The point about sampling as it relates to the go survey seems dubious, as the survey is not a random sample, but instead represents those who are most motivated to comment. It seems likely that that population would be more likely to want changes.

Same thing for opting in telemetry.