You know, it's funny when people talk about job disruption from tech, one of their key platforms is that things are changing so much faster now than they used to be. That's one of the things that's supposed to be really different about 2017.
If you accept that as valid, then 20 years is a preposterously long timeline, and nobody could take seriously a prediction that far out. As an example -- in 20 years time, Hyperloop could easily become the dominant form of transportation, leaving people to shrug off as irrelevant the self-driving car stuff. Not saying that's going to happen, just saying that 20 years is so long, that level of change could happen in the same timeframe.
Me personally - in terms of the economy, available goods & services and such, I think you can take a really good guess 1 year out. 2 years out it's a bit fuzzy but still OK. 5 years out you've only got the broad outlines. 10 years out, you're just straight-up guessing.
20 years out? Unrecognizably different.
I accept it ("things are changing so much faster now than they used to be") as valid AND I don't find 20 years a "preposterously long timeline".
"Much faster than they used to be" is not instantaneous, it just means that in the ancient world it took millennia, in the industrial world centuries, in the electronics world decades, and in the internet world years.
But self-driving cars are more in the electronic's world than the internet world. It's not like they are some new website or web business that takes over in a couple of years or so.
They need engineering work, safety tests, government compliance, infrastructure, etc, which take decades.
>in 20 years time, Hyperloop could easily become the dominant form of transportation
No, it wouldn't.
>20 years out? Unrecognizably different.
Hardly. 20 years is a perfectly good horizon to make realistic predictions. 1996 to 2016 for example: more mobile, more internet (which tons of people rightly predicted) would be totally correct. Policy decisions work in even larger timelines than that. Only superficial crap wouldn't a 1996 be able to recognize about 2016.
Are 20 years a "near-term timeline"? Because all it takes is for self-driving cars to be approved for market, and today they are under heavy testing, and we even have some (Tesla etc) where while not "self-driving" they are very close.
It's not like some feature cpu/storage/battery/etc technology, where some lab breakthrough or another frequently gets reported in tech outlets, but it never arrives.
This is a mature field, with multiple multi-billion investments and heavy activity.