To clarify, I'm not talking about simply exceeding the speed limit. I'm talking about reckless driving. There isn't going to be a situation where you *need* to go 106+mph.
On public roads, no. On private roads, yes. Manufacturers will claim that "Montana could always increase the speed limit", or cite the need to sell the same model in Germany, where there are no speed limits, and talk about dirty data and whitelists and people who take their car to a new track only to find that they don't have the maps update for it yet (most manufacturers do not offer OTA downloads). They will cite safety issues and whatnot. Weak claims, but since high speed crashes are relatively rare, the courts find there is no reasonable way to prevent reckless speeding while not compromising traditional legitimate uses.
Clearly, though, Japan has already moved into this kind of regulatory framework, and GPS and OTA technology is relatively new, so eventually you may a court rule for negligent design--especially as autonomous cars show the world what's easily achievable. And I suspect fear of lawsuits is why there is a "gentlemen's agreement" among manufacturers to limit top speed to 155 mph, rather than, say, 200 mph.
Citation required for low number of high speed crashes (it is very easy to google crashes of cars going 100mph), but if you are using that argument, the number of crashes caused by automakers not limiting lane keeping to divided highways is relatively low. By that logic why wouldn't "It would likely take a federal law mandating (lane keeping) governors before manufacturers would comply." apply?
Well, I should be more clear: I don't think there are many lawsuits, because it's a cultural myth that you can't create a good limiter, and no legal precent has been set. But assuming these lawsuits could proceed, you'd see far fewer reckless speeding liability claims than autopilot claims. How many people take their cars above 120 mph? One in a thousand, I'd guess. How often? Not that often. What percentage of autopilot users will engage a autopilot inappropriately? I'm going to claim 50 percent. How often? Almost all the time, because it's used for regular driving. That's at least 500x more hours/miles, likely 10000x; accidents would be far more prevalent, and the liability begins to add up. That's sufficient for multiple class action lawsuits.
I'm talking about exactly the same type of limiter. There are only going to be a limited amount of tracks in the US where you can go significantly higher than 100mph (I know this from threads discussing suitable locations to safety test the top speed of the Model S). A GPS based limiter of this type would be very easy to implement. The same reason that automakers don't implement it (and nobody expects them to) would be the same reason why people are arguing for Tesla not to put a restriction.
Clearly you're not buying my reason, so I'm at a loss: why don't automakers put in a GPS based governor?
I think you'll see a cultural shift where inappropriate autopilot use is going to be viewed a lot more like drunk driving. From the evidence we have right now, it appears that a substantial portion of the population both engages Autopilot inappropriately and is unable to control Autopilot in all situations.
If society had a magic technological switch that would eliminate all drunk driving, they would throw it. And that's what Autopilot restrictions are: a means to consistently, and reliably, prevent Autopilot from driving drunk in a situation it was never intended to handle.
That said, this argument only has weight if Autopilot is substantially less safe in some situations than human drivers acting alone. I think that's the case right now, but ten years from now it most certainly will not be, so what we have here is a temporary phenomenon, and thus temporary restrictions.
People keep saying that "other manufacturers" are doing things "cautiously" but I found that to be false (the Infiniti video is a prime example). For this specific scenario, none of the other automakers have implemented locking out of lane keeping based on divided highways, even though they have the same (or superior sensors) that can detect a divided highway. There is a double standard here in terms of Tesla.
It is a funny thing that the better an autopilot system is, the more is expected of it, because it opens peoples' eyes as to what is achievable. These other systems are so primitive that nobody uses them and nobody expects anything of them (similarly, nobody expects car manufacturers to be able to speed limit their cars). As they get better and more popular, I think you'll see more people pushing the automakers' to further improve.