Data Mining Exec Pays For Burgers In Cash To Avoid Insurance Company Snooping

http://www.techdirt.com/articles/20120619/04094319383/data-mining-exec-pays-burgers-cash-to-keep-his-insurance-company-knowing-his-bad-diet-habits.shtml

by ridruejo • 14 years ago

134 116 14 years ago

techdirt.com

nivertech• 14 years ago

I guess he 100% sure, that his insurance company do not data mine techdirt.com ;)

planetguy• 14 years ago

Ehh, the miles which I get by using my credit card wind up being worth about 2%, so you'd have to convince me that each five-dollar burger raises my insurance premium by more than ten cents.

excuse-me3• 14 years ago

I suspect that if you eat enough burgers that a 2% cash back is significant - it's not just your insurance company that should be worried!

sliverstorm• 14 years ago

2% on any one item is of course not all that significant, but like any other salami-slicing deal you have to make a habit of taking advantage of it wherever possible.

ojbyrne• 14 years ago

That's not a completely correct analysis, as the insurance premiums are a recurring payment. So it would have to increase the present value of the stream of recurring payments by more than ten cents.

spindritf2• 14 years ago

Discounted to the present value.

prodigal_erik• 14 years ago

If habitual eating is more predictive than occasional binging, the model won't be linear. There may even be some threshold that simply makes you uninsurable in the distant future.

But this may turn out to be a bad example, if medicine continues finding that heart attack risk is mostly linked to cholesterol you produce yourself (this is apparently hereditary), and not much affected by cholesterol you eat.

Jare2• 14 years ago

The original article linked talks about some company buying an old lady's health insurance (they end up not buying because they datamine and find the lady is not unhealthy). That such a thing is possible blows my mind - a company placing an explicit bet on someone's death?

yummyfajitas7• 14 years ago

That such a thing is possible blows my mind - a company placing an explicit bet on someone's death?

Um, life insurance is also a explicit bet on someone's death, right (a short bet, but a bet nevertheless). Also, a defined-benefit pension plan is a long bet on someone's death.

tankbot• 14 years ago

a company placing an explicit bet on someone's death?

This is what the entire life insurance business model is base on. What's amazing is the level of information gathering possible now, not the fact that they're (still) doing it.

chris_wot4• 14 years ago

Yes, but interestingly they aren't very happy about Viatical Settlements, because it's a guaranteed loss for the neurone company. But totally legal.[1]

1. http://www.lisa.org/content/51/Life-Settlement-History.aspx

jessriedel• 14 years ago

Life insurance is a bet on someone's life, but it's one that the insurance companies pays out if you die. Annuities are a bet that pay out when if you don't die.

How else would people hedge against this stuff?

mikecane• 14 years ago

Betting on Death http://www.washingtonpost.com/wp-dyn/content/article/2009/06...

Mz4• 14 years ago

I am reminded of the stories of insurance policies being sold on Jews in WWII knowing it was pure profit for the company since there would never be a death certificate issued with which to collect on the insurance.

Jare2• 14 years ago

Thank you, that article describes the matter pretty clearly. Now excuse me I have to go throw up.

(to clarify my comment to others, what I'm amazed at is the idea that some company bets to EARN money upon someone's death. With any 'normal' insurance policy the company wants me to live long and prosper and never have any problem)

ck2• 14 years ago

Not only to avoid data-snooping but a great way to stay on budget is to once a week buy a ~$1 item and get cash withdrawal at the register. Then use that cash for the week without going over because you cannot use your card again.

jmount• 14 years ago

Just a marketing stunt.

jackalope• 14 years ago

I have a very healthy diet, but I buy all of my produce at markets that only accept cash. I wonder what conclusions an insurer would draw from my supermarket purchases (mostly meat & dairy)?

kintamanimatt3• 14 years ago

That you're a fattychunks and need to be charged a higher premium to compensate for your vegetable-free diet. /s

Data mining is very hard to get right. Suppose I swipe my card at McDonalds every few days. That charge alone doesn't even mean I'm eating unhealthy food. I could be buying food for the homeless dude outside, or just buying bottle after bottle of Dasani for lulz.

People do unusual and unpredictable things all the time.

jrockway6• 14 years ago

Remember, this isn't a court where you get to argue your individual case. If data mining starts affecting insurance premiums, you probably aren't going to notice. The insurance company is not going to say "because of your daily McDonald's purchases...". They're just going to say, "effective 6/6/2020, your rate is $x".

People that do "unusual and unpredictable things all the time" are so rare that they don't matter at the scale of data mining. Most people do usual and predicable things, and that assumption is probably worth money.

However, a number of factors have to come together for this to work. First, the insurance companies need to correlate purchases and insurance claim rate. Then insurance companies need to secretly implement this plan. Then when you get the letter that says your premiums are being raised, they need you to not switch to a competitor that is undercutting their new rate. And finally, they'll have to avoid the rage of citizens and legislators. With their upcoming special status in 2014, it's unlikely that lawmakers are going to let insurance companies do whatever they want.

So ultimately, being different isn't going to protect you from data mining, but I think the market will. Do you want to be the first company to announce they're spying on your purchases? Do you want to own the grocery chain that's known as the spying one?

Probably not.

netrus• 14 years ago

This seems to be common behavior (many people want to buy their healthy food fresh and local), so the insurer will be aware of it.

neutronicus• 14 years ago

They'd probably notice the absence of e.g. green bell peppers and onions, which even dyed-in-the-wool carnivores tend to buy to go with meat.

spindritf2• 14 years ago

> I wonder what conclusions an insurer would draw from my supermarket purchases (mostly meat & dairy)?

"Follows paelo diet."

keithpeter• 14 years ago

Yup, I am in the same position. A scan of my debit card purchases would reveal a very low calorific intake however. The obvious solution is just to use cash for everything.

rhplus• 14 years ago

I buy fruit and vegetables in bulk from a small local store and pick up beer and convenience food from nearby supermarket when I need it. It's crossed my mind more than once that the supermarket is the one that's most likely to sell my purchasing history, and within that data-set I'm a junk food addict. Which is why I sometimes throw some fruit into my basket, just to skew the data a little.

excuse-me3• 14 years ago

There was a US supermarket that threatened a customer who was suing them after he slipped in their store - they said they would present the history of his liquor purchases to suggest to the court he was alcoholic

stfu• 14 years ago

Any source for that? Sounds almost too good to be true.

haberman3• 14 years ago

That is exactly the kind of insinuation that evidence rules are designed to protect against (in my layman's understanding of the law). From Federal Rules of Evidence, rule 404 (http://www.law.cornell.edu/rules/fre/rule_404)

"Evidence of a person’s character or character trait is not admissible to prove that on a particular occasion the person acted in accordance with the character or trait."

citricsquid• 14 years ago

Key point:

> he predicts

He's avoiding it because he thinks it's the natural progression of the product these companies offer, not because they do, it might never happen.

dredmorbius2• 14 years ago

Here's the bit about all of this data tracking that bothers me the most.

The data is, in large part, a one-way street. The companies which are most aggressive in collecting information are also the most aggressive in not disclosing how they use, share, distribute, aggregate, correlate, etc., this information. Look at the NDAs and "no talk" policies of Google, Facebook, any of the credit bureaus, to say nothing of the vastly less visible enterprises which service the B2B markets of data mining and information. It's an area in which I had some experience and washed my hands in disgust years ago -- it's also an area in which I'd very much like to leverage open channels and tools to provide the public with the ability to fight back against the problem.

I've had a very clear view for over a decade now that this will not end well.

The truth is that the public has little understanding of what data are being used, how, or by whom. Much more pernicious is how data can be aggregated. Your insurance company gets information about your car and license. Your smog check station runs a standard battery of tests against your vehicle and reports this to the state in a large electronic record. Your state turns around and sells this database, at a very, very low charge, to companies providing services to the insurance industry, so that based on VIN and license information, a huge dump of data form your car's onboard data collection systems is now available to your insurer. They're mostly interested in total mileage, but as car data collection systems advance, there could be a great deal more information there -- accelleration/decelleration, speed, conceivably in the future, GPS waymarks.

And you'd never know about it.

The mileage stuff? That's for realz and has been for a decade or more.

Edit: and just to put a face on this: http://www.iso.com/

The truth? That few even within the datamining field know what others within the field are doing with data (see above WRT one-way information), unless that information is being directly marketed/sold. Where derived products (e.g.: risk/profitability profiles based on models in which individual inputs are not disclosed) are sold, even uses which are directly sold may be using information in way undisclosed to the data suppliers / users, let alone the members of the public who are being profiled.

What could possibly go wrong?

cs7024• 14 years ago

I've been using cash for certain purchases for several years now, because consumer-protection laws, regulations, and policies are far, far, far behind the tectonic shifts that have already occurred in the world of data mining.

jrockway6• 14 years ago

Care to elaborate?

cs7024• 14 years ago

Here are some examples of things I prefer to buy with cash: purchases at airports (junk food, junk media, junk items); purchases while traveling by car (mostly junk food); purchases on riskier/dicier neighborhoods and areas; certain occasional purchases at the grocery store or pharmacy.

Also, on occasion, I will refrain from buying or getting certain things if I can't buy or get them anonymously.

Until the legal, regulatory, and cultural environment meaningfully change, my presumption is that every single non-anonymous purchase I make can and probably will be used against my economic interests, sooner or later.

pasbesoin3• 14 years ago

I have to clean up a bit of mold in the (detached) garage. It suddenly occurred to me that perhaps I'd better purchase the application-specific cleanup materials using cash.

ams61102• 14 years ago

Why is it not fair for an insurance company to charge you a higher premium if you are leading a lifestyle that statistically makes you more likely to have health problems?

Mz4• 14 years ago

Insurance theoretically spreads risk and operates on laws of large numbers -- in other words, the company is betting that if 500,000 people buy this policy, statistically it will only pay out on a relatively small number of them, thus covering operating expenses and turning a profit. But, for health insurance,* it tends to not really work that way. You don't get 500,000 completely random people buying the policy. People are more likely to buy the policy if they have reason to believe they will need/profit from it. So insurance companies try to account for that reality. I think it is a fundamentally broken system.

* Car insurance, which is basically required across the U.S., seems to work a little closer to the way it is "supposed to".

olefoo• 14 years ago

1. You are making the assumption that the insurance companies models of what is and is not a risky lifestyle are correct.

2. For the same reason that your employer is not entitled to put cameras in your bathroom to make sure you aren't doing anything that might impact your job performance.

31434• 14 years ago

1. All other factors being equal, insurance companies with more accurate statistical models will do better.

2.. Because many employees would choose to not work for such an employer, and this negative impact would outweigh the potential benefit? I haven't seen anything to suggest that people would choose not to be insured by a company that profiles its clients in this way.

Cadsby3• 14 years ago

I think on a principled level, many are becoming concerned about the ever increasing volume of data being collected on all of us. It isn't necessarily about any particular subset of data, or all the plausible abuses of it, it's just the idea that some would rather not have every move they make being recorded and logged into a system they have no information on or ability to influence.

More specific to insurance companies, the potential/plausible abuses to increase profit don't take much effort to ferret out. And to make matters worse the data they use will invariable flag a lot of false positives that you'll have little to no way of fixing. For example, my girlfriend regularly picks me up a pack of cigarettes on her way home from work. She usually uses her credit card instead of cash. Ok, now she's flagged in the system as being a chronic smoker and her rates get jacked up to holy hell. Except, of course, she's not actually a smoker and I'd be super skeptical about her ability to reasonably explain this to some random customer service rep over the telephone.

Data mining usage habits rarely result in lower prices for people making positive, less risky choices. Non-smokers will probably be giving a trivial token discount, but unlikely anything more.

yummyfajitas7• 14 years ago

Data mining usage habits rarely result in lower prices for people making positive, less risky choices.

This is because we live in an inflationary economy. Most prices go up. As a healthy guy, I'd like the price increases to be inflicted on the fattie smokers rather than me.

Not that it really matters in the context of health insurance - the laws surrounding it are structured in such a way as to subsidize the fattie smokers at the expense of healthy gym guy, regardless of what info the insurance company has (and they will become worse in a couple of years, unless the supreme court stops it - good luck with that).

sehugg• 14 years ago

Living your life according to contradictory actuarial tables cooked up by people running an anti-competitive market captured by government regulation? Sounds like fun.

rogerbinns3• 14 years ago

If you are going to charge people for what they use then it isn't insurance any more. That is the really silly thing in the US - the "insurance" system is really just a mechanism for obfuscating how much things cost.

And if you want to start down the "lifestyle" road then why shouldn't people who have kids pay more (it is a choice too), or the people in LA (who are more likely to get skin cancer). And remember back in the 60s people didn't think cigarettes were that harmful. Soon we'll find out that fat is good, or bad, or was it salt? Do we get to retroactively charge and refund people as scientific evidence mounts up?

ams61102• 14 years ago

I guess I disagree to some extent. Insurance is designed to protect you from rare but catastrophic losses. It's not (originally, anyway) intended to step in and pay for reasonably forseeable expenses (routine health care, for example).

A contributing reason that health insurance is so expensive is that people have been using it as a way to only pay 20% or less of expenses they know they are going to have.

By the way, if you buy your own insurance, and you have kids, or want maternity coverage, you pay more.

mikeash• 14 years ago

Even if it's fair, that doesn't mean we have to willfully help them do so.

lcargill993• 14 years ago

So is salt still bad for you this week?

Because you don't hire them to nag you, you hire them to finance certain forms of consumption. None of the numbers about even smoking have held up very well over time. "Lifestyle policing" is, like exercise mania, about vanity, not health. The exercise industry is there to pump up sports medicine spending while assuaging the Baby Boomer terror of death. Go look up which is faster rising, sports medicine or oncology. People have enough information about consumption choices to make informed choices. We should leave them be.

But the point is you hire them, they don't own you.

"The servant is the one that takes the money" - Lawrence of Arabia.

hexagonal• 14 years ago

Double blogspam: three-paragraph summary of a five-paragraph summary of a six-paragraph Economist article.

kirian4• 14 years ago

This is why I think something like Bitcoin is so vital. As physical cash is more and more being replaced by digital money the ability to have financial privacy is imperative. Think about all the data mining that would be possible if your every transaction ever was recorded in some database. The financial privacy possible with Bitcoin is one of its best qualities in my opinion.

rogerbinns3• 14 years ago

What is the difference between a bitcoin transaction and a theft? If you have total financial privacy then the law can't help, and a heck of a lot of people are going to want police/law intervention when they get robbed or any other injustices.

kirian4• 14 years ago

I don't really get what you are trying to say, why is a bitcoin transaction the same as a theft? A bitcoin transaction is the digital equivalent of a cash transaction, you can keep your financial privacy so that all your transactions are not stored in a database linked to your identity.

lolcraft• 14 years ago

That is almost the same problem as having to prove one has been robbed at a solitary place. The police would have to look for a fishy-looking transaction, and see if there were indicators that the machine were bitcoins resided was hacked.

Yeah, there would be slim chances of recovering the stolen money, but that already is the case with cash, given a smart enough robber. The right way to tackle that problem is not to make the stolen goods easy to track, but to have preventive measures. In Bitcoin's case, these measures would be good security practices.

geoka9• 14 years ago

They also mention going through social media profiles for insurance data mining. I wonder what this practice can lead to, if taken to an extreme.

For example, if I apply for credit and I can't provide some/most of the information they ask in the application, my request will probably be denied. Suppose in not so distant future I apply for credit and I don't have any social media accounts. Will I get denied because the bank could not dig enough information about me that they expect to be able to find about their applicants?

I don't like the world this social media craze is making us into. Being concerned about your privacy has always been a rule, not an exception. But that is changing now.

jandrewrogers• 14 years ago

Even if you can't provide some information on an application, they usually already have access to most or all of that information once you give them just a few facts so that they can uniquely identify you. And it takes surprisingly sparse facts to identify individuals for this purpose. They do not make you fill out the application for the information on that application per se; their sources of that info are more reliable.

In terms of risk mitigation, they are probably interested in comparing the extensive profile they have of you with what you put on the application. Significant discrepancies would be a reasonable red flag for risk, more so than having incomplete information.

maxerickson• 14 years ago

The only way that happens is if social networking behavior is a far better predictor than things like income.

That doesn't seem all that likely to me.

gcb• 14 years ago

All insurance companies are/will have to spend billions on data mining...

...says the guy who sells data mining.

kleiba• 14 years ago

Is this so unusual? I try to use my bank cards only at my bank's ATMs to keep profiling to a minimum. Sure, every now and then I end up at a register and discover that I am out of cash. Makes me feel uneasy every time. But it surprises me that this is on the front-page of a tech-savvy community - I thought we were all like that guy anyway.

jezclaremurugan• 14 years ago

This article is based on Economist's www.economist.com/node/21556263, reading that would show that the 'data mining exec' pays in cash because data mining software found a correlation between paying in cash and living longer.

mseebach• 14 years ago

Huh. So the title of the piece should be "Data mining exec doesn't understand the difference between correlation and causation". Which is pretty stupid.

chris_wot4• 14 years ago

In Australia, divulging this sort of information is against he law.

beaker• 14 years ago

The last thing I need is this kind of kruft polluting my brain causing me to continuously second guess myself. I have a hard enough time deciding if a purchase is "business related" or not, now I have to ponder the future implications that every purchase I make will have on my permanent record? Sorry but I can't play this game without encumbering my mind with an amount of stress that completely outweighs any future benefit this bit of clever purchasing discretion would bring. I surrender to our big data overlords and beg for mercy on my consumption footprint. Maybe someday someone will be able to provide a technical defense to all of this nonsense: e.g. http://www.theatlantic.com/technology/archive/2012/06/apple-...

pasbesoin3• 14 years ago

Here's another weakness in all this analysis: The (presumed) value of the data and analysis.

For years -- decades -- the medical professions have told us that "salt is bad". Now, we're finally learning that, specific and limited medical conditions or extreme diets aside, this is not so much the case. In fact, to little salt -- even when not "extremely" little -- may present significant problems.

Imagine insurers having had purchase/dietary information available during this "salt is bad" phase. How much harm might their resulting "persuasion" have caused? (As a simplistic argument/point.)

Taken to extreme, such controls are like "best practices". They can trend towards mono-cultures, which nature shows us tend to be fragile things.

Humans need to understand that our population has diversity, that this is a good thing, and that, wedged together in our cities, towns, countries, and onto this blue marble, to some extent we are "all in it together". Like it or not.

TL;DR: Some of these data analysis practices should spend more time worrying about their own garden.

P.S. I'm not against analysis, in general in life. I am against biased analysis warped by self-serving motivations.

Data Mining Exec Pays For Burgers In Cash To Avoid Insurance Company Snooping

Related Stories