You--a human driver--have no idea what the other cars will do either. Some people run yellows, some people don't. The only reasonable thing to assume the other car to do is follow the law, but be prepared for the unexpected.
Unless Waymo cars are doing something unnatural like braking significantly harder or faster than a human would, the Waymo car is not at fault for being rear-ended.
Once we pass ethical muster (meaning don't ship knowingly buggy code, don't sell cars you expect to crash a lot, etc), I have a hard time seeing what the moral issues are here.
Waymo cars are often unnaturally cautious, which is usually a factor when they get rear-ended. Ex: vehicle behind expected waymo car to turn left when there was an opening large enough for it, but it didn't.
These are probably worse because there's a safety driver who is probably actively engaged. If you can see the (apparent) driver looks actively engaged, you'll expect them to see the gap, and go for it. If the car had no one in the driver seat (or they were looking at screens/passengers), you might be more cautious.
Added: of course, that doesn't mean it's waymo's responsibility; this type of collision occurs with human drivers too, and it's the driver to the rear's fault, but either driver could have avoided the error: the front driver by going when it was safe to do so, as expected; and the rear driver by waiting to confirm the front driver actually went. Sometimes it's tricky because the front driver releases the brakes and then reapplies them; the rear driver saw the brake lights turn off and moves forward while watching oncoming traffic, but doesn't see the brakes were reapplied.
Added later: I read the verge article. It sounds like Waymo is working on not getting rear-ended, so they acknowledge there are things they can do (and that, in many cases, the safety driver was able to do).
No one has managed to rear end me, but I obviously don't have 6.5 million miles driven. And even if they did, I would trade a high speed t-bone for a low speed fender bender any day of the week.
If traffics is light, it's pretty easy of course. But get into congested areas and you pretty much have to be at least somewhat aggressive and even take things like Pittsburgh lefts. And if you're not at least somewhat aggressive, cars behind you can start honking horns and taking dangerous actions because, from their perspective, you're blocking the road. (No criticism intended; you have to do what feels comfortable for you.But it's probably one of the routine driving tasks that takes the most judgement.)
But one of them may be more likely, and therefore result in more collisions if you assume it isn't.
In addition to that, many yellow lights are timed too short. That makes it possible to be far enough from the light that you won't make it through the intersection before it turns red, but still close enough that stopping before the intersection would require braking rapidly enough to yield a significant probability of being rear-ended. The problem there is the light timing, but the car can't fix that and still has to make a choice.
That's right, but not the point. They are optimizing for not being at fault, instead of optimizing for not getting into an accident.
Our phones should SUGGEST slight changes to our driving to improve efficiency. For example there's a jam ahead. Google/Apple know our phone's speed and the speed of those around us already.
A lot of backups are due to a lack of information transfer. The wave of stopping/slowing is the propagation of that information. If our phones informed us beforehand to slow slightly we could smooth the way out much faster.
I know it's a privacy nightmare right now but it's doable and everyone that's driving would potentially benefit.
That doesn't mean that it can't still get rear-ended by being more cautious than some impatient and unsafe human drivers, for example when it decided there is plenty of time to stop safely.
I don't think there are that many really. Speeding is one, but most of the time following the speed limit doesn't dramatically increase the risk of accidents. Then you have stuff like turn signals. Using them isn't going to hurt anything, whether drivers expect them or not.
Another thing to take into account is that not all crashes are equally bad. I wouldn't be surprised if getting rear ended 10 times was safer than a single t-bone or head on collision
This presents an interesting problem. It's obviously easier to program something to follow the law, given it's unambiguous. But the question is what are we optimizing for? The fewest crashes? That's probably the right thing to do given crashes are bad. In that case, isn't it better to do what people would expect other cars to do? But are Waymo constrained by the fact that if a self-driving car is programmed to get a ticket they could be held liable? Probably.
I think the moral side of self-driving cars is just as hard or a harder problem than the technical side, and we haven't made the decisions as a society that we need to. If the government doesn't step up soon to lay out how this is going to work, the corporations will. And guess what: they'll choose whatever costs them the least amount of money. Not what's best for society.