[AnotherUnixGeek]

How are the potentially driverless cars doing in their testing? Awful. For example, in the first week of March, Uber’s 43 test cars in three states logged some 20,000 miles on public roads. Their drivers had to intervene and take control away from the software, an average of once every mile.

An incredibly complex new technology doesn't work flawlessly right out of the gate and needs further development? Shocking, just shocking - must be doomed to failure, let's call the whole thing off.

Betting against technological advancement is a fool's bet. The bugs will be worked out and driverless cars will become a reality in a decade or two.

[AndyTheBear]

I was a part of the GPS map revolution, which is a much simpler technical problem. The first working nav system was the Etak developed in the 85. But the hard technical problem was good digital turn-by-turn road maps. But after the industry worked on them for 30 years, and had engineers even as brilliant as me helping, there is no way to even get the map data to be 100% reliable over all the area one may drive so that it is safe to trust not to suggest something unsafe.

But happily for maps you don't have to trust them 100%. You have a driver who is supposed to use common sense.

The self driving stuff will keep improving over the next decades I am sure. But getting them to the point they can safely be driver-less in 20 years seems about as likely as flying cars being common in the year 2000 (which didn't happen btw). My guess is we will get more enhanced breaking/cruise control/driver supplemental type stuff over the next 20 years.

I don't bet against technology...just saying what it produces and when it produces it will frustrate intuition. The devil is in the details.