These are interesting, and clever, and I agree difficult, but I don't think they necessarily need to slow down the progress of self-driving cars much at all. T

These are interesting, and clever, and I agree difficult, but I don't think they necessarily need to slow down the progress of self-driving cars much at all. Tesla's data so far suggests that their current autopilot implementation is reducing crashes by 40% - http://www.theverge.com/2017/1/19/14326258/teslas-crash-rate-dropped-40-percent-after-autopilot-was-installed-feds-say http://www.theverge.com/2017/1/19/14326258/teslas-crash-rate... - and while these cases are all problematic, they're all fairly rare, and the cases in which they come up and the car reacts wrongly and that's a major problem are going to be even rarer. On top of that, self-driving car performance is only going to improve. It doesn't really matter (in terms of the value of self-driving cars - there's a mostly independent marketing question) if there's a couple of cases where self-driving cars make the wrong choice and kill you, if there are many thousands of cases where they save your life. It's effectively changed a great many risky situations into some new and different but less likely ones. If driving a self-driving car significantly increases my odds of survival while driving, in addition to giving me huge amounts of bonus free time, then I'm definitely interested, regardless of risky but rare cases like this.