> One of the things the article names is the challenges caused by snow, ice, etc. I realize I'm in the minority, but I actually think there's lots of conditions

> One of the things the article names is the challenges caused by snow, ice, etc. I realize I'm in the minority, but I actually think there's lots of conditions that we shouldn't be driving in -- conditions where the obscured lines of sight, slipperiness of the road, etc, actually mean that we should either be off the road entirely, or traveling at speeds far far below normal. I don't think snow, ice and rain will be as problematic as predicted for self-driving cars. Altogether I expect them to perform much better than humans - more so than in clear conditions. Self-driving cars have more sensors than humans - lidar, radar, cameras, possibly at multiple wavelengths, which should let them see through conditions which are difficult for us. And they can control braking, acceleration and perhaps steering on each wheel separately and faster than us, allowing better control in slippery conditions. --- But what I expect the fundamental problem to be is risk vs speed trade-offs. Humans routinely drive beyond safe speeds in all conditions. People tailgate and drive close to the speed limit in snow storms, in a way where stopping in time is impossible if the car in front brakes hard or an obstruction appears from the blur. People drive the city speed limit next to parked cars, unable to stop in time if a child steps out between them. People drive the city speed limit across blind intersections if the crossing traffic has stop signs, assuming that the crossing traffic will give way (but emergency vehicles don't have to give way, there was a bad crash just like this a few months ago in my city). All these are relying on the prior probability of a dangerous situation occurring being low enough to drive at "beyond safe" limits. So self-driving cars will have to choose to match the human behaviour, and accept occasional crashes - or appear infuriatingly slow.