[munk-a]

I think there's an epistemological issue in your statement. When a human does it we say "what an idiot" because the driver is performing at a level below the generally accepted proficiency for driving. I think our reasonable expectations for autonomous driving is around an average level of proficiency. I also don't think it's reasonable to delay implementing technology until it's better than the best of us - but this was an utter failure and is not within the bounds of acceptability.

I do think it's fair to argue that this is probably an oversight that can be corrected now that it has been revealed, though.

[arcfour]

But in many ways, self-driving cars are better than or equal to the average driver. If I make a mistake once but am otherwise an exemplary driver, am I a bad driver? Same goes with these. The question isn't are they perfect (which is unfair), it's whether they are in aggregate safer than humans, which is objectively true, but people making big examples of issues like this serve only to muddy the water and scare the public needlessly.

So, in the interests of avoiding needless regulation that would make us less safe, I think it's important to point out that these comparisons are unfair & examples are typically extremely rare edge cases.

[munk-a]

I agree that in many ways self-driving cars are better than average. And I selfishly want to accelerate the adoption due to the fact that I'll appreciate it the most when other drivers are using it to reduce the chances to getting hit by a drunk or highly distracted driver.

However, I think that driving on tram tracks is unacceptably bad - it is something that a good driver would simply never do outside of really strange circumstances (like complete vision loss due to heavy storm weather). This example shouldn't be used as a single example to bar autonomous vehicles but it should also be properly recognized as unacceptable to be repeated.