Of course it's theoretically possible to create a third pedal that is programmed to feel exactly like a clutch, with tangible resistance at the friction point and the feel of stalling out if you aren't timing it properly. It's just a pretty complicated engineering task, especially given that it's not just a software question it's a physical engineering question, an it has an audience of uncertain size. But maybe we will see a niche enthusiast EV in the future that tries to fake the feel of a manual.

> I think we're still in too young a phase with autonomous vehicles. Theoretically, computers should be able to avoid pedestrians/cyclists much better than (very distractable) human drivers, but I don't know if the technology is fully there yet.

If we define "autonomous vehicles" as something like "full self driving cars" then yeah obviously the technology isn't there yet. That said, I think there is a lot of technology available right now that is well short of that threshold that could save plenty of pedestrian lives. Automatic emergency braking already exists in a number of cars and hopefully already helps avoid some pedestrian collisions (especially in daylight). We also definitely have the technology to actually enforce speed limits by software if we wanted to, we just lack the will.

On the whole, I would not seek to discourage vehicle autonomy features right now, despite the existence and well-publicized failures of of Musk the Moron. In the long run I'm sure they will be a good thing, and they are already probably having some meaningful positive effects.

> slowing down EV adoption isn't a bad thing

Disagree. Rapid EV adoption is an imperative. But it does create urgency to improve road safety in the meantime, because EVs are heavier and faster (and therefore more dangerous) than ICE counterparts.

> Semi-autonomous driving, however, is an absolutely awful idea and will produce even worse drivers than we have now. Tons of people are getting their driving reflexes accustomed to adaptive cruise control and lane keep assist.

IDK, this isn't obvious to me, it just seems like a narrow intuitive argument and I don't think I am fully on board with your intuitions. Most "semi-autonomous" features are features that engage under very rare and extreme circumstances, 99+% of the time drivers are driving their cars the exact same way we always have. It's not obvious to me that having some semi-autonomous safety features that engage under rare circumstances will radically change the driving behavior that drivers develop as they learn to drive. And even if it did, it would still be a question of how many accidents were avoided vs. caused by that change.

> What happens if they are in a situation where they have to drive an older car?

Sure, maybe they would be worse drivers in an older car. That seems like an edge case, though - as above, it becomes a question of whether accidents caused in those circumstances are more frequent or worse than the accidents avoided the rest of the time. It is also something that will gradually go away over the course of the transition towards vehicles with more semi-autonomous safety features, rather than a permanent problem with semi-autonomous features. Like, I'm sure that drivers today with backup cameras would be really shitty at parallel parking a car if they had to drive an older car with only a rear-view mirror... but we are rapidly approaching the point at which the likelihood of them ever having to drive an old car without a backup camera is getting pretty low.

No, I assumed -- incorrectly, I guess -- that anyone reading that comment would know that GPS is an important function that other people wouldn't be dumb enough to propose shutting that off.