Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I was part of a project to measure reaction times in a primitive driving simulator. Accelerator, brake, steering wheel from a gaming setup. You drove down a straight road, some roadside scenery went by. Every so often you went through a gate. The gate would sometimes close just as you got there. Measured whether you hit the brake in time. Simple, a little challenging. The idea was, for drivers to self-measure if they were still able to drive safely. Tried to pitch it to an insurance company that rhymes with Slate Charm.

They acted like it was poison. They didn't want to even appear to be trying to select out older customers. The legal trouble they'd have, discriminating according to age (or even appearing to try) would have brought a landslide down on them.



Huh, if they can legally justify explicitly discriminating based on sex and age (younger men generally pay more than younger women, while older women generally pay more than older men) I’m surprised they couldn’t figure out a way to discriminate based on age alone.

Edit: here [0] is a breakdown of car insurance costs, stratified by age/sex. Young people pay by far the most, with premiums steadily declining until age 50 and then rising again. So I’m not sure what their objection might have been, since they already do charge old people more (presumably at a level accurately reflecting their relative risk).

[0] https://www.valuepenguin.com/how-age-affects-auto-insurance-...


Worse: couldn't figure out a way to "discriminate" based on a bona fide measure of driving ability that (I imagine) only loosely correlates with age.

But age discrimination laws/enforcement are seemingly one-way, and the elderly vote and sue, so...


Of course they discriminate on age. They don't want to make it clear to the people they are doing it to.


> They don't want to make it clear to the people

Wouldn’t a driving sim be the exact opposite of that?

“To make our premiums accurately reflect your driving habits and avoid lumping you into crude demographics like the other insurers, we offer all our customers an opportunity to take a spin in our driving sim and get a truly personalized rate reflecting your excellent driving habits.”


But most people wouldn't get a rate based on excellent driving habits, they'd get a rate based on their actual performance.


That's the exact argument they make for those drive tracking apps, isn't it? I don't think many people are buying that one, either.


Why wouldn't they? If I can prove I'm a better driver than the average Joe/Jane, and pay a lower premium as a consequence, why wouldn't I?


There’s very little upside - 5% discount for good driving- and lots of downside - 50% premium for bad driving.


I looked into getting one of those once, figuring I’m a safe driver and therefore would only benefit from it.

What I found on the Internet were (self selected, admittedly) stories of the presence of the device making your driving less safe. Since it’s basically an accelerometer, sometimes it’s “safer” to take a curve fast than to brake for it. Or fly through a yellow rather than stop at it.

I have no doubt there’s correlation between those devices’ scores and accident proclivity; but I’m confident my driving is safe as-is and would rather let premiums go down after accident-free miles than on the say-so of a nanny device.

So I bought a dashcam instead, hopefully to avoid being ruled 50/50 at fault if someone hits me.


My premium dropped more than 20% after adopting the tracker.


Those trackers see if you can drive calmly most of the time, which I imagine isn't nearly as age-dependent as a sim that tests emergency reaction time.


Which is interesting, paired with technology in cars that add additional braking force when they detect an emergency braking situation (quick lift off the throttle and immediate application of the brakes). Apparently studies have found that people don't brake hard enough before an accident.

Good thing we're encouraging people to be even MORE afraid of the controls in their vehicles, lest their insurance companies ding them for hurting the accelerometer's feelings.


In the US, it is legal (in a hiring context) to discriminate against young people, but not against old people. For better or for worse, it's not parallel.


Old people are much more politically active than young people.. So everything tends to favor them.


Sounds like they don't want to base cost on actual driving ability so that they can keep using the models that make it seem like 7 years of increased rates are justified when it's pretty well known that reaction time peaks early in life - many professional F1 drivers go pro before age 20. Therefore, your program would call attention to the fact that young drivers with no accidents who are competent are heavily overpaying and they weren't interested because it would prove that they are discriminating against the young.


Reaction times aren't everything when it comes to driving. You also have to factor in tendency to speed, tail-gate, drive drunk, or other dangerous driving habits. There may be correlation between those habits and age.


After someone turns 30 (according to AAA [0]) their likelihood of entering an accident decreases until they reach ~70 years old.

[0] https://aaafoundation.org/rates-motor-vehicle-crashes-injuri...


The variance overlaps from age 20-59 so you can't statistically differentiate anything conclusive about the difference between those groups. An alternative conclusion is that retired people (65+) don't drive as much and are able to avoid rush hour / high risk times to drive.

https://www.fhwa.dot.gov/ohim/onh00/bar8.htm


Variance overlapping isn't a reason to discount this analysis. Also note that the variance narrows substantially for older age groups. So effectively what the variance says is that some younger people are vastly more responsible drivers than other young people but older people tend to be more responsible in general.


There could be a number of factors that have nothing to do with responsibility, that's a narrative that gets added to explain the data. You need to do ANOVA test to verify any significant difference in data sets, not look at avg - esp when its not a controlled experiment.


Sure, that is the justification used to continue basing prices off statistical modeling instead of individual testing.


Of course it is! What do you think actuaries do? Insurance is entirely about sharing risks across groups of people.


What’s the argument against this exactly? Your doctor measures your healthiness and if you try to get insured, it’s priced accordingly. Older driver feel like they’re being discriminated against? Just take a simulation like this and prove your reflexes are within their defined spectrum of acceptability.


The argument is the AARP is a powerful lobby that you must not cross.


Could be a market for BMVs since they're responsible for license renewal. And they already test eye sight.


How long ago was this?

Now, insurance companies put trackers in cars and adjust rates based on actual speeding and braking.


That doesn't test reaction speed or driving skill. A slow elderly driver who is inattentive/has poor vision and pulls out in front of people would receive lower rates than someone who drives with traffic, but maintains proper following distance and is forced to brake quickly to avoid the inattentive elderly man who just pulled out in front of them.


Wow I suppose its been a few years. 2010?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: