The Micromort concept is explored thoroughly and very enjoyable in the book 'The Normal Chronicles' by David Blastland and Michael Spiegelhalter. It's well worth checking out.
Blastland was one of the creators of the excellent 'More or Less' on the BBC that looks at how numbers are used and mostly abused by the media, politicians and everyone else.
Spiegelhalter is a prof of stats.
The book addresses a number of the criticisms that have been leveled at the concept in comments here. They also talk about the 'microlife' which is what you get by exercising or whatever. They also acknowledge the problems with these ideas as well.
The Wikipedia article claims air pollution from living in New York or Boston is 30 times worse than the risk from living with a smoker, though that was in 1979. Presumably air pollution has become much less severe since then. Does anyone know the figure today?
These days people vastly overestimate the risk of second-hand smoke. Based on the numbers in this article living with a smoker is equivalent to smoking a pack of cigarettes every two years. i.e. not much.
Second-hand smoke does have some health consequences -- for instance poor air quality is strongly linked to childhood asthma. It's also certainly considered a nuisance by most people.
Anyway, I'm not surprised at all that living in a polluted city would expose you to far more danger than a smoker at home. 30x sounds fairly reasonable.
Indeed, the EPA has steadily increased the stringency of particulate matter (PM) standards over the years (these have an impact on car/engine design and fuel (sulphur and nitrogen content)); you can find more information here:
It should be noted that used in Thinking Fast and Slow context this micromort induce can be used to show how much we human have very hard time grasping low probability events.
All the lottery and insurance business is just a scam against us taking advantage of this miss computation of micro risks.
Or put another way: In your life you are optimizing over more than just money. Insurance is a way to reduce variance, and thus stress, and is therefore worth a nominal fee.
Another reason is what I believe economists call it "the law of decreasing marginal utility". Every extra dollar you have is (usually) worth less and less, since you pay for the things you want/need the most first, and money is only worth what you buy with it.
Your expected utility is not your expected dollars. The average person will pay more into insurance than they get out of it. However it's worth it for the small chance that disaster could strike and you could lose everything. Better to sacrifice a small amount of utility to prevent losing a large amount from becoming homeless.
My point is not that insurance in itself is a scam. It would be rational to pay a small fee to a common account and get back much more when needed. It is how most communities work.
My point is that the current insurance system and price is a scam, because it is much too expensive and covers to many silly cases. That's because it is very hard for human beings to evaluate the risk of something rare and its price. While this evaluation is exactly the job of insurance companies. It is the classic situation for a scam: the buyer can't know the real price. And we also have the symptoms: way too much money is spent on luxury in this industry, they have the most expensive tv ads, etc.
I am a credentialed actuary responsible for the pricing of insurance risk. It is absolutely true that the consumer cannot price his or her own insurance policy (and I can). However, the end result of this is not some nefarious scenario where insurance companies are charging consumers ten times the fair price to insure their car or home. There is a functioning market for insurance, and consumers are going to tend to select the lowest-price option from amongst their choices in that market. This means that if you overcharge your customers, you will lose them to a competitor. Systematic mispricing of policies relative to the competition will lead to adverse selection, which is even worse - the insureds that you were making money on leave, and the insureds that you were losing money on stay.
Because of these factors, the insurer's goal is to price your policy as accurately as possible. Profit margins in the personal lines are so thin that many insurers engage in what's called cash-flow underwriting. The only money they make on the policy is the investment income they earn on your prepaid premium.
On top of all this, insurance (especially insurance marketed to consumers) is heavily regulated. Rate changes and new rating plans are scrutinized by each state's department of insurance. These regulators function like you wish the banking regulators did. They have enormous authority and their relationship with insurers is adversarial.
I could go on at some length but I will cut it off here. Suffice it to say that insurance, particularly property/casualty insurance in the United States, is about as far from a scam as you can get.
Like any other product you buy, there is a reasonably transparent and competitive market for personal insurance.
Unlike any other product you buy, the price of insurance may not be excessive, inadequate, or unfairly discriminatory. This is the law. Unlike any other product you buy, personal insurance prices must be filed with regulators who have the power to block the sale of any insurance product that does harm to the public. Unlike any other product you buy, an entire profession is devoted to the pricing of insurance. You cannot even propose to sell an insurance policy if your pricing scheme has not been signed off on by a credentialed actuary. Those credentials are not easy to come by, and actuaries are bound by standards of practice that preclude us from doing anything unethical.
If my employer asked me to violate an actuarial standard of practice, I would quit on the spot, and I don't know any other actuary who wouldn't do the same. And finding an actuary willing to throw away their livelihood would only be the first step in the process of attempting to charge a consumer an excessive rate. There are so many safeguards in place that bypassing them all doesn't seem like it would even be possible, and even if it were, the insurer would not reap any rewards due to the force of adverse selection.
I don't often post on HN, and I know that the actuarial profession is not very well known, but arguing with an actuary about the pricing of insurance is like arguing with a heart surgeon about where the aorta is. I can tell you with the authority of an expert that you are mistaken.
Well I certainly have no special authority in the matter, but maybe Kahnemann has? He got a Nobel after all. To be true he didn't say insurance was a scam, he equated it to lottery in the fact that it plays on the difficulty to evaluate rare events probability, and also the psychological reward in buying a lottery ticket (temporary dreams of massive wealth, e.g. how many Ferrari will I buy...) or insurance (peace of mind).
I have friends in the insurance business and I think it is a very honorable profession in most of the cases but you can't wipe out 1) the door to door insurance salesman scamming fragile old people, 2) the possibility that, much like finance, the whole insurance profession is based on wrong equations that makes it apparently robust but inherently fragile (in the sense of Nassim Taleb).
The buyer never knows the real price in any industry. How many people know how much it takes to build a factory, to pay the wages of all the workers, transportation costs, material costs, etc, etc. Competition is supposed to lower the price to it's actual amount. I don't see why this isn't true in the insurance industry.
Insurance isn't a scam. You pay a little extra to ensure that you can get the money before you would have saved it (assuming you put your entire ensurance premium into savings) in the event that you (or your family, in the case of life insurance) need it. Sure, in infinite (or sufficiently high) number of lives/events, you'll always end up behind, but that doesn't mean it's a scam.
It is true, though, that many people fall into the trap of over-insuring. For example, people who could easily afford to replace their car yet pay for comprehensive insurance.
For risks you can self-insure against, you should do so. Otherwise you're just gambling against actuaries, and they're better at the game than you.
For catastrophic events, insurance is a completely sensible bargain for both parties.
Catastrophic event; reminds me of black swans. The more catastrophic, the rarer, but also well be the one with the most damage. As every insurance is itself insured, I'd say if there is something really catastrophic, all crashes and no-one gets money at all.
Financial insurance also build a corset to let the card castel grow higher, until it falls apart on the floor. Idea taken from Antifragile.
I agree with both your points there, but disagree with the premise that insurance is a scam - the insurance companies provide a valuable service, and they charge what people will pay - the way any reasonable business is going to behave.
From there, it's up to you to determine whether it's worth it. I personally don't have health insurance (though I live in a country with universal healthcare), but I have 3rd-party car insurance in case I accidentally drive into someone's Lamborghini, and it was fortunate that my girlfriend's dad had life insurance so that her mum had some money after he passed away prematurely last year. Just because money is being made doesn't mean it's a scam.
But why are you worried about driving into a Lamborghini and not the fact it would mean you've had a car accident and you could be dead?
This to me is where our perceived realities differ with real life.
Spending the money on defensive driving and/or car safety makes much more sense, but we can only see the future where the accident happens (And we survive) not all the accidents we can stop.
I have no idea about your girlfriends father and what happened but I'd prefer to spent money on not dieing than paying for what happens if I do.
non-lethal car accidents happen more often than those with casualties. And they often are fairly expensive, especially if you manage to involve more than your own car. That's not to say that you shouldn't drive carefully (and I guess your insurance company certainly raises the price once you start driving into other cars twice a month).
Even if all insurance companies are using fear to sell their product, they still have to compete with each other. The insurance industry is the boring financial industry, but is very vast with many nooks and crannies. The companies are making financial gambles estimating risk. And not every company has the same model or the same strategy. Some risks pay off, and sometimes they don't. And from what I understand, in the property and casualty business, the profit the companies make typically comes from investments of their capital and not their loss rates.
If you don't see the value of insurance, then don't buy it unless the law requires. And if you don't like the law, then do something about it.
But if you're family has money, or if you do, it would be smart to put a few bucks into your insurance policy and relax knowing that your capital is not at risk. And if you don't have a lot of money, its nice to know that there's cash available for you when something bad happens.
I gladly pay my life insurance premium because I want to know my loved ones will be taken care of. It's not a scam. I fully intend to pay in more than I get out of it (I hope!!).
Lottery isn't really a scam either if you derive pleasure simply from playing the game.
Yes but a fair insurance would be the exact price of the risk leveled among insured, plus a small processing fee. There should be not one penny left for luxurious offices or prestigious ads or art mecenat.
Lottery pleasure is empty. Play poker with friends, you may loose money but at least you get friends.
The problem pointed in Kahneman book is that we are irrational, and (AND) it's usually the less irrational among us who win.
For example, an insurance company has to pay for investigators, the number of which would likely scale linearly with customers.
Nice offices and an advertising presence are also likely necessary to entice and maintain customers, probably more-so an an industry like insurance in which there aren't many tangible differentiators between competitors (see Coke vs Pepsi).
I don't see why insurance should be any different than any other industry regarding these issues and also the existence and legitimacy of profits (assuming they weren't derived from fraud).
It is different in this that you can't evaluate the price of the thing you buy. It's like contemporary art if you will, but art is more obviously a scam, while insurance have an aura of solidity and serious.
"fair insurance", "There should be not one penny left ", "Lottery pleasure is empty"... You are stating a lot of opinions but not many facts. Also, why isn't it "fair" for insurance companies to turn a profit while it's fine for other types of businesses?
You can have a look in a book titled Thinking Fast and Slow. It is strongly backed by multiple experiences. Profit: it is ok to make profit, but where ifs the limit? Would you agree that companies should not take advantage of children weaknesses to force sell them useless thing (a lot do)? Or mentally deficient ones? I wouldn't, because somehow we are not in a jungle. And the point is that humans are not grown ups in front of rare events. Everyone fears an earthquake during the week following an earthquake, and completely forget about it the next month.
Moreover the fact that insurance are highly regulated just makes my point. We are defenseless and the law puts some limits because a scam is too easy to build.
Sure and that's why they are easy to beat. How many industries would not even exist if human were rational? And again, the most rational ones have less loss aversion, take more pondered risks, invest in what has real value (education) and usually win the game in the end.
Insurance is about managing risk. I willingly and happily pay extra money spread over many small payments to eliminate the risk of a single rare event (e.g. A car accident that's my fault) from totally ruining my life.
The car accident is different, insurance is mandatory because you also risk other people's lives.
And I don't say you shouldn't pay for your own protection, I say you can't evaluate the risk correctly and are very likely to be ready to pay too much. Which is how insurance make money.
Agreed, and the fact people here think insurance is not a scam is actually proving your point how they can't measure risks well even when pointed out to them.
If you house burns down, sure that's reasonably catastrophically bad, But if you house burns down, there's a fair chance it'll happen at night, and there's is a fair chance you'll die.
Don't spend money on insurance, spend it on stopping your house burning down.
If there's a risk of it happening then the money should be spent on reducing that risk, not wasting it on, if I survive at least I won't be poor.
> Don't spend money on insurance, spend it on stopping your house burning down.
Or do both (it's funny how IT types love those dichtomies).
And if you tell your insurance about your extra measures for fire safety, they likely lower your rates. They might even consult you about which fire safety measures are actually useful (and they have the data: data driven fire safety, how does that sound?).
Since fire insurance tends to pay for damage repair when things didn't burn to the ground, they're also useful in situations that are not as life threatening as your example. We had a warehouse burn down in town, and the neighboring house was affected in that the windows on two sides of the house melted enough to be useless. The cost of replacing all those windows can easily exceed an individual's short-term liquidity (a single event might even exceed the rates paid for the insurance over its life-time) - for the insurance that was pocket change.
Lottery - sure, that's a scam, barring the odd situation where a positive expected value is possible. Insurance on small events (i.e. the extended warranty from Best Buy) could be described as a scam, but homeowners insurance/car insurance isn't.
The whole reason insurance exists is because of differences in risk tolerance. What is a huge risk for me, such as a fire destroying my house, is a relatively small risk for an insurance company that is insuring against fires across the entire state. What I pay the insurance company for is to assume part of that risk.
Consider homeowners insurance, and more specifically fire insurance, in this admittedly contrived example. Suppose that in the next year there's a 1/1000 chance of a fire that will cause damage that will cost $100k to repair. That has an expected value of $100. Well, since $100k is a lot of money to me, I'd rather pay someone $200 than take a bet with an expected cost of $100, even though paying $200 has a negative expected value. That means I am risk averse for potential gains and losses on the order of $100k, and would rather take the more certain side of a bet, even if it means it has a lower expected value.
Take another example. Suppose I'm worried about losing or breaking my cell phone over the next year, and it would cost $500 to replace. AT&T charges $6.99/month for insurance on the phone. Over the course of a year that's about $84. And furthermore suppose there's a 1/20 chance that I'll lose/break/etc my phone during that year. Without insurance, the expected value of the loss is $25. Unlike the $100k example, $500 isn't that big a deal to me, so the insurance is a horrible deal for me, because I'm risk neutral for a $500 loss.
Of course, real life is more complicated. Homeowner's insurance protects against risks other than fire. Risks to the insurance company can be correlated - something on the order of the 1906 SF fire is a large risk, even to an insurance company, which is why there is reinsurance. There are deductibles that change the pricing. But still, as a simple example, that's how insurance works.
Yes, you describe exactly what is in Thinking Fast and Slow. So you are not rational, neither am I. But as you repeat the same patterns of risk aversion for all occurrences of a choice in your life, the sum is that you paid too much for insurance.
A rational agent would pay 100$ plus the processing fee for the home insurance. It would not pay for "piece of mind", just as it would not pay for the "excitement" of a lottery ticket.
We are not rational agent, but my point is that those who are closer to rational choice based on statistical truth are the one who win the game (in average).
Uh, no. Risk aversion is rational. It seems like you're equating rationality with risk neutral preferences, which I, and a lot of other people disagree with.
Besides, it's all about risk preference. As long as my preferences are consistent and transitive, I think it's safe to say they are rational. Take the fire insurance example. Suppose I have a job that pays $10k/year. I would gladly pay $200/year to avoid the possibility of a $100k loss. Those are my preferences, and as long as you can't a non-transitive loop, it's perfectly rational of me to have those preferences. In this case, I'd value the guaranteed loss of $100 to be a much better outcome than the risk of loss of $100k.
Except if your entire wealth is at stake, being risk adverse is not rational. That is the whole point of Thinking Fast and Slow.
Proof is easy: I give you the choice to play on the flip of a coin, one side you win $1200 other side you loose $1000. You'll likely refuse because of fear of loss. Then consider I propose the bet 1000 times in a year, under different disguise so you don't recognize it. You lost 200 000!
It is not rational to refuse a net positive bet it of irrational fear of loss. It is even more obvious for the lottery, because we would all agree that paying lottery ticket is irrational, right?
A lot of this seems weird. If I imbibe 0.5 liters of wine exactly once in my life I doubt there's any effect. Same for 2 cigarettes. For a lot of things there are thresholds that must be crossed before any effect is created.
It seems pretty clear that, to come up with these numbers, they took mortality figures from people with lots of exposure, divided their exposure down into small chunks, and assumed that everything scaled linearly.
I mean really, 1 hour in a coal mine and you get black lung? Absurd. Would 1 minute in a coal mine be 1/60th of a micromort? Would 1 second in a coal mine be 1/3600th of a micromort? Of course not, and of course not. Black Lung is caused by long term exposure.
You are not rolling a million-sided die every hour you are in a coal mine.
The statement >>1 hour in a coal mine and you get black lung<< is completely wrong.
By staying in a coal mine for 1 hour you increase the chance of death by 1 micromort, that by definition means if you did that for about a million times you would very probably die. If the average life expectancy is about 613000 hours, and if you spend every living hour in a coal mine you would still live every other lifespan.
Some of these fall into the old correlation-causation trap. People who wear leather jackets are more likely to get into motorcycle accidents, but putting on a jacket alone doesn't change your likelihood of an accident.
There's also contexts missing from many of the other statistics: where you drive and the kind of car affects the mortality rate of driving, for instance.
The micromort is still a useful concept in specific controlled conditions (radiation exposure), or as a first approximation of quantifying risk, but the general statistics should definitely be taken with many grains of salt.
A counterpoint - the first cigarette you smoke is probably the riskiest one, if you factor in the long-term likelihood of becoming addicted to cigarettes.
If you had two identical groups of a million people each and one group received 0.5 liters of wine per person, would you really be shocked that the group that got the wine had a single extra death?
I would be shocked to see any statistically significant[1] difference. The things about alcohol that kill you are cumulative. 3 drinks isn't going to kill your internal organs, not even 1 in a million times.
[1] I would be shocked if, give two groups of a million, both groups had the exact same number of deaths in a predetermined time period. 1 death difference between two groups that large is pure background noise.
I'm not sure I buy that, but either way you're not visualizing it quite right: both groups go on to have regular lives... it's just that one group drank an additional 0.5 liters [1]. Like you said, death by alcohol is a cumulative risk, so that means each and every drink must raise your risk a teensy tiny amount.
How about this way: you agree that if you drink enough alcohol it will kill you, right? So you could surely find someone who almost drank themselves to death. Perhaps they got liver disease and came very very close to dying, but survived. So there must be a line somewhere. Given a huge enough pool of people, it should be possible to find someone who came so close to death that a single extra drink could have killed them.
([1] It's a thought experiment. The two groups lead otherwise identical lives besides the additional alcohol.)
You mean what sort of relationship between alcohol and death do I propose? One that isn't a neat mathematical line for sure, particularly since it seems possible that moderate amounts of alcohol are actually beneficial.
If the relationship were linear, then you could spread that 500 milliliters out over 100 days and see the same effect, which is plainly ludicrous. The rate of alcohol consumption plays a large role, a single binge can kill you, slow and steady drinking can plausibly extend your life. A single extra drink is also going to do way more harm to a long-term alcoholic than it will to a young healthy Mormon; that Mormon is at absolutely no chance at getting cirrhosis, the risk to the alcoholic is non-zero.
The relationship between alcohol and mortality is something that you could probably fill damn near half a medical library with. "half a liter is a micro-mortality" is silly oversimplified nonsense. It might work for radiation exposure (http://en.wikipedia.org/wiki/Linear_no-threshold_model) but it is nonsense for alcohol.
Is there some amount of cigarettes that is beneficial too?
So long as the line is going up and to the right, it's specific shape only matters if you want to debate the micromort value, not whether the concept exists.
If you go up to the top of this thread, I think you will find that we are bickering about specific micromort values listed on the Wikipedia page.
Some of these things plausibly can be modeled as LNT. Radiation possibly can, perhaps airplane rides too. Some of them cannot be, like alcohol consumption.
> Isn't it obvious that alcohol can cause certain fatal diseases in proportion to consumption?
Its obvious that consumption is positively correlated to certain fatal conditions, but its not at all obvious that the risk is in proportion to (i.e., linearly correlated with) total lifetime consumption.
Yes, I would be shocked if an extra millionth died, because we have done many many studies of alcohol consumption, and they show that below a certain level (40ml a week or something?) there is no measurable effect on health.
That specific mechanism? Probably not. But I wasn't here to say the effects were linear. I was just pointing out that complaining about sample size in a theoretical experiment where you can multiply the size in your head is a nonsensical objection.
Perhaps. I was merely trying to be precise with my answer so that I wouldn't say something silly (expecting the exact same number of deaths). I wasn't trying to be bitchy, though I fear it came off that way.
If you consider the underlying mechanisms, it becomes clear. Cigarette smoke contains mutagens. For those who are genetically predisposed, it might only take a single point mutation (change of a DNA base pair) in a single oncogene to cause a tumour. So while unlikely, there is no need to suppose a threshold.
In low quantities of a toxin, up-regulation of repair mechanisms can sometimes improve outcomes (i.e. hormesis), but the risk of damage to critical genes still exists.
If a million people each smoke two cigarettes in their lifetimes, it's not impossible that one of them might suffer a cigarette-related health detriment.
The question is whether that person would suffer from that same problem anyway, due to some other cause. This is where statistical significance comes in.
If you smoke 2 cigarettes on one day, chances are that you are actually a smoker and therefore you fall in the category of people that are more likely to die from cancer.
The actual impact of the cigarettes you have smoked isn't really taken into account here, it is the fact that you are likely be a smoker that increases the number of microdeath.
The discussion here is interesting. One of the reasons I hardly ever submit Wikipedia articles to Hacker News is that very few Wikipedia articles have gone through enough thorough editing to be worth discussing among the professionally edited sources submitted here. I'll note for the record, as a Wikipedian, that the article talk page
for the article kindly submitted here suggests that the article still needs A LOT of work (as most Wikipedia articles doe) and that some of the statements now in the article may be flat wrong, misrepresenting the micromort concept as it was pioneered by Ronald A. Howard. See his presentation on microrisk analysis
Looking at the source, he seems to be lumping shorter plane rides in with the long-distance (safer) ones. The danger is largely in takeoff and landing, as far as I understand. I also can't tell if the source takes into account private jets, which are less regulated and likely more risky than passenger jets. I'd also imagine the statistic varies a lot of you limit yourself to flights involving the US and our colonies[1].
Or the source could just be wrong - he could have mistakenly used the chance of being involved in a crash, rather than the chance of dying. That would put it in the right order or magnitude, I think.
[1] With all the "evil empire" rhetoric flying around lately...
There are no reliable statistics for the safety of commercial air travel in the US these days.
It's not for lack of trying. It's simply that commercial air travel in the US is so safe that there isn't enough data to compile anything like a reasonable risk figure. Prior to the Asiana crash, the last fatal airliner crash in the US was in 2009, and there were no fatalities in 2008 or 2007.
So, yes, they're much safer than that number would imply, but it's hard to say exactly how much.
Furthermore each listed risky activity should be balanced with the micromort risk of not doing it.
For example suppose I take a plane, then ride a bike to go hiking for a week with friends. Maybe this trip will save me many micromort from health and mental perspectives. Having a deeper friendship is certainly also a net positive.
I think you mean 1-1/e (~= 1 - (1-10^-6)^(10^6)). But yes, the idea that you can just divide a statistic sampled over thousands of events linearly to get a meaningful statistic for one occurrence of the event is ridiculous.
Neither. 1/e would require that they all be independent random events which is not true in the general case (and mathematically 1/e would only be an approximation anyway). 1 would require that they be entirely dependent, which is also not true in the general case.
2 micromorts does not mean 2 separate risks, each of which has a 1 in 1 million chance of killing you. It means a single risk that has a 1 in 500 thousand chance of killing you. Similarly, it would be absurd to define 1 mort as the combination of 1 million separate risks, each of which has a 1 in 1 million chance of killing you. Whether 1 mort is guaranteed death or 1 - 1/e chance of death (my previous mention of it being 1/e was mistaken) is a property of how we define the unit to work, not a property of how correlated 1 million separate risks happen to be. Definitions that give answers other than 1 - 1/e or 1 don't make any sense.
> mathematically 1/e would only be an approximation anyway
If we're going with the definition that treats micromorts as independent, then it would make sense to define a mort as exactly a 1 - 1/e chance of death, and a micromort as a 1 - e^(-10^(-6)) chance of death, rather than a 10^-6 chance of death, but the difference is smaller than the degree of risk that we can measure or care about, so the difference is inconsequential, and it would still make sense to describe a micromort as a 10^-6 chance of death.
Blastland was one of the creators of the excellent 'More or Less' on the BBC that looks at how numbers are used and mostly abused by the media, politicians and everyone else.
Spiegelhalter is a prof of stats.
The book addresses a number of the criticisms that have been leveled at the concept in comments here. They also talk about the 'microlife' which is what you get by exercising or whatever. They also acknowledge the problems with these ideas as well.