Hi-tech product management for cars
AC - upvoted, since my own experience with "smart driver assist" gadgetry matches yours.
It seems to me that the typical "technology startup" attitude of doing 80% of what is needed in 80% of most common cases is enough for a good enough product is woefully inadequate for making cars.
In "normal" situations any alerts or automatic actions are either not needed or not helpful. In abnormal - typically dangerous - situations a decent driver will be not just alert but tense, and any additional beep triggers (for me, at least) "Oh, shit, what else is wrong now?!?!", while struggling with a car that wants to do something else leads to a feeling of loss of control. I have yet to see an automatic driver assist or alert system not doing the worst thing possible in a hairy situation - but that's never in the 80% most common cases Product Management considered.
I could offer lots of examples from personal experiences. Just one: I had to drive a rented car with active lane assist (impossible to switch off) in Italy last autumn. I spent a couple of days driving on narrow mountain roads with clear lane markings. The car's sensors thought I was too close to the markings most of the time (I was - the roads were narrow) and "pushed back". I felt totally out of control with a wall on one side and an abyss on the other - not pleasant. On autostrade the system was all but non-existent - the lanes were wide enough and I wasn't asleep.
I remember MobilEye marketing their gadgetry as something that will keep you alert when you are tired behind the wheel (in the past, to be fair). If you are tired you are not supposed to be behind the wheel in the first place - are you saying it is OK now? This, IMHO, relates directly to the "autopilot+moron" use case.
Dear software product managers, don't even think of the "good path" until you've figured out all the cases where your shit may cause real harm. That's a lot more important in a car than in a smarphone app.
Biting the hand that feeds IT
