How do these things even exist? Obviously the people in the cars don't want them, and will never use them. Even the people A/B testing them should eventually figure that out.
I suppose it's the "better horse" paradox. A/B testing can only get you incremental improvements, to make the leap from a really good horse to a car, you need to make a bet and sustain it for a bit to get people comfortable enough to decide to try it out.
That's not to say that whatever Waze is doing is a good idea, but that may well be the underlying mechanism for why they're trying.
