Car Insurers Find Tracking Devices Are a Tough Sell | Hacker News Reader

Car Insurers Find Tracking Devices Are a Tough Sell

wsj.com

55 pointsshahryc83 comments

Once interviewed with a company writing the web interface for users with these devices.

Them: So as you can see here, we have a user in Florida who had 3 hard brakes and an excess speed reading in the past month.

Me: Interesting.

Them: Ooops. Clicked wrong button. She's going to have reset her password next time she logs in.

Me: That was an actual user?

Them: Yes, that's the live system.

They told me they were popular with drivers looking for a discount with online insurers.

There's a lot of pessimism ITT, but I'd like to say: As a new driver, I love these things. I started off with a multi-thousand insurance bill yearly, even with the cheapest insurance I could find. This is simply because I am young, male, and had to buy my own car - I don't think the insurance company was screwing me over, I'm just a high-risk demographic.

However, I'm a fairly cautious driver - why should I have to pay for the indiscretions of other people in my demographic? I signed up for a UBI program and received a device that tracks

* How long I drive over 120km/h

* When I drive

* How often I brake hard

I don't consider this a huge breach of my privacy, and it has saved me over $5000 so far. If it required GPS my tune might change, but across 3 different insurance companies I've never seen a program like that. Maybe if I had the luxury of belonging to a low-risk demographic I would be more sensitive, but for people like me these programs are a godsend.

(Full disclosure: I have not read the linked article due to paywall)

I know it feels really rad and smart when we see this headlines and say "finally Americans are waking up to privacy conerns!!!11!", but the mere existence of such a headline only indicates Americans' growing apathy to digital footprint of any kind. And for good reason: if Skynet/Matrix/Hal were close to such a self-awareness, we would have already solved some pretty legit issues (e.g., poverty, greed, insecurity, life, etc.).

That is the most optimistic outlook on the AI apocalypse I have ever heard.

How much were the premiums inflated in order to convince you to opt in to the tracker?

How do your premiums change when you get several years of no fault claims or no claims at all behind you?

Well, companies that didn't offer UBI-based systems (and therefore have no motivation to push me to UBI systems) offered me similarly ridiculous insurance rates. So if I'm getting screwed over it's nothing to do with the UBI systems.

do yourself a favor and buy a dash cam that you control, because i guarantee that tracking device will be used against you if you get into an accident.

trust me on this one, "new driver". insurance companies never do anything for your benefit, only theirs.

How would his insurance company "use it against him"?

Proving he was at fault would also prove his insurance company had to pay for the accident.

"I'm sorry sir, the system shows you were driving 2mph over the limit when that drunk driver wrecklessly hit you, therefore you're partially to blame and we won't be paying out the full amount..."

If they wrecklessly hit you it won't be a big deal, since there was no damage. (Recklessly vs. wrecklessly. Sorry, the pun was too good to pass up)

Well, the whole point of auto insurance to pay out when the driver is to blame so.....

>> "...pay out when the driver is to blame..."

That's what auto insurance marketing says, and sometimes it even works that way. But a key part of the business model is to take your money for years, and then when you do have to submit a claim, they welsh on a technicality.

"Your vehicle is licensed in Seattle, but our records show you spent more than 50% of your time in California this year, therefore your coverage is retroactively void even though we took your money for 11 months."

If you actually read the giant auto insurance contract you signed, it is full of this and other similar loopholes, which are quite deliberately inserted to allow the insurer to manipulate the magnitude of its payouts.

I dont know. I always read my policies. My auto insurance policy is about 13 pages of large font which is written in plain, easy to understand, English. All the "loopholes" seem pretty straightforward and reasonable to me. They seem pretty easy to avoid to me.

Yea, it stands that lying about where you live to an insurance company would be a good reason to deny your claim. You were paying the (probably cheaper) Seattle rate, not the CA rate. If you split your time between two different places just get the proper insurance for that.

From my personal experiences with vehicle insurance, I think you're a bit optimistic. Does your policy have an "act of god" clause? Because that's a loophole big enough to drive a truck through. (Which is what happened to me, when a shipping truck driver totaled my car when driving it through a hurricane without covering their load, and the insurance company refused to assist because it was an "act of god")

i can say this as sure as anything else in this world: anyone who defends insurance companies has surely, surely never actually dealt with one in a time of need.

hn is full of hypersmart young people with very weak mental models of how things actually work in the real world beyond the tech industry, and probably have had very few if zero bad things happen to them (yet).

Yes, surely I have. I've had nothing but positive experiences with auto insurance companies. Health insurance is another story.

One problem with the pessimistic mental model is that it provides nowhere else to go with reasoning. For example, how can a tracking device cause problems with insurance claims, if insurance companies never pay claims anyway? If the worst is always expected, then any heuristic returns the same result.

The other problem is that it's contradicted by mountains of evidence--insurance companies pay millions of claims ever year. I know a lot of people who have been paid for losses by insurance companies, including me, even when they were at fault.

Extreme pessimism is itself a weak mental model in that it does not adapt itself to evidence. It takes no effort at all to parachute into any given conversation and say "that'll never work."

That implies an exclusively pessimistic model, however. My suggestion was to simply have a more pessimistic prior. As you say, insurance has to pay out _something_ _sometime_ by the data, and I've certainly seen some of those payouts myself (and sometimes even had them be advantageous in my favor), but if there's enough prior for adversarial interaction it seems a more more robust approach to plan for a worst case scenario even if it doesn't end up happening; or at least don't be exclusively optimistic, for this provides the same but inverted issues as exclusive pessimism.

It is a small step for a longer tail.

An insurance works that way, that a whole bunch of people pay a small "fee" into a teapot. If something happens to one person, than this person gets the teapot. So, it is a little like a lottery, but with the assumption that you will hopefully never win the price. That would mean, you never had an incident.

This general principle is already violated, when insurance companies tweaked that system to give some people within that bunch an advantage. This advantage is given by demographics (young vs. old; male vs. female; ethnic; experience; …) to those some people. That means, that all those others will have a disadvantage.

Now, this technology provides another possibility to tweak the rules, to give some people an advantage over others. That system is tweaked so long, until you have one person who will definitely pay for nothing. That means, this person is separated from the pool and pays for nothing. Because the pool still pay to cover up their ass and when something happens, it will be paid out of their pool.

I hope you can see the spiral. Basically, you will now have a advantage over others in the pool. In the long run, you and we all will loose.

That is (next to privacy concerns) one of the reason, why I am opposed to any of those tweaked insurances or valued-added insurances. We need to come back to the original idea of an insurance. That is also the reason, that I always think about and ask myself: "Do I really need this one?"

Of course, there are some, like car insurance, which you need to have. I know, I am also rated with those insurances like my age, that I don't own property but I own a registered historic car, and such things.

> If it required GPS my tune might change

What are you talking about? Of course they do. They either use GPS or GSM or both to track your position and speed. How do you think it determines what speed you are doing? Magic? What brand of device do you have? It may be true that you have one of the only systems in the world doing this (e.g. Metromile) but it is not true for 99.9% of other people with trackers.

The only reason it is cheaper is so they can gain customer base to justify their investment in the technology & infrastructure and hope for a legislative enforcement of the uptake of the devices.

They connect to the ODB port to monitor speed, and braking.

From Progressive's site: What information does the device collect? The Snapshot device collects information about how you drive, how much you drive and when you drive. It also collects your vehicle identification number and triggers an email to you if it comes unplugged. Some devices collect location data: this is only for research and development purposes—we don't use it to calculate your rate.

Sounds like these systems look for all the things that make driving fun

True story, last night I was driving on the motorway, just cruising at about 60mph. "You are breaking the speed limit!" my GPS announced. She thought I was on a 30mph limit road that runs parallel for a bit, about 50ft to the left. Do I want my car snitching on me to the insurance company or the po-po? Not a chance I will ever risk getting a tracking device fitted.

Where I live car insurance is very comprehensive and one of biggest reasons why people insure their cars is as protection for theft.

For the most popular cars, the ones at higher risk of being stolen, tracking is mandatory. All insurance companies require it, otherwise they won't insure your car.

So, with the justification of a higher risk and by acting in a coordinated manner, peruvian insurance companies (4 which control the market) have effectively deployed a vast tracking network.

Don't know if something similar could happen in the US, there are too many insurance companies as to easily establish common policies across the board and I suspect that it would be harder to come up with a credible justification for it.

So what's the car theft rate there? What about RF key validation?

There are different types of tracker, the ones used for theft prevention / detection tend to be the type that are only activated once a theft is reported. The article is about active type ones that record all driving behaviour to assess your driving.

Once the (very closed source) metrology hardware and transmitter are installed, you have no way of knowing if they decide to track you intermittently, constantly, or never. Can they remotely and silently activate the tracking system at arbitrary times? Probably yes. Can they push software updates over the air to drastically change the functionality of the tracking system? Also probably yes. At the risk of sounding like RMS, an insurance vehicle tracker is not your computer - it doesn't do as you tell it, and it is not on your side. Furthermore you have no idea what they do with the data downstream, particularly wrt giving it to the law and storing it in very insecure ways such that (other) malicious actors can get it.

Unless you're the kind of person that believes in the truthfulness, good faith, and technical competence of insurance companies and the police, in which case I have a bridge in Brooklyn that I'd like to sell you.

Normally, trackers design for theft detection and reporting can not be re-used for active tracking without changing firmware.

The first one is triggered by a dedicated signal from within the car to be activated. Otherwise they are offline to conserve energy. Energy is a very very limited resource in a car. When they are activated, they send a text message for the GPS position. Only a vary view newer once can trigger a web request to report the GPS position. When they are activated they can receive text messages to execute commands like disabling the engine.

Versus active trackers are the once mostly designed and engineered within the last 5 years, are able to consume very low energy while activated. They report the position (and may acceleration, deceleration and other parameters) via a web request.

Theft prevention are mostly established legacy systems more than 10 years old. Thats a different technology then we are used today.

The ones used here are activated remotely. If we assume the tracker to be similar to a cellphone and that it goes into standby mode after the car is of it will be drawing in about 2 - 4 ma, considering that a small car battery has a capacity of 45Ah the consumption would be negligible, it would deplete the battery in 625 days.

You are not working in automotive? mA is a considerable amount even when running. In switched off mode, any ECU (with very few exceptions) should not draw more than 100 nA.

Remember, there is not only 1 ECU. There are around 40 to more than 100 ECUs. A battery should deplete not more than 50% in 8 weeks.

There are in fact some ways..

1) Measure power draw from the car battery (not 100% conclusive probably but a 24-hour measurement might be)

2) Get a 20$ RTLSDR stick, drive somewhere remote with no cell coverage and watch for the tracking device in the GSM/3G frequencies - if found to try to connect it is at least powered on and trying to listen to commands

3) If activated and running on GSM, there might be a chance to listen to and decrypt its communication with the cell tower, especially if text messages are being used - and if not or it's running on 3G you could still monitor how active and frequent the device is in sending and receiving data

1) Would require for you to establish a baseline current draw. If your car has already been equipped with the tracking device, it might be very difficult as they are extremely well hidden and the owner is rarely informed of the location. Current draw also varies depending on aftermarket accessories installed, such as car alarms.

2, 3) Like in a multiple level underground parking lot?, that might work.

It is technically possible but probably beyond what the average new car buyer is willing to do.

What car thieves do is to drive the newly stolen car as far as they can in as little time as possible and park it somewhere public. And wait. If its not recovered after a day or two, they assume its not being tracked. (they remove the stereo, ecu and other valuable components in the meantime)

Over the last 2 decades, technology has secured many products from common criminals. Think of Internet banking or the authentication provided by mobile phone SIM cards.

Yet cars theft is still surprisingly common. Why is that ?

Car theft has virtually fallen off a cliff in the United States. In 2013, the rate was 1/3 of what it was in 1991 (the peak for auto thefts in the US).

https://www.nicb.org/newsroom/news-releases/historical-look-...

Because car manufacturers are idiots who, in the race for the most "ooh, shiny" features don't appear to understand what "improvements" are actually useful and appreciated. If you look at many of the nice features on an S-class Mercedes, for instance, they're neither expensive nor technologically advanced, yet other car manufacturers don't see them as selling points. E.g. the ability to get cold air in your face and warm air on your feet simultaneously. Or the tiny fabric flap between the driver's seat and the central console that stops stuff from your pocket falling down into that hard-to-reach gap.

To wit: look at the amplifier attacks that got hundreds of Range Rovers stolen in London last year [1]. Is the tiny effort saved in not getting the keys out of your pocket worth a 50x increase in the chance of your car being stolen? And the subsequent large increase in insurance premiums? Most reasonable people would say "No".

[1] http://www.theguardian.com/money/2014/oct/27/thieves-range-r...

Here they are under the guise of roadside assistance and can silently listen in on you. Google stuff about ONStar for details.

Indeed, wonder if they have been tapped of if there are backdoors built in.

This made me think of the "official cars" the peruvian government gets gifted from time to time. These cars are used by ministers, the president and important people in the government.

Shortly after china gifted a batch of cars an limos the japanese government did the same. Wonder if they are tapped?

I have one and don't mind at all. My insurance company is also my health-insurance provider and would know immediately if I have an accident, so they'd call me and if no answer would send help automatically. If my car is stolen, they can track it. They have a clause that prevents them from using the data against me in e.g. an accident claim. I get up to 50% of my fuel cost back monthly if I drive well and e.g. check my tire condition yearly (although linked to household insurance as well). I have friends with fast cars that wouldn't touch this and that's fine - keeps my premiums down because I'm not subsidizing their behavior.

I'd rather have them tracking me than my Android device, say:

https://www.google.com/maps/timeline

"And here is the Apple Watch that tells us every time you smoke a cigarette, or think something shameful. Healthy mind healthy body."

I wonder if one way to make better use of this data is to restrict what detail the insurance company can see. For example, you could imagine an arrangement where a third party vendor subject to strict controls gets the raw data and produces summarized scores, for example the frequency of hard brakes. Insurance companies would then only see these summaries, which could be challenged by customers, and can only vary premiums based on the summaries.

So basically the automobile equivalent of credit reports/scores?

When you apply for a loan in the US, most lenders these days will only look at your credit score(s), but they also have access to the full credit report and will sometimes have a credit analyst look at the report manually and form a judgment. So this would be like a scheme where the lender/insurer would only have access to scores, and not the full report.

At the risk of sounding cynical: how long before the insurance providers lobby the government to make these trackers mandatory because they enforce "safety?"

I wonder if the data can already be sold from cellphone carriers.

Someone should build an insurance company for Teslas. Build it as a Tesla app connected to the log data. No need for creepy trackers, they're already built in ;)

Honestly, the tracking is the biggest thing that would stop me from getting a Tesla. Is it possible to completely sever the data connection for the car?

Could probably route it through a non-existent proxy.

I was more thinking pulling a fuse or unplugging an antenna, but that could work too. Any idea how that would be accomplished?

Teslas connect via AT&T, so I doubt that one could add a proxy.

Find the SIM card and take it out. They might have soldered it on though.

So you don't get critical updates and your car blows up.

OR you don't get updates and your car is the only one that doesn't blow up.

"Tesla will have an enormous amount of data, of real world data, because of the millions of miles accumulated per day from our fleet of vehicles. Probably Tesla will have more real world data than any other company in the world." - Musk [0]

All things being equal, they're probably one of the best groups of people to hand this power to, but I would take an open source EV over a Tesla. Some (most?) other car companies track their new cars too (eg Jeeps broadcast GPS coordinates [0]).

[0] https://medium.com/backchannel/how-elon-musk-and-y-combinato...

[1] http://www.wired.com/2015/07/hackers-remotely-kill-jeep-high...

The tracking systems is one of the reasons I ordered a Tesla.

Google (location awareness) and T-Mobile (cellular provider) already know my locations at all times, I'm fine with Tesla collecting that data in return for software updates, Autopilot refinements, and for the vehicle returning data to them to enhance their own Autopilot algorithms.

Sell my privacy for a tiny discount of giving away money and getting nothing in return? Sign me up.

you get a discount on insurance

until they make it mandatory and then the discount will be as good as gone.

The discount is partly there because tracking makes it cheaper for them to insure you, because they can estimate risk better. The better you are at estimating risk, the narrower your margins can be.

Consumers don't win however if they are judged as now higher risk (which should be approximately half of them).

Do you want to pay for someone else to be a bad driver? Or should that person put down the beer and the phone?

Insurances should be there to cover risks, and they work because you have sufficient base size to cover for the % of people who will have accidents. Once you start discriminating it destroys the purpose of having an insurance for everyone in the first place.

Consumers on average win, because average rates go down, because insurers have better estimates of risk. Or do you think that the good drivers should subsidize the dangerous ones? (Privacy issues aside.)

I don't agree. The trends of insurers is to cover less and less and to raise their premiums as soon as they can. They are clearly profit maximizers, not "consumer-centric" organizations at all. Just like banks.

A naive assumption of incentives would make one assume that tracking devices would make people drive safer and save everyone money. More likely consumers will not drive any safer[1] and instead not only have to pay for normal insurance, but also the tracking devices and than data analytics.

[1] Difficulty the incentives people already have to avoid accidents is likely cognitively saturated. See red light cameras.

Or, isomorphically, you pay a premium to avoid being tracked.

Good. I hope this idea dies quickly.

My brother recently got one of these on his car, but he found he really didn't save anything by having it. Based simply on the amount of time driving during "peak hours", namely rush hour/when everyone is leaving work, he lost any discount he was pitched by the insurance company. This isn't really anything he has any control over.

In that case he seems to be getting a snow job from an insurance company that is using the the idea of quantified self/living to seize a larger part of its customers' lives, likely without really paying them back anything for the intrusion.

Similar to my own experience. For the first 6 months, I got a 10% discount just for having it. But immediately after that, the discount dropped to just 1% - regardless of how little I exceeded 80 mph (never) and how little I hard braked (once or twice a week at first, once or twice a month once I got a feel for where it detected braking) - it didn't matter. I was "grade A" for those categories, but C- for distance. Anything over 7500 miles per year basically removed any chance for a discount.

Now that I live much closer to my job, I could probably qualify, but I also found a little stressful to always be worrying about maybe triggering the "hard braking."

If car insurers want to make them an easy sell try solving a harder problem than just small discounts. Register a car in the city of Detroit and your insurance rate goes up 200-300%.

How about a device that you can activate from your cell phone that not only tracks your car but actually shuts it down? People would be creeped out by tracking but if you can save 70% on your insurance and at the same time more likely save your car from the chop shops it would be a win for everyone. Then have a private service that retrieves the cars so the police can work on more serious crimes.

These devices would be easier to swallow if you could view the data that's being collected, knew how your score is being calculated, and had a chance to appeal errors/anomalies.

When I used one with Progressive I had the first two.

You can. I've used them from about three auto insurance companies. All of them had a web interface with exactly that. One you could set text or email alerts on certain events also.

  "These devices would be easier to swallow"

Don't give them ideas.

From the article:

"Smartphones have made Americans more comfortable with location tracking."

"Privacy already has shifted from being a right to a good that is purchased."

Right.

This article touched on the fact that additional information these sensors provide is interpreted in a traditional context.

"Early company surveys of people’s interest in usage-based insurance revealed that about 40% of people had a viewpoint that was some variation on “No way in hell.”

followed by

"Progressive concluded Snapshot was overestimating the potential for accidents from midnight to 4 a.m. driving on week nights, and now considers late-night driving to be high-risk only on weekends."

That's why people don't want to tell their insurance company anything more than the minimum. The insurance companies thought driving at night was risky and it wasn't until they had snapshot telling them that just because someone works 3rd shift isn't an inherently higher risk did they not pull their head out of their butt. So now, via snapshot they can offer competitive rates to people on 3rd shift whereas before they'd just have jacked the rates on the few that were unlucky enough to file a claim for something happening in the evening. Um, hooray?

Hard-braking: A+ Speed: A+ Peak risk hours: A+ Distance: C-

Because I drove over 7500 miles per year, I got the minimum discount (1%) - apparently experience decreases skill level ;)

As soon as the initial 6 month 10% discount was reduced to 1%, I dropped the device.

In the UK these are being pushed as a smartphone app

Reply on news.ycombinator.com