That said, I still consider an "active" steering decision by Autopilot to go outside of a lane a type 2) event, separate from "passive" events like the Joshua Brown incident, fire trucks - and others where lane simply ends at an obstacle - where mere predictable lane following have resulted in disaster through no fault of Autopilot... that I categorized as type 1) above. The Mountain View event certainly seems to be an active event where Autopilot failed.
There is a high dependence on the exact sequence of events/ positioning of vehicles in the Mountain View accident.
With a following distance of 1, the worn right gore line, and a leading vehicle, it is questionable whether what AP could see and what it crossed met the criteria of a lane (realizing of course that there was a lane line which is discernable with ground truth data or sufficient field of view).
Side thought/ realization: at T-8, the car was going 65 and following a lead vehicle at T-3 it was reported to be accelerating from 62 MPH. That means the lead car was either going slower than 65 (and the Tesla slowed to match), or the lead car slowed down in that time span.
Possibly, the lead vehicle made a maneuver that necessitated a lower than limit speed and/ or the Tesla was closing on the vehicle, in which case the following distance may have been even less...
Made up senario:
Tesla following vehicle in 101 fast lane.
Lead vehicle slows to cross gore
AP slows but still gets closer to lead vehicle, this reduces camera field of view of the road.
AP now follows lead car to the left over the worn right gore line (limited view).
AP sees the left gore line and stops following the lead car.
Lead car continues on 85 offramp.