Thanks for posting the sample interaction chart. Seeing how interactions vary over time for one individual driving with primarily the same level of competence and care can indeed tell us a story. Multiply this type of interaction chart over multiple individuals, comparing past versions of the software to present, could be useful. My concern is that we beta-testers differ tremendously in temperament, experience and driving environment. What happens, for example, when the number of beta-testers increases four-fold with the introduction of new beta-testers? I appreciate what you're trying to accomplish but I think disengagement reports would make Tesla look unfairly bad compared to some competitors that have trained safety drivers in the seats and who are more concerned about letting the vehicle prove itself than showing courtesy to those behind. There's too many variables at times to draw short-term conclusions. Over longer time, the data may be useful, but again I think it could be used against Tesla by equating every disengagement with a failure.
