While that is true, that’s only a small percentage of drivers. Most Tesla drivers do not pay for FSD at all. About 12% of Tesla owners pay for FSD in one form or another [0].

So even though paying monthly is more amenable, the vast majority don’t want to pay anything.

[0] https://www.businessinsider.com/tesla-full-self-driving-sale...

I guess males wreck more. I pay double now for myself as a single man with a car than I did for both me and my spouse and our two cars. Went from $100/month to $200/month overnight.

I get your joke. But I like paying more to ride by myself.

It’s the driving I don’t like.

> This, of course, will require some kind of centralized control over entire convoys, and a way to coordinate them. Railways and airways definitely can offer examples of how to handle that.

Not at all. A simple peer to peer protocol based on proximity and mixing in traffic data distributed like the national weather service will do just fine.

These convoys seem like a perfect example of swarm algorithms fitting well where you don’t need a central coordinator.

Within a convoy, yes. Between convoys, a dispatcher service could be beneficial, distributed and federated, again, like air traffic controllers and railway dispatchers. The same self-driving car companies that produce the software and require subscription could offer it.

No, GP is right. Check out the urban/rural populations in 1900 [0].

Cars allowed for suburban sprawl but the country was already really spread out before cars.

Maybe if cars didn’t exist we would have eventually consolidated into dense population centers.

You’re right that US cities were large and well populated, but that’s not where most people (60%) of Americans were.

[0] https://en.wikipedia.org/wiki/Urbanization_in_the_United_Sta...

Same in Atlanta. Best ethnic food is in the suburbs.

I think the point is that it’s much less than all the cars.

And a hailstorm that knocks out 10,000 cars is very rare. But hurricanes or fires that knock out billions in homes happen almost every year.

Or isn’t the alternative a local AI that runs autonomously?

We’d may for updates and new models, but no need for a subscription.

Of course the manufacturer makes more money with a subscription and this is the reason they want it.

The first point is exactly my thought. Self-driving cars are completely different to human drivers. We should not hold them to the same standards while simultaneously holding them to much higher standards. There are many driving violations that are just laws because they could lead to an unsafe scenario that is purely the fault of the driver.

Eg; stop signs. The only reason a full stop is required is to ensure that drivers are taking a clear observation and to give way to other stop signs. If there are no other traffic and no other drivers to give way to. Why do self-driving cars full-stop

You’re probably right in the long term. So, when the world is 100% self-driving cars, we can probably change the rules to favor the machines. In the near-term, however, it’s probably good to make the robots obey the human laws so that the humans don’t start getting the idea that they can disobey them, too.

laws of physics still apply. Car still takes time to slow down, even with perfect reaction times. Well, maybe you could get it to stop in time, but it might break the necks of everyone in the car.

Given how hidden children are walking in front of the bus, if the AI instantly applied breaks upon seeing the child, would the car slowdown in time? probably not. Better yes, good enough? no.

Their license would be suspended if this were any of us. There's been at least 30 documented instances of this behavior.

In 1998 I worked for a small nasdaq company that had a successful software as a service product that was growing quickly.

We used Clarion and MSSQL7 on windows because it was cheap. Since we started making real money, some figured we could finally afford Oracle and Sun (back when they were different).

I was a junior so my job was to evaluate the migration of one of our sql servers to oracle to test it out. I talks with the Oracle team who helps people plan purchases. They took my transaction level (~100M/year) and size (1-2GB/year) and came back with $1M for the system. This replaced a functioning $10k server. And we had maybe a dozen that would have to eventually move.

When I told them the current server was $10k, they revised their estimate to $100k. I recommended we not move.

I left the company a little while later and I think they ended up buying lots of Oracle.

Companies have money and don’t mind spending on useless stuff.