[sofon]

Interesting thread.

I think self-driving cars are short term doomed, and we've doomed them.

The problem is we view them as a purely technical problem where-as the real issues are sociological. It's not possible to create a self-driving car that acts perfectly on roads designed for humans (arguably, it's not possible to have humans acts perfectly either).

There are two solutions to this problem. Either shift human expectations such that they accept the failures of self-driving cars, or change the roads to make it easier for automated cars to drive on them.

I believe we should overwhelmingly be doing the second. We should be augmenting and instrumenting our road infrastructure to make self-driving easier. And in particular, we should be concentrating our efforts where this is easier, and has the most value (probably transit).

[protomyth]

Infrastructure spending is problematic. Heck, many states cannot even keep their bridges in good repair. I get the feeling that to enhance the roads, big tech needs to come up with a standard and finance it’s inclusion when major road work is done. Otherwise, I don’t see a way forward with augmented roads.

[sofon]

Yes, or some startup figures out that they can start doing it and charge the tech companies for access to their platform.

An open standard would be better of course.

[protomyth]

An open standard is easier for lawmakers since they just quote the standard document.

[Nition]

Possibly true, but some of the biggest hurdles to self driving are unexpected road conditions. Debris on the road, people stepping out into traffic, temporary roadworks etc.

[coatmatter]

I believe the biggest hurdle is that the bar has been set too low for what constitutes an acceptable standard of driving/casualty rate. There's so much that can be done to improve the situation but few answers are politically palatable (except for Netherlands, Sweden, and a few other countries around there, I believe). Car owners also seem to like driving, despite being stuck in traffic Mon-Sat.

I also think people will soon realise the limitations of trying to implement too much of this kind of tech as an answer to solving political and social problems. I'm not saying that the tech will never be good enough, but right now it's hard to see when that future will arrive in time before other pressing matters take hold.

However, what I do like about the self-driving car "movement" is that it's forcing more people to ask questions about what is acceptable when it comes to trauma rates. But for the time being, I see a lot of risk compensation ahead and fiddling around the edges.

More can be done a lot faster for less money if we instead were to focus all our efforts on reducing the population's overall dependency on private vehicle use and ownership. But in many parts, there's just too much money tied up in motordom with powerful parties are at play - the same parties who decide where to place big infrastructure spends and tax concessions. After all, always making new cars is one way to drive the economy - but at what externalised costs?

[CardenB]

That’s actually not the case. It seems the biggest hurdles are actually quite expected at this point, and enumerated by the article.

Self driving cars are actually quite good at spotting something unexpected with a lot of diligence.