That said, I don't think LIDAR is key. I fully agree with Elon - if you're going to have a transmit-and-receive system, why would you double up on the same spectrum as your cameras, and face that spectrum's same limitations? Radar is the way to go. Sure, the radar spectrum has its own weaknesses, but they're different weaknesses from the optical spectrum.
I was at an autonomous vehicles conference over the summer--there was a speaker there from Sony who challenged the conventional wisdom for the need for Lidar. He proposed video, specifically next-gen sensors, being a better long-term sensor strategy. Granted that it was a bit self-serving, but at the time, I thought that he did make some good points and aligned with Elon's prior comments.
One of the interesting points he made was that driving is currently built around visual input (street signs, lane markers, traffic cones, school bus lights, etc) so any sensor strategy should take that into account.