> If it's really so simple to prove, why hasn't anyone done a carefully controlled experiment proving this once and for all?

They did, and found no listening being done. It’s in the article under “The data doesn't add up”.