Operationalizing this as follows: Will I personally, by 2025-12-31, call a car (rideshare, whatever) and have it show up to my house in Portland, Oregon with no one in the driver's seat?

1. Does this have to be open to the public?

Mostly, yes. Some kind of restricted beta is ok if I don't have to pull strings.

I have to be able to take a trip that's faster than walking and cheaper than renting a car. Low bars but if that much is possible then presumably practicality isn't far behind.

3. What if people own such cars but it's not available as a taxi/rideshare?

If I call a friend and they make the car show up sans driver, we'll count that. Again, as long as I'm not pulling strings to make it happen. The car can't be more expensive than, say, a top-end Lexus that a (somewhat) normal person might buy. If I've bought such a car myself but I'm the only one I know, that doesn't count. 

So mostly this is about taxis/rideshares but in the unlikely event that private ownership of fully self-driving cars becomes somewhat common first, we'll count that as sufficiently in the spirit of the prediction and resolve YES. Again, using the heuristic of me personally being able to make such a car appear at my doorstep without elaborate arrangements that few others could reasonably make.

4. What if self-driving taxis totally happen but you don't ever use one?

Oh I will, trust me. But ok, if I move away or something, I'll ask a friend to try. If I'm certain that I could've done it and just have someone else do it instead, that still counts as YES. Again, this is very unlikely to be an issue.

I don't think we have to worry about this since I believe latency/connectivity questions make this unsafe and/or illegal. But we'll say it has to have similar autonomy to what Waymo currently has in Phoenix and San Francisco.