My favorite stat about micromorts is that motorcycle travel and canoe travel pose the same risk when measured by distance. That is, one mile of motorcycling has the same risk of death as one mile of canoeing.
I was hoping to learn something like this. I was hoping to see other comparable activities though. Driving is actually much less dangerous than I expected in comparison to riding a motorcycle but I think there are other factors, mostly that injuries are not considered, only fatalities.
I've ridden many many miles on a motorcycle and maybe one mile in a canoe and this statistic surprises me. I would have assumed canoeing would be much, much safer!
That's partly because you travel so much more distance in a motorcycle per unit of time. (Probably at least 20x depending upon how the statistics were derived.)
It also points to another problem with this sort of data. While there are certainly deaths in challenging whitewater, a lot of canoeing deaths involve combinations like alcohol+no PFDs+cold water at night.
I think of all canoe trips as "gentle trips on the lake" when I read this. I suppose most canoers think of motorcyclists as "that guy doing triple the speed limit on the crowded interstate"...
I don't remember the exact numbers anymore, but in Australia at least the statistics suggest that if you:
Don't ride under the influence, Don't ride without a license, Ride with a helment, and Don't ride an unregistered bike,
Then you have decreased your risk by more than 50%. So using the articles methods, you will be exposed to less than 200 micromorts.
Servicing your motorcyle and having taken a motorcycle saftey course just once also increases your liklihood of survival.
Another key part of the dangers are where you ride, being more likely to die in the CBD than anywhere else, and likely related, just under 50% of accidents are caused by other motorists than the rider. Many riders seem to think it's skewed much more toward cars being at fault, it's useful to know that's not the case.
A final tidbit I recall is that the biggest risk group is 40-60 year old, returning riders. Not younger new riders as you might have expected.
The interesting part of all of that though is knowing such broad stats can not truly reveal your personal risk factor, and that there is a lot you can do to help your cause.
> I don't remember the exact numbers anymore, but in Australia at least the statistics suggest that if you: ..., Don't ride without a license, ..., and Don't ride an unregistered bike
How does lack of a piece of paper increase your chances of dying?
I suppose there is some correlation there, but stating it like that makes it sound like causation since there is a causal relationship between deaths and the other two things presented in the same statement.
That was the understanding I had, that not being in the group of people who do those types of things is what lowers your risk. But all of this is inferred by "what we know about the people who died" which as the article alludes to, is not a perfect measure.
That's the important bit to remember for a lot of these motorcycle statistics - so a squid with no gear might think "whatever man, I took the motorcycle safety course and I wear a helmet! I can lane at a delta of 40mph no problem!" Rather, all of those things combined (helmet wearing certificate-bearing registered-bike-owning) are the traits of a safe rider anyway, not necessarily the things that cause them to be a safe rider.
It does not, it tags you with attributes usually associated with the safer group. That's how statistical correlation works. It does not have any power, not in this sad amagic tyrannical causal universe we find ourselves in. Its a pity really, I'd sacrifice a horizontal vestal virgin for a piece of paper like that.
There are other factors too. The less powerful your bike, the safer you are. That one is actually a pretty significant factor.
I was interested in riding a motorcycle once, but I found that even if I assumed I was at the safest end of the spectrum, my risk was still many times that of driving. And the comparison is not apt because many of the risk factors I mitigate on a motorcycle I also mitigate in a car, for e.g., by being a careful driver, minimizing the amount of driving I do, driving slowly, and never driving drunk. So it's not like the motorcycle has become safer relative to the car. It's just that both skew less risky for me.
So I decided I would not do it. Maybe when I'm older and other risk factors are more pressing.
I had a similar experience. The impetus for me was always building bikes rather than actually riding them though, so the compromise was getting a project car instead! I haven't calculated the micromorts of circuit racing mind you..
I probably put in about 100k miles on motorcycles over the years and never had an accident/incident. I do not agree that a less powerful bike makes you safer (I only rode sports bikes btw). The slower bikes (250cc or so) generally had much worse brakes and couldn't accelerate if needed. I found 600cc bikes were the easiest to control and "safest." That of course is only true if you are going the same speed as the counterpart (which is difficult especially if you're riding a 1000cc).
What if Shark attacks are so low, because people pay attention to the risk? Just like airplane travel is safe because of all the regulations, not just out of itself?
And I dislike measuring the micromorts per population, it should be divided by those people actually going swimming/surfing, just like riding x miles with a motorcycle increases your risk of dying by y micromorts, going swimming for a hours should be increasing your mms by b.
I think it's just hard to estimate. How many visits-to-the-beach go into that shark attack number? It'd be useful to know the per-visit risk for the same reason it'd be useful to know the per-mile risk for driving, but is harder to estimate:
"It’s important to be aware, however, that being based on population-level data, the micromort is not a measure of your personal risk. For example, the risk calculation for fatal shark attacks is based on the average risk across the whole population.
So it effectively estimates a risk based on the assumption that all Australians swim in the deep ocean a similar number of times per year. But if you live in Alice Springs, you would not expect your risk of being killed by a shark to be the same as for a surfer who lives on the coast. Similarly, if you only wade in water up to your knees and don’t swim in deep water, your personal risks would be different."
Which sort of amounts to treating oranges as apples in the case.
I always find it super silly when people apply population data to themselves. There's always much better data to go on if you know a bit about the individual.
As others have said, possibly to make very disparate events comparable across a population, like the risk of skydiving vs. a shark attack. Unfortunately, this results in an almost useless measure for any particular individual when the probability of being in a risk group is small. Here is a breakdown of the risk for a random person in the US dying from skydiving in 2012 [0]:
(1 death / 100,000 jumps) x (3 million jumps / 350,000 jumpers) x (350,000 jumpers / 313 million people)
The last two terms give us the prior P(jump) ~= 0.01 and the first is the conditional probability P(death | jump) ~= 0.0001. The non-skydiver's risk factor is, of course, zero while the skydiver's risk is about 100x greater than that of the average person (who doesn't actually exist). As you can see, the population measure isn't very informative for either group.
As a statistician once told me: "the odds of running into a grizzly bear are miniscule, but if you're the one who runs into it, they're 100%." So it's important to remember that when evaluating odds and probability. Many things are unlikely to occur, but they do occur to some. I realize this is only tangential to your question, but wanted to put it out there.
But wouldn't it make sense to group people who are risk aware together?
Like 90% of a population will live risk free lifes, and only 10% have high risk.
I mean, depending how risk friendly 10% of the population are, they could screw up the result and then it seems all people of that population are crazier than most of them really are...
Lots of answers about apples-to-applesness, and such, but from a very practical viewpoint, the "said activities" part of "participating in said activities" is hard to define and measure. Should you condition on swimming, swimming in the ocean, swimming in the ocean in california, swimming in the ocean in the Santa Cruz area, or swimming in the ocean at Año Nuevo (or even some part of that beach)? However you decide to break it down, do you even have that data? In stats, lots of questions begin and end with "do you have data on that?"
Pretty easy to divide through by how much time the average person spends in ocean water to get your hourly risk. If such a number could be found, anyway.
Most shark attacks turn out to be attacks of people surfing, boogie-boarding, or otherwise making themselves look like seals or similar mammal from the shark's perspective. Your risk while surfing is much higher than the average swimmer, never mind the average beachgoer who doesn't spend much time in the water.
yep, and as I mentioned above, what percentage of those surfers are actually regulars in sharky places? We're talking about a very small slice of the total surfing population.
Out here in California, most people never go more than ankle deep in the water if that. Those who do surf usually only do it occasionally, in very un-sharky places.
How many people really surf regularly in sharky places like Ano Nuevo, Russian River Mouth, Bolinas, and so forth? My guess is that you should probably be dividing the number of shark attacks by something measurable in the thousands, not 36 million.
Whatever that number is, it's still damn unlikely. But still, if you're in the habit of surfing in known shark feeding grounds while dead seals are floating around, I don't think shark attacks divided by the total population of California is a good reflection of your risk.
There's a wonderful (and heartbreaking) essay, "The Median is not the Message" by Stephen Jay Gould about cancer survival statistics. It expresses some of those points beautifully.
If you know more information, you can always change your estimates. Even if you're a surfer in a "sharky" place, what percentage of those are there at the same time as a shark? What percentage of those are chased by a shark? Bitten by a shark? If you're not one of the few that actually get bitten, then you're safe from death.
Agreed as a spearfisher (and a spearfisher in a sharky place), I am well aware that an overwhelming majority of attacks happen to spearfishers or related activities (abalone divers). It is not uncommon for a freediver with some sort of food to have an encounter with a shark and most people that have been doing it for a while have several stories. It comes with the territory but those who do it accept the risks. If you factored out spearfishers, abalone divers and surfers the numbers would look drastically different. The reality is it is very, very rare for some random person in the water to experience an unprovoked attack.
They could make units of "kilo-terrors" and "mega-terrors", units of a thousand times the risk of dying in a terror attack and a million times the risk respectively.
"Interestingly, kangaroos (approximately 0.1 micromorts) pose a risk of death that is similar to that posed by sharks, but our cute national emblem does not evoke quite the same fear in us."
The fear of running into Kangaroos is real. Hitting an 80kg buck at 70km/hr looks like this:
" crashing a car into a kangaroo is not considered a kangaroo-related death"
Roo ran into me. Ran up an embankment. I've been in other situations where the kangaroos (multiple) are window level travelling along roads at speed. Chance of kangaroo-related death is there.
That part confused me. It is suggesting that flying around the world is safer than sitting in a chair to watch tv.
1 micromort for 10K km of air travel, and 1.3 micromorts for sitting in a chair in your own home.
Considering all the standing, sitting, stairs, ramps, overheads, etc. we do while traveling, this must be referring ONLY the the flying portion of flying... very misleading.
The silliness is even clearer in the finale, the part about "every day you get out of bed...". You're virtually certain to die if you don't get out of bed for enough days.
Their skydiving estimate and comparison shows some faults with this approach. If 10 people die a year from x. But those 10 people were people pushing the limits of skydiving by doing something unusual or intentionally chumming waters to swim with sharks the numbers are really all wrong
Their skydiving estimate and comparison shows some faults with this approach. If 10 people die a year from x. But those 10 people were people pushing the limits of skydiving by doing something unusual or intentionally chumming waters to swim with sharks the numbers are really all wrong
> you would divide the number of people who have died due to a shark attack each year (on average three to four each year based on recent data) by the population of Australia
This method is meaningless, as it assumes the risk of dying by shark attack is constant regardless of your behavior (e.g., daily swim in chummed waters vs never entering the water).
The wikipedia article does a better job of micromort-per-unit-of-exposure.
The problem with a "unit" of risk like this is that probabilities don't simply sum like this.
Assuming the probability of dying in each activity is independent, planning to do a 7000-micromort activity A and then (after appropriate recovery time) an 8000-micromort activity B is not an overall activity with 15000 micromorts of risk. (The math is close enough when talking about single-digit micromorts, but it doesn't work in general.)
Pr(dying when doing A followed by B) = Pr({doing A and dying} or {doing A and surviving and then doing B and dying}) = Pr(doing A and dying) + PR(doing A and surviving and then doing B and dying) = 7000 / 1000000 + (1 - 7000 / 1000000) * 8000 / 1000000 = 14944 / 1000000
The larger the risks or the longer the series of events, the less accurate the assumption of linearity. This also doesn't take into account the fact that dying in extreme activities is not independent; by failing to die in A, you've demonstrated some small amount of survival fitness.
If the micromort was sold as an expected number of deaths, then I wouldn't have as much of a problem with it. (My major problem is that you can't die more than once, so probability seems to be more of a concern than expected number.) The article calls it "a one-in-a-million chance of dying", though.
For example, playing Russian Roulette once is about 166667 micromorts. Playing it 6 times is 1 mort of risk, but the probability of dying after 6 plays is not 1. (It's 1 -
(5/6)6, or about 2/3.)
Really, anytime we move faster than we can walk and are on the ground (car, bus, bike, etc.) we're likely to be involved in an accident that may kill us.
Approximately half a million people died of malaria in 2015, and that was after a long trend of decreasing deaths (it used to be much worse)[1]. That's deaths from malaria, not cases of malaria. If we naively assume everyone on Earth has equal odds of dying from malaria, that means sharing our planet with mosquitoes carried a risk of 59.2 micromorts for every human on Earth in 2015. It's probably gone down slightly since then, but it's still a lot worse than sharks or skydiving.
Obviously, your real risk is lower in most places, but dramatically higher if you live in sub-Saharan Africa.
That's very high for the overall population, but I'm guessing the actual number is close to 0 for say Americans who never left the country (we don't have malaria deaths contracted from inside the states, do we?). If you are someone that surfs 4-5 times/week during the season, I guarantee the odds of you dying from a shark are much higher than dying of malaria.
The most fun thing is that such risks are generally additive. Not being tied to number of participating in a given activity makes it a bit better for this purpose.
That said, instead of frequentist approach a bayesian one would be more relevant. This would allow including fuzzy participation data like never, once per year, once per week, daily etc.
Is anyone else really fascinated by how this article exposes how human beings are simultaneously irrational and yet, also completely rational when it comes to fear?
Humans as a whole are irrational when it comes to fear because their fear is almost never proportional to the actual risk. We're afraid of the one-in-a-million occurrences, but not the one-in-one-hundred. We don't (yet should) fear the mundane and the superficial. Humans always strive to be rational, so why are we so irrational when it comes to fear?
And yet, arguably this is also simultaneously a perfectly rational reaction to fear. Think about it. If humans were hard-wired to be completely rational when it comes to fear, i.e. our fears are proportional to the risks involved, we would be unable to function in our daily lives. We'd be paralyzed by fear from the moment we wake up. We wouldn't drive cars, or step into bathtubs, or eat fatty hamburgers. The explanation as to why humans are afraid of shark attacks, but not driving can be as simple as this: it allows us to exist.
>> We wouldn't drive cars, or step into bathtubs, or eat fatty hamburgers.
Is it really rational to do those things? I put a non-slip mat in my tub, eat less meat, and generally avoid driving if at all possible. I'm certainly not unable to function nor continually paralyzed with fear, but my risks of death from some very common causes are greatly reduced.
The risk of skydiving into a shark's mouth are extremely low. Ironically, if you were assaulted by a skydiving shark, you would probably die of a heart attack.
64 comments
[ 2.5 ms ] story [ 146 ms ] threadIt also points to another problem with this sort of data. While there are certainly deaths in challenging whitewater, a lot of canoeing deaths involve combinations like alcohol+no PFDs+cold water at night.
Then you have decreased your risk by more than 50%. So using the articles methods, you will be exposed to less than 200 micromorts.
Servicing your motorcyle and having taken a motorcycle saftey course just once also increases your liklihood of survival.
Another key part of the dangers are where you ride, being more likely to die in the CBD than anywhere else, and likely related, just under 50% of accidents are caused by other motorists than the rider. Many riders seem to think it's skewed much more toward cars being at fault, it's useful to know that's not the case.
A final tidbit I recall is that the biggest risk group is 40-60 year old, returning riders. Not younger new riders as you might have expected.
The interesting part of all of that though is knowing such broad stats can not truly reveal your personal risk factor, and that there is a lot you can do to help your cause.
How does lack of a piece of paper increase your chances of dying?
I was interested in riding a motorcycle once, but I found that even if I assumed I was at the safest end of the spectrum, my risk was still many times that of driving. And the comparison is not apt because many of the risk factors I mitigate on a motorcycle I also mitigate in a car, for e.g., by being a careful driver, minimizing the amount of driving I do, driving slowly, and never driving drunk. So it's not like the motorcycle has become safer relative to the car. It's just that both skew less risky for me.
So I decided I would not do it. Maybe when I'm older and other risk factors are more pressing.
Cool!
> I do not agree that a less powerful bike makes you safer (I only rode sports bikes btw).
You don't need to agree. It's a statistical fact that power and danger are associated. May not be causal. Probably isn't, in fact.
And I dislike measuring the micromorts per population, it should be divided by those people actually going swimming/surfing, just like riding x miles with a motorcycle increases your risk of dying by y micromorts, going swimming for a hours should be increasing your mms by b.
"It’s important to be aware, however, that being based on population-level data, the micromort is not a measure of your personal risk. For example, the risk calculation for fatal shark attacks is based on the average risk across the whole population.
So it effectively estimates a risk based on the assumption that all Australians swim in the deep ocean a similar number of times per year. But if you live in Alice Springs, you would not expect your risk of being killed by a shark to be the same as for a surfer who lives on the coast. Similarly, if you only wade in water up to your knees and don’t swim in deep water, your personal risks would be different."
I always find it super silly when people apply population data to themselves. There's always much better data to go on if you know a bit about the individual.
(1 death / 100,000 jumps) x (3 million jumps / 350,000 jumpers) x (350,000 jumpers / 313 million people)
The last two terms give us the prior P(jump) ~= 0.01 and the first is the conditional probability P(death | jump) ~= 0.0001. The non-skydiver's risk factor is, of course, zero while the skydiver's risk is about 100x greater than that of the average person (who doesn't actually exist). As you can see, the population measure isn't very informative for either group.
[0] http://adventure.howstuffworks.com/skydiving8.htm
But wouldn't it make sense to group people who are risk aware together?
Like 90% of a population will live risk free lifes, and only 10% have high risk.
I mean, depending how risk friendly 10% of the population are, they could screw up the result and then it seems all people of that population are crazier than most of them really are...
Out here in California, most people never go more than ankle deep in the water if that. Those who do surf usually only do it occasionally, in very un-sharky places.
How many people really surf regularly in sharky places like Ano Nuevo, Russian River Mouth, Bolinas, and so forth? My guess is that you should probably be dividing the number of shark attacks by something measurable in the thousands, not 36 million.
Whatever that number is, it's still damn unlikely. But still, if you're in the habit of surfing in known shark feeding grounds while dead seals are floating around, I don't think shark attacks divided by the total population of California is a good reflection of your risk.
https://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=...
Higher in Middle East, but I lack precise enough death data for those countries.
The fear of running into Kangaroos is real. Hitting an 80kg buck at 70km/hr looks like this:
- https://www.flickr.com/photos/bootload/5037959035
- https://www.flickr.com/photos/bootload/5037959047
- https://www.flickr.com/photos/bootload/5038593608
- https://www.flickr.com/photos/bootload/5147428118
Roo ran into me. Ran up an embankment. I've been in other situations where the kangaroos (multiple) are window level travelling along roads at speed. Chance of kangaroo-related death is there.
> sitting on a chair (due to the likelihood of falling off it) increases your risk of death by approximately 1.3 micromorts.
Is this per day? Per hour? Per year? How does this compare to laying down? Standing up? Walking around?
1 micromort for 10K km of air travel, and 1.3 micromorts for sitting in a chair in your own home.
Considering all the standing, sitting, stairs, ramps, overheads, etc. we do while traveling, this must be referring ONLY the the flying portion of flying... very misleading.
This method is meaningless, as it assumes the risk of dying by shark attack is constant regardless of your behavior (e.g., daily swim in chummed waters vs never entering the water).
The wikipedia article does a better job of micromort-per-unit-of-exposure.
https://en.wikipedia.org/wiki/Micromort
e.g., Travelling 17 miles (27 km) by walking (accident) = 1 micromort
But it still doesn't account for differing risk levels, like walking on an indoor track vs walking in city traffic.
And it doesn't include shark attacks.
Assuming the probability of dying in each activity is independent, planning to do a 7000-micromort activity A and then (after appropriate recovery time) an 8000-micromort activity B is not an overall activity with 15000 micromorts of risk. (The math is close enough when talking about single-digit micromorts, but it doesn't work in general.)
Pr(dying when doing A followed by B) = Pr({doing A and dying} or {doing A and surviving and then doing B and dying}) = Pr(doing A and dying) + PR(doing A and surviving and then doing B and dying) = 7000 / 1000000 + (1 - 7000 / 1000000) * 8000 / 1000000 = 14944 / 1000000
The larger the risks or the longer the series of events, the less accurate the assumption of linearity. This also doesn't take into account the fact that dying in extreme activities is not independent; by failing to die in A, you've demonstrated some small amount of survival fitness.
For example, playing Russian Roulette once is about 166667 micromorts. Playing it 6 times is 1 mort of risk, but the probability of dying after 6 plays is not 1. (It's 1 - (5/6)6, or about 2/3.)
Frequentist approach like this micromort unit ignores conditional probability.
Obviously, your real risk is lower in most places, but dramatically higher if you live in sub-Saharan Africa.
[1]http://www.who.int/gho/malaria/epidemic/deaths/en/
https://www.cdc.gov/malaria/resources/pdf/fsp/cdc_malaria_do...
That said, instead of frequentist approach a bayesian one would be more relevant. This would allow including fuzzy participation data like never, once per year, once per week, daily etc.
He has some good stuff on this for more everyday activities like drinking beer, being 5kg overweight for a day and smoking a cigarette and things.
https://understandinguncertainty.org/microlives
Humans as a whole are irrational when it comes to fear because their fear is almost never proportional to the actual risk. We're afraid of the one-in-a-million occurrences, but not the one-in-one-hundred. We don't (yet should) fear the mundane and the superficial. Humans always strive to be rational, so why are we so irrational when it comes to fear?
And yet, arguably this is also simultaneously a perfectly rational reaction to fear. Think about it. If humans were hard-wired to be completely rational when it comes to fear, i.e. our fears are proportional to the risks involved, we would be unable to function in our daily lives. We'd be paralyzed by fear from the moment we wake up. We wouldn't drive cars, or step into bathtubs, or eat fatty hamburgers. The explanation as to why humans are afraid of shark attacks, but not driving can be as simple as this: it allows us to exist.
Is it really rational to do those things? I put a non-slip mat in my tub, eat less meat, and generally avoid driving if at all possible. I'm certainly not unable to function nor continually paralyzed with fear, but my risks of death from some very common causes are greatly reduced.