[Gustomaximus]

I think you can break it down to:

1) Price

2) Reliability

3) Self driving

The first 2 are already standard.

Self driving will be a mix of capability and safety record. The latter being more important I suspect. There will be endless websites reviewing which one is the safest. And when you choose a robot to let you drive that is going to be a massive part of the equation. Even if accidents are extremely rare across all platforms, social media will make accidents feel far more common than they are + they will likely be for stupid 'human avoidable' error. This will drive fear, one of the most powerful marketing tools out there.

[asdff]

I fully expect self driving to be regulated to the point where it's more or less adaptive cruise control. At which point it will be as ho hum on the feature list as cruise control today. Because lets be honest, this is mission critical software being written by bleary eyed, sleep deprived kids fresh out of school, ultimately, and these companies would prefer to keep it this way in order to "move fast and break things" and maximize profit. In this case, I'd rather not be the one broken. Someone else can beta test your for profit software and be the unfortunate name in the next tragic newspaper article.

As self driving becomes commonplace we will see deaths go up both actual, and the coverage around deaths as you note. Building safer software that receives input from a chaotic, random, ever changing, heterogeneous world is a much harder problem than a lawmaker stepping up to regulate self driving and becoming the public hero of the story in the process. Personally, I also think there are more important problems for these engineering minds to be working on.