetry itself, is how it is used, and the way the proposal is worded makes me very optimistic. They are trying 200% hard to make sure we understand that it will never be used in violation of the users’ privacy.