Self Driving Autos

[Ron Weasley]

Originally Posted by MountainHawk

I think the GL form would give coverage here. No underwriter worth his salt would actually bind a policy right now, I would hope.

Under today's laws and forms, I agree to both points. At the time prior to the existence of WC, both points would have been valid also, that is, too expensive and no sane underwriter would bind it. However, legal strict liability for workers' injuries was established and WC still came into existence.

[MountainHawk]

Originally Posted by JasonScandopolous

I was ignoring inflation when I said premiums would drop 90%. Comp will be down too, because collisions with animals (significant in wooded areas) will be greatly reduced as well.

I doubt collisions with animals will drop that much. Most of them are generally unavoidable. Google can't change the laws of physics and make a car moving 100mph (since they will presumably go faster) stop in 15 feet before it hits the deer.

[MountainHawk]

Originally Posted by JasonScandopolous

I was ignoring inflation when I said premiums would drop 90%. Comp will be down too, because collisions with animals (significant in wooded areas) will be greatly reduced as well.

Also, premiums can't drop 90% in reality. Assume the loss cost drops from $500/yr to $50/yr. Assume $50 fixed costs in policy issuance and 25% variable costs.

$550 / .75 = $733
$100 / .75 = $133

Decrease of 82%.

I don't think a 90% drop in loss costs is even that likely. Maybe a 90% drop in frequency, but severity costs will be up.

[JasonScandopolous]

Originally Posted by MountainHawk

Also, premiums can't drop 90% in reality. Assume the loss cost drops from $500/yr to $50/yr. Assume $50 fixed costs in policy issuance and 25% variable costs.

$550 / .75 = $733
$100 / .75 = $133

Decrease of 82%.

I don't think a 90% drop in loss costs is even that likely. Maybe a 90% drop in frequency, but severity costs will be up.

I disagree with the latter; severity should be down as well. Any accidents (those not involving complete failure of the sensors) that do happen are going to happen at LOWER speeds, not higher, because the computer's reaction time in applying the brakes is ~0, as compared to whatever a human's reaction time is.

You do make a good point about fixed expenses. But, even this is overly optimistic... auto insurers will have far less to compete on once driving behavior is taken out of the equation, besides service and expense costs (bc every driver is nearly identical from a loss cost perspective given a certain milage if the tech works correctly; only comp loss costs would be worth doing real analysis on). I'd imagine that fixed costs would decline (in real dollars) as companies paid more attention to that as it became more and more of a % of premium. But you're right that it'll still be a big part of premium and will keep premiums from declining as much as I had thought. All of this assuming that no radical insurance industry changes occur, of course.

[MountainHawk]

Originally Posted by JasonScandopolous

I disagree with the latter; severity should be down as well. Any accidents (those not involving complete failure of the sensors) that do happen are going to happen at LOWER speeds, not higher, because the computer's reaction time in applying the brakes is ~0, as compared to whatever a human's reaction time is.

You do make a good point about fixed expenses. But, even this is overly optimistic... auto insurers will have far less to compete on once driving behavior is taken out of the equation, besides service and expense costs (bc every driver is nearly identical from a loss cost perspective given a certain milage if the tech works correctly; only comp loss costs would be worth doing real analysis on). I'd imagine that fixed costs would decline (in real dollars) as companies paid more attention to that as it became more and more of a % of premium. But you're right that it'll still be a big part of premium and will keep premiums from declining as much as I had thought. All of this assuming that no radical insurance industry changes occur, of course.

Most of the fixed costs are driven by regulations now, I think. Unless the regulation loosens up because of this (possible, but not likely I would think), I don't see too much dropping off.

On severity:

Let's say the autocar can drive safely at 100 mph, and the manual car drives at 65 mph. The autocar's extra equipment probably weighs a little more, but we will ignore that or assume they make the brakes better so it cancels out.

The autocar will stop after 4.6 seconds and 330 feet, assuming a 0.00 second reaction time. The manual car will stop after 4 seconds and 240 feet, assuming a 1.00 second reaction time. IANAP, but I don't the the autocar is ever travelling at a slower speed than the manual car based on those numbers. (from wolframalpha.com, use stopping distance as query).

So the crashes will likely be higher speed, and based on that, we may even have a few MORE animal collisions with this technology once it is in full use and the cars can go faster.

Plus, the other part of severity is fixing the car, and that is clearly going to be more expensive with the extra sensors and gizmos.

Edit: A speed of 85 mph with 0 reaction time is the same a 65 mph with 1s reaction time ... they both stop around 240'.

[JasonScandopolous]

Originally Posted by MountainHawk

Most of the fixed costs are driven by regulations now, I think. Unless the regulation loosens up because of this (possible, but not likely I would think), I don't see too much dropping off.

On severity:

Let's say the autocar can drive safely at 100 mph, and the manual car drives at 65 mph. The autocar's extra equipment probably weighs a little more, but we will ignore that or assume they make the brakes better so it cancels out.

The autocar will stop after 4.6 seconds and 330 feet, assuming a 0.00 second reaction time. The manual car will stop after 4 seconds and 240 feet, assuming a 1.00 second reaction time. IANAP, but I don't the the autocar is ever travelling at a slower speed than the manual car based on those numbers. (from wolframalpha.com, use stopping distance as query).

So the crashes will likely be higher speed, and based on that, we may even have a few MORE animal collisions with this technology once it is in full use and the cars can go faster.

Plus, the other part of severity is fixing the car, and that is clearly going to be more expensive with the extra sensors and gizmos.

Edit: A speed of 85 mph with 0 reaction time is the same a 65 mph with 1s reaction time ... they both stop around 240'.

I don't think autocars would drive significantly faster than they do now; if anything, they'd be slower because the cars wouldn't speed (because you can't program illegal behavior into them). I'm thinking that optimal fuel economy would determine the speed and would leave cars driving around 65-70mph on highways. That's the difference in our assumptions at this point, and I'm less sure about what'll happen in this regard than on other points.

[Darkness Falls]

In terms of driverless cars, until a foolproof antivirus comes on the market I doubt we'll see them without manual options (and therefore a need for insurance). It's one thing when your cpu crashed ... completely other when it's your car and you're inside.

[MountainHawk]

Originally Posted by JasonScandopolous

I don't think autocars would drive significantly faster than they do now; if anything, they'd be slower because the cars wouldn't speed (because you can't program illegal behavior into them). I'm thinking that optimal fuel economy would determine the speed and would leave cars driving around 65-70mph on highways. That's the difference in our assumptions at this point, and I'm less sure about what'll happen in this regard than on other points.

I can't imagine why people would routinely turn their life over to a computer when they didn't have to and it didn't get them there faster.

[nonactuarialactuary]

Originally Posted by Final Boss

Navigation is the easy part. Driving among ALL the other perils of the road is the hard part. The DARPA Challenge took place in the desert because it was devoid of stuff yet in 2004 not one vehicle could even finish the 150 mile course. And only 5/23 could finish in 2005. The fastest was Stanford's team (I think the same guys involved with the Google driverless car) completing 150 miles in just under 7 hours... I think it's safe to say that this technology is a ways away.

That's actually a point arguing against your conclusion. 2004 was fairly recent, and the technology back then wasn't even good enough to navigate a desert with no obstacles. Less than 10 years later, the technology has improved beyond anyone’s wildest expectations, and there’s no indication that improvements are stopping any time soon. If improvements from 2013-2022 are anything like the improvements seen from 2004-2013, we’re in for a treat.

[CuriousGeorge]

Originally Posted by MountainHawk

I can't imagine why people would routinely turn their life over to a computer when they didn't have to and it didn't get them there faster.

Because now I can spend an extra hour a day reading, working, studying, or playing Words With Friends, instead of staring at a stripe of pavement. I can't imagine why people would routinely forgo that, unless they are a car enthusiast. And even then, I can't imagine they get much enjoyment from the daily commute.