Hacker News new | more | comments | ask | show | jobs | submit login
NY Insurers Can Evaluate Social Media Use If They Can Prove Why It’s Needed (wsj.com)
83 points by TuringNYC 24 days ago | hide | past | web | favorite | 88 comments

I would like to see the increasing granulization of insurance put to an end, because otherwise it won’t stop. There should be one pool, everyone’s in it, we spread risk, and the whole damn thing is simpler and easier to regulate. Sure, Mr Base Jumper is going to raise your premiums when he breaks his leg, so what. Maybe he doesn’t drink and so has less health affects than you. Or you’ll raise his premiums when it turns out you carried a gene for disease-X and got it at 45.

What we have now is an ad-hoc nonsensible set of what you can and can’t risk profile on, and every time we learn some new way to risk profile then we’ll have to decide again if we want to regulate it. On top of that this just increases the risk of insurance companies getting it wrong and needing to be bailed out. I’d rather we had a conservative, boring and simple insurance industry and just get off the treadmill.

Unfortunately “fair” insurance - one where everyone pays the same premium - is an unstable equilibrium without regulation. As others mentioned, self selection without regulation leads to the fair carrier getting the highest risk population, effectively leaving it no money to pay out. It is definitely a race to the bottom.

There is the other unstable equilibrium if insurance is fully fair but non compulsory - it’s participation bias. What’s the benefit of getting insurance now when I can get it right before I go to the doctor (health) or on my deathbed (life)? Obviously at that point there’s no insurance - risk dilution - it’s a subsidy.

Which makes the only way to have fair insurance is to require everyone to have it and enforce very simple underwriting criteria. It will also require most folks to pay more than they could otherwise. Unfortunately the US has historically been bad at attempting to pass a law with that kind of effect, though the ACA got closer than most attempts before the recent changes.

I believe this is more or less how it works in the Netherlands. Insurance is private and not directly paid by the state unless you're under 18 or you cannot afford it (but then subsidy is given to buy private insurance). Insurance is compulsory for everyone. It is well regulated. Insurers compete on premium options, hospital choice, etc.

In practice, monthly insurance premiums are around 110-130eur per month with a minimum deductible of 385eur. GP visits don't cost anything additional and only specialist care eats into the deductible.

We do have a technique for battling the second issue you point out at least: open enrollment periods. Of course they are kind of annoying and aren't generally viewed as acceptable for individual policies.. Maybe the some of effect could be mimicked with coverage start delays and/or extending the minimum length of coverage you can purchase.

If that is what you want then that is okay. You can pay for that type of insurance. I prefer to buy life insurance from a place that will charge me a lower rate because I don't drink or smoke. I assume that most people who don't drink or smoke would join my type of insurance as well, leaving your type of insurance full of drinkers and smokers. That would make your insurance cost more (the insurance is paying more) while mine would be cheap.

Now, if you are talking about requiring everyone to use the same insurance for the same price (maybe a government insurance) then I would want to ban drinking and smoking. Because now their actions directly impact my by rasing my insurance cost.

You can't have both types of insurance. If you try, the one that discriminates less just becomes the highest-risk pool. So telling someone to choose the one they prefer doesn't make any sense.

That's my point. How would you like it if you had to pay just as much in flood insurance as someone with a beach front property in Florida? If I had to, I would try to make beachfront properties in Florida illegal.

The other extreme is a scenario where insurers magically have perfect knowledge about your individual risk and insurance becomes completely pointless.

You will not receive exactly what you pay. Some people will receive much more, and their payouts are subsidized by the others who receive much less. That is the nature of insurance.

What's the difference between your question and "How would you like it if you had to pay just as much flood insurance as someone down the street whose house actually gets flooded next year?" The only difference is that you can't reliably predict things with that level of accuracy.

Insurance companies and low-risk insurance buyers are both incentivized to predict individual risk profiles as accurately as possible, but this gradually drives the system towards uselessness.

The only way insurance becomes pointless is under absolute certainty of future events. If you merely have personalized calculated risk using all the information available, insurance still makes sense. For example, you have total information about yourself, and there's a minimum price you'd be willing to sell car insurance to a copy of yourself, and it makes sense for you to buy it, too.

One could argue that the social value of insurance is to incentivize you to not take risks we know about, while protecting you from risks we don’t know about.

This assumes that conformity is the best way to increase value for society which I would argue is wrong.

The more information they have the better it is for me. The reason that I use insurance is to lower the variance significantly while only slightly lowering the expected value. Now, if they had an oracle that could see the future, then there would be no point in using them. But that is not the case now, nor do I expect it to be the case any time soon.

Currently I want the insurance company to know what my expected value is properly so that I can get a rate that is appropriate for my needs. Of course I would prefer that they make a mistake and assume that my expected value is lower than reality, but if they make to many mistakes like that then either they will go out of business or raise their premiums.

> There should be one pool, everyone’s in it, we spread risk, and the whole damn thing is simpler and easier to regulate.

Call me naive, but I also always thought that's kind of the point of it: spreading the risk, or solidarity to use a less fashionable term. People can be very sick through no fault of their own, and those who are healthier should look out for them. If you calculate that "fair share" according to exactly how sick people are, then what's the point, wouldn't that just be a needless middleman that gets paid upfront?

The idea is correctly price the consequences of lifestyle choices.

Cautious Prius owners probably wouldn’t be happy sharing equally in a risk pool with enthusiastic drivers of sports coupes. If insurance rates were standardized, there’d be a lot more sports cars out there, until the Prius constituency got them banned altogether. Charging different insurance rates gives all parties what they want at a cost that creates appropriate incentives.

You might be able to say something similar about health insurance and sedentary smokers vs. runners. If an extra hour of work is less economically valuable than a gym visit, the incentive structure should reflect that.

I agree that insurance rates should not be affected by luck. One of the reasons the preexisting condition part of the ACA is important.

> You might be able to say something similar about health insurance and sedentary smokers vs. runners. If an extra hour of work is less economically valuable than a gym visit, the incentive structure should reflect that.

If someone gets sick and thus becomes sedentary due to their illness now you’d be effectively discriminating against them due to their preexisting illness because they can’t go be a marathon runner.

So we better create a rule that somehow excludes ill people from being discrimated against like this. Oh and then we’ll force the sick person to go through a hellish process to prove it too. Etc.

Wrt health insurance we should just get off that train. It only leads to a worse and worse place.

1. It's hard to disentangle lifestyle choices from genetic ones. If someone is born with an insatiable appetite that means it's 10x harder for them to stay at a healthy weight compared to someone else is that a lifestyle choice of genetics?

2. Being healthy already has quality of life consequences. And the extra cost to insure those lifestyle choices are probably pretty low, and most of that cost is just shifting it from primary insurance to medicare. Because everyone dies of something, and a lot of times it's really expensive. Insurers care a lot about whether that happens at 64 or 66. But as a society it doesn't really matter, and the insured bears most of the impact.

In support of your comment, when talking about weight it's common to talk about "fast metabolisms" keeping people skinny. Studies have shown that metabolisms don't vary much at all. What does vary is how hunger signals work with different people. The people who complain about being too skinny and can't gain weight simply find it difficult to eat enough to gain weight. Their body tells them they're not hungry. If you force feed them they'll still gain weight.

I'm one of those people who gain weight easily due to always wanting to eat. I was morbidly obese at 19 despite coming from a household of healthy eating. I'm now 33 and have maintained a healthy weight for years but it isn't fun. For me to maintain a constant, healthy weight I need to be hungry. Despite 5+ years of maintenance, I am still hungry constantly.

> If insurance rates were standardized, there’d be a lot more sports cars out there

Is insurance that large a cost factor? I would imagine the increased cost of the car and fuel costs would still be strong incentives against upgrading to sports cars.

I thought the point of insurance is buying lower variance with higher expected costs. You can have that even if the insurer has perfect knowledge of the likelihood of whatever event you're insured against. If you have probability p to have to pay $X, I can sell you an insurance for $(p+my profit margin) * X. You get the benefit of lower variance at the cost of $(my profit margin) * X.

This is about life insurance, not health insurance.

Sure, but the increase of granularization in insure in general was mentioned in the comment I responded to.

You could say there's a spectrum, between stuff like healthcare and genetic diseases, and someone insuring their pinky finger not because they need it, but because they're vain and think it's an astonishingly pretty finger, and they're just that rich. But somewhere on that spectrum, there are also families that lose bread winners, and I don't see why solidarity wouldn't also apply there.

I can sympathize with that argument, but unless everybody is mostly the same (within, say, about 10%) it won't be worth it for some people to be insured.

How does this work when people want to buy different amounts of life insurance?

Pooling risk disincentivizes people from making the changes they need to be healthier. And people not bearing the full cost of their decisions makes it easier for them to displace the negative consequences of their choices on others.

I bust my hump in the gym a few days every week because I want to age well - I make an effort to eat healthy food and now I gotta pay for an cookie addict who would rather slurp down medication than make lifestyle changes? Screw that.

I made my choices, I can live with them. Others make their choices, they get to live with their choices. If they get to take good stuff from me to support their bad choices, I expect some reciprocation.

How much do you think pooling risk disincentivizes people from exercising or eating healthy? It's probably not a lot because the U.S. is one of the most unhealthy and least insured places in the West.

That cookie addict will be paying for your medicare and social security for 30 years, while you pay for their medicare and social security for 5.

Everyone dies, and healthy people live a long time, and can die very die slowly and expensively. He'll pay for your full time assisted living while you die from Alzheimer's at 95, you'll pay for some critical care after his heart attack at 65.

>That cookie addict will be paying for your medicare and social security for 30 years, while you pay for their medicare and social security for 5.

Why can't we just pay for our own if we have the means and give those without means basic, cost-effective care for things that can't be solved with lifestyle changes?

And why do I have to be forced into this? Why can't I just chose my relations and affiliations for myself? Besides, bold of you to assume that social security will be there for me when I'm able to retire. That money is gone, I'm not going to see a dime of it - it's just an extra 15% tax on my first 110k.

Cost, Quality or Access. Whatever you get on one comes at the sacrifice of another.

This is called the Iron Triangle of healthcare.

Furthermore, you can have public insurance that charges co-pays, high-deductibles etc. Public Healthcare does not guarantee 'free healthcare', and if it did it would do so at cost of access (rationing) or quality (cost-effective care).

You can do that today with an HMO or a High-deductible plan if you wish.

There are many issues with healthcare in the us, but this is not a solution for them.

This story is about life insurance, not health insurance.

That Iron Triangle assumes that you have an efficient system. If you have a wasteful system (say, you pay twice as much per person as other places do without any advantage in quality or access) then you can reduce costs without hurting other things.

What about (at least partial) cost reduction due to decreased bureaucracy of collecting data and segmenting population demographics and customizing premiums?

If you expand a government program 10x fold, its bureaucracy costs will expand >10x fold. Medicare as it is today is already bigger than its national counterparts, and it still has great administrative costs.

There are some benefits in government administration, but "efficiency" is not one of them.

You seem to be thinking about something different wrt cost sharing and public healthcare. Cost sharing structure is a separate matter to risk pooling.

Taken to its conclusion we should all be in risk pools of 1.

The only reason this works now is that people behave naturally (or at least natural enough) when posting on social media. If social media profiling becomes more prevalent (like this article suggests), everyone is going to be much more careful when it comes to what they're posting. All their real activity is going to be driven into pseudonymous (or at least private channels) that can't be easily monitored, and their "real" accounts is going to be so carefully curated that the signal to noise ratio is abysmally low. Or at least, this is my hope. I could be overly optimistic, and people will continue posting their true selves against their own interest.

It's helpful to contrast what sorts of information people will put on LinkedIn to see a nice demonstration of this... some people are a bit odd but most people keep the "social" portion of LinkedIn to a minimum since they are aware what sort of effect it can have on their livelihood.

Social media is just _super weird_ and I think as a societal experiment we're closing in on the realization that it doesn't have any value, people like to communicate but - just like in real life - people generally like to partition their lives. Social media companies gain no value in allowing a partition of activity, the only social media-ish company that seems to make an effort to actively support it is reddit and that's mostly because the focus is never on you as an individual.

People are pretty dumb.

There’s another side to this though, why should a low risk person be in the same risk tranche as somebody who has lots of guns in plain sight, is out drinking 3 nights a week or has high-risk hobbies?

Life insurers have pretty transparent interests. A middle aged guy with guns and heavy alcohol use is a more expensive risk for death. Right now they only use age, occupation, specific medical and smoking status.

Insurance is weird.

It operates by taking events that are unpredictable individually and grouping them together such that they are predictable in aggregate. The larger the groups are, the better this works.

At the same time, there’s a competitive advantage to using smaller groups, if you can reliably discriminate. Except that once you take it too far, you approach the individual case again and insurance ceases to have a point.

A life insurance company that buckets everyone together will lose out to one that divides people up into groups based on their risk. That one in turn would lose out to a company that has an accurate death clock they can use on every customer to provide perfectly individualized coverage. And yet the death clock company won’t be able to sell any insurance at all, unless they can keep it a total secret, because people could accomplish the same thing on their own.

I think you're missing something important in this.

If you make the fundamental assumption that what happens to people has a level of randomness to it (don't think we'll ever be able to escape this assumption) then insurance always will have a point even if your bucket size is the size of the individual.

This is because insurers provide a hedge against a certain amount of negative randomness over a time peroid. They calculate the risk on a policy as if they provided it to you a million times and then base their pricing on that. Since they do this across a body of customers the pooling system still functions, it just is more accurately priced.

So as long as there is a risk instability over time they eill have the ability to provide a service worth paying for.

You're right that there's always some utility as long as there's randomness, but that utility decreases the more accurate your predictions get.

Rather than a magical death clock, let's say that actuarial science gets to the point where it can figure out an individual's life expectancy ±5 years with a 90% hit rate. The possibility of being hit by a bus or having some unknown wild-card disease means you could die much younger, and you might beat the odds and die much older, but 90% of people die within their predicted range.

Now, Bob goes to sign up for life insurance and he's predicted to die at age 50. He really wants this insurance to ensure his family can survive without him, but the premiums are completely outrageous, so he passes. Alice is predicted to die at age 100 so her premiums are cheap, but she sees little point in it, so doesn't sign up.

In this world, I see a lot of accidental death insurance being sold, but almost no general life insurance. Some people would buy it, as it would have some utility, but not a whole lot.

Bob needs to participate in a group policy.

Do you know if the factors insurance companies can consider is regulated? It seems like it would have to be to ensure pools stay large.

Yes, definitely. This story is about the decision of one such regulating body.

I wasn't able to read the entire story because WSJ wants me to subscribe, but it starts saying they can use the information as long as it doesn't unfairly discriminate. I wonder what that means? The fact that it reduces the pool size at all seems like a negative.

That’s referring to things like racial discrimination, which is definitely illegal. Insurance companies sometimes try to use proxies for race or other protected classes to get around such restrictions, and that has resulted in a situation where any technique for evaluating people needs to show that it’s not being used as such a proxy. The goal isn’t to keep pools large, but rather to keep pool assignment fair in some sense.

People would need to have sufficient cash on hand to be able to accomplish insuring themselves.

With an accurate death clock, the premiums on a life insurance policy will cost even more than that.

Let's not forget about malice. If social media profiles become more important, they'll be broken into more often. The people doing ransomware can make threatening to raise your insurance rates a feature of their software.

Of course. In my country, the following is mostly common knowledge:

- Don't showcase more wealth on Facebook than your tax reports suggest you should have, in particular if you're behind on taxes or otherwise in debt to the government. It's been confirmed that (our equivalent of) the IRS looks at Facebook profiles to determine whether a debtor isn't lying about being poor.

- Don't post photos (or get yourself tagged on them!) that imply you were anywhere but in your bed, if you have a medical leave with "stay in bed" clause. Since the insurance is paid by the government (out of social insurance), officials do check whether or not you're lying to them - especially because medical leave abuse is a common phenomenon in this country (in big part because of many, if not most, companies that are abusing their employees, forcing them to go on fake medical leaves just to catch a breather).

People I know are extra careful about the two things above, and no doubt there will be more things like this to come.

People regularly confess very serious crimes on social media. It's quite astonishing.

A bunch of people have pointed out that, for all the criticism we’re leveling at China’s nightmarish social credit system, we’re well on our way to building the exact same thing but with corporations in charge. And while you can talk about how this is totally different because the corporations can’t put you in jail, that really doesn’t matter at all if I have to constantly self-censor all my words and thoughts or lose my health insurance. That’s an equally shitty outcome.

True, I never show my real name in any public platform unless it's my "official" one, which is whitewashed. I go through pseudonymous accounts every 2 years for my privacy. We already do have social credit, it's called your credit score. Self-deception is a powerful force. Everybody rise up to the polls! Oh wait, it's both sides that are abetting this crap. Just vote with the closest to freedom (including a social safety net) and don't worry, cause you can't control it.

Unless I’m mistaken, your credit score is your financial credit rating, not your social. It isn’t affected if you get a ticket jaywalking, or acting like a jerk.

There are interconnections between the two, but there are pretty clear distinctions.

A matter of time, I guess. Your "insurance score", as seen by insurance companies, provides information on how likely you're to be able to pay back a long-term loan, or even to be alive long enough to pay it back. And "insurance score" is based on increasingly deep insight into your lifestyle. It's only a matter of time before someone will have the bright idea to use "insurance score" as an input to credit score - and then, suddenly, you have social credit (sans the politics part).

Jail and losing health insurance are not equally shitty outcomes.

Correct. The latter is a lot more likely to kill you.

This story is about life insurance, not health insurance.

It's quite hard to avoid "disguised and illegal race-based underwriting", so expect some lawsuits soon. Even something as simple as a zip code is too biased, so an attorney shut it down in a project I was working on.

NY has some odd laws. On the one hand they are progressive (energy, human health), on the other hand they allow some very neo-liberal things to happen like where debt collectors can continue suing you, if your bank has one branch in NY, and now this nonsense.

It's almost as if NY has a huge financial industry that lobbies an otherwise liberal government.

Yet they overcome the auto industry lobby’s efforts and many others’...

The financial industry being in NY means actual voters. While a company cannot force their employees to vote in any particular way, they can make their employees aware that some politician did something... They can't actually say raises and bonuses were affected but if you read between the lines you can see the relation.

The auto industry isn't large in NY. When enough voters in NY decide the auto industry is doing something bad, no amount of lobbing from the industry will prevent the laws they don't like.

Politicians need votes. Money is not nearly as powerful as a voter who is informed.

Not particularly surprising after the recent NY bill which will require gun owners to share their social media logins and search engine history with the state. Sometimes the slope really is slippery.

Then spying, tracking, censorship of "offensive" or "unpatriotic" words depending on your political side for little gain of the little people. Of course, the state won't do anything to actually reduce mass shooting, they profit of fear and hate. History doesn't repeat, it is remixed.

Mass shootings are vanishingly rare. Your odds of being in one are microscopic.

Risk assessment fundamentally about probability and impact.

We took measures to get serious about drunk driving because it increased the probability of car accidents, especially at night,and was killing people.

On the flip side, we mandate tamper evident seals in medicine because the rare event of tampering hurt a small number of people, but undermined the citizens faith in the institutions they depend on.

The mass shooting issue is like the pill bottle problem. It’s a serious matter with an obvious solution. At the end of the day, it’s a problem unleashed by marketing — as hunting and shooting sport participation evaporated, gunmakers were left with a problem. Now we have batshit crazy people with political power, which is a genie that will be difficult to put back in the bottle.

> It’s a serious matter with an obvious solution.

This is a bold statement. Why do you think the solutions are obvious? Do you believe this is a subject without nuance?

No gun, no bullet, no shooting.

Look at the gun death rates in the northeast compared to the south.

However, it’s a shame that it happenes with frighting regularly in 2019! I probably won’t involve me but it’s a outrage.

Lots of things are vanishingly rare, the odds of being in a mass shooting are a lot less vanishingly rare then I am comfortable with. If the odds were up near where tornado deaths are I'd prefer it. The numbers I saw gave a 1/11125 chance per year of being killed in a mass shooting in the US.

That number looks wrong. It can't be the case that thirty thousand people die in mass shootings every year in the US.

Yeah that number is outright false for "mass shootings". This figure originates from the approximate 33k people that are killed by firearms per year. Of that number, somewhere around 23k are individual suicides with firearms. Of the remaining approximate 10k deaths, 80% are gang related incidents. Source is the FBI UCR.

There are roughly 30000 gun-related deaths per year in the U.S., but far fewer from mass shootings as you say https://fivethirtyeight.com/features/gun-deaths-mass-shootin....

You can also see the gun deaths broken down here. https://fivethirtyeight.com/features/gun-deaths/

I think they're probably quoting the chance over a lifetime, assuming rates stay the same. According to Wikipedia, 382 people died in mass shootings in 2018.

Gang violence is often included in mass shooting statistics, and rightfully so, but depending on the depiction or commercial are not shown as such. On the other side of the coin, often times, gang crime is taken out of mass shooting statistical data, making it look far less likely, but more realistic to rich, white, suburban targeted audiences that politicians tend to advertise to. Neither is a lie, but both data sets are "massaged" to suit political narratives popular with particular bases.

Only if you use the perversely tight definition of 4 or more people killed (not including the shooter), not in a robbery, not as part of gang violence, not as part of domestic violence, and not including sponsored terrorist events, on the same day in close geographical proximity.

As soon as you drop some of those requirements the numbers increase dramatically.

In fact they do not

"Hey, why did my rate go up?"

"I'm sorry, we cannot offer you a CleanSocial(TM) discount because your social media footprint is insufficient to properly inform us of your risk profile"

I don't imagine this will go over very well with most people.

However, headlines like this are a good reminder to practice good social media hygiene; I've been careful about what I post and have a habit of deleting old Facebook posts from my teenage days.

That reminds me I’d like to erase my HN comments over a few years old.. oh wait.

(Seriously, can we please make it easier to remove comments/accounts from non-archive sites? There’s not a single person on earth that can survive a Willie Horton style campaign where everything you say is examined without context - it’s an abdication of responsibility on those running a community)

I wonder how compliant HN is with Article 17 of the GDPR[1]. It's the "right to erasure".

[1] https://gdpr-info.eu/art-17-gdpr/

Personal data is not the same as comments posted on a public forum, but its not to deter the most determined actors - its more to indicate that I no longer want to have any association with the comments.

Hadn't heard of outline before -- very cool. Thanks for sharing this link.


I bet that you're into free content too, when you're not on your high horse. The public domain is a great thing and we should feel entitled to it.

Don't forget that a paywall only enriches the owners of the site; it does nothing to ensure that journalists are compensated. If you want to fund artists, do it directly; paying an artist's owner and hoping that the owner is a good person is a losing strategy.

Looks like it's time to either fudge all my social media with unicorns and rainbows or start operating purely behind pseudonyms?

When social media use and friends you have there affects the price of the services, it's already a type of social credit.

Ah, yes, this is when the real dystopia begins

Joke’s on them. They’ll be lucky to uncover a handful of uncorroberated shadow profiles and spam catcher email dead ends that flout “real name” policies.

Meanwhile, I scarcely have a need for insurance. Come up and get me.

> Meanwhile, I scarcely have a need for insurance. Come up and get me.

I don't live in the US but I have 5 insurance policies - home insurance (mandated by my bank for my mortgage), car insurance (mandated by having a car), health insurance (optional but less so if you live in the US), Travel insurance (traveling for work mandates I'm covered by a policy) and life assurance (the only actual optional one) - should I have to provide all of these groups my social media details even though I don't "want" to have them?

Further, whos to say they decide that by me not participating in "normal" activities online, that they won't decline or offer me a substantially worse policy?

Joke's on you! As we've seen in the credit rating system if there is a fake profile sharing your information the company is pretty likely to assume it's just you - I'm not certain how being known as a v1agr4 reseller will impact your insurance prices but it'll be fun to find out together.

Applications are open for YC Summer 2019

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