Join 3,433 readers in helping fund MetaFilter (Hide)


Individual free choice but collective predictability?
December 28, 2008 1:54 PM   Subscribe

Can someone explain why despite the fact that every one of these incidents occurs at a unique time and place, each involves a complex history of events and personal decisions leading to its very unlikely outcome, that the death toll on the roads year-on-year is so predictable? See here and here.

Why doesn't it jump around from 100 one year, to 8000 another year for example? What is it about human nature in particular that makes the rate of error leading to death so predictably in the same narrow range? This is not a question about law enforcement, personal practices or the nature of death. It's more a question of individual free will, statistics and aggregate regularity. Does anyone even understand the existential puzzle I am trying to unravel here or is it really just unmysterious?
posted by zaebiz to Religion & Philosophy (47 answers total) 5 users marked this as a favorite
 
I think it really comes down to the fact that every car has the same basic controls, more or less, and that human reaction times and ability to perceive the environment around their cars stays, statistically, more or less the same.

Are you asking if the thought processes between humans should vary so much that human error should be impossible to judge?

I think I read once that human drivers make 32 decisions per second, and they make about 2000 mistakes on each trip. However, they recover and keep driving. It's when mistakes pile up that an accident happens. I think it was from an article my mom read aloud to me while I was learning to drive. I don't know if that helps or not.
posted by mccarty.tim at 2:06 PM on December 28, 2008


You could apply the same thing to pretty much any human phenomenon --- consumer bankruptcies, college degrees earned, babies born, etc. In a given population, of course the circumstances leading to each individual event will vary, but they tend to occur in predictable numbers. Are deaths on highways any different from any other predictable number?
posted by jayder at 2:18 PM on December 28, 2008


mccarty.tim: "Are you asking if the thought processes between humans should vary so much that human error should be impossible to judge?"

Yes I guess I am asking something like this. Your explanation about human error rates may be the right answer but for some reason I find it unsettling and unsatisfying. In that split second an error was made, the individual driver could have made an alternative decision which would have avoided the event leading to death. Doesn't that kind of suggest we don't really have free will or choice? We are all just machines and most of us are lucky enough to avoid low probability deaths? But it's more than that too - each of these events are so unlikely and so unique - yet their aggregate occurrence is very predictable. To me it just says something about collective freedom. Maybe it's not as troublesome as I think.
posted by zaebiz at 2:18 PM on December 28, 2008


jayder: "Are deaths on highways any different from any other predictable number?"

No not at all. And yes Im sure you are right about those other statistics. Road deaths are just so instantaneously determined (and so tragic and random). That's kind of what holds the fascination and highlights how unlikely the aggregate predictability seems.
posted by zaebiz at 2:23 PM on December 28, 2008


- each of these events are so unlikely and so unique - yet their aggregate occurrence is very predictable.

The average amount of miles driven and hours spent per person in cars traveling at speed multiplied by the range of human reaction times, wakefulness, sobriety, and other behaviors, make auto accidents hardly unlikely or unique.
posted by kuujjuarapik at 2:32 PM on December 28, 2008


It might help to think of a situation where the individual cases are the result of complex sets of events but which don't involve a tangled idea like free will. How about the rate of mechanical failure of aircraft (not due directly to human error)? The reasons vary widely and hugely complicated sets of circumstances contribute, but the rate no doubt stays similar year-to-year. Because even if the factors involved were completely unpredictable (random), it'd just be like rolling dice, which of course is predictable in the aggregate.
posted by abcde at 2:32 PM on December 28, 2008


Law of large numbers and the central limit theorem start to answer your question. I think these are pretty well the things you're asking about - not why do people crash their car, but why are aggregate crash statistics so predictable.

The assumptions that allow these theorems to apply to car accidents are that
  1. car accidents are independent events,
  2. the rate of car accidents per mile driven doesn't vary much,
  3. and there are many car accidents and very many miles driven per year.
If you dispute or are curious about any of those 3 assumptions, be more explicit about that and we can try to clarify it for you.
posted by ikkyu2 at 2:35 PM on December 28, 2008 [10 favorites]


zaebiz: we may have an infinite variety of choices available to us at any given time, but only a very small number of them make much of any sense if we are, for example, hurtling forward at 60 miles per hour in a metal box. The statistical consistency accounts for the probabilities of all the different choices. Aggregates display behaviors not shown in their constituent members. Sometimes the whole is more than the sum of its parts, in this case and many other cases of very large groups of humans, the whole is less than the sum of the parts.
posted by idiopath at 2:45 PM on December 28, 2008 [1 favorite]


most of us are lucky enough to avoid low probability deaths

If such a death is low probability, you don't need to be very lucky to avoid it.

What is it about human nature in particular that makes the rate of error leading to death so predictably in the same narrow range?

It's nothing to do with human nature... you'll find similar effects with any other (sufficiently large) set of independent phenomena. By the central limit theorem, the occurrence of such phenomena will tend to approximate a normal distribution, which has a more tightly defined confidence interval with increasing sample size. Since a lot of people spend a lot of time driving, you get an appropriately large sample size for this effect to be seen, because any variations will tend to be "ironed out" across the vastness of this segment of the human population.

The premise is satisfied by that auto accidents are (to a reasonable degree, across the whole sample size) independent, such that my driving my car off a cliff in Colorado wouldn't conceivably cause anyone in Australia to rear-end the car in front of them. This is different from systems like the stock market, where past price fluctuations will tend to influence future price fluctuations, and you can see things like massive sell-offs and the accompanying tremendous variability in price over an annual timescale.
posted by 7segment at 2:52 PM on December 28, 2008 [1 favorite]


The motion of a particular molecule of gas is reasonably random and complicated, given all the other gas molecules and funny shaped objects and so on it's bumping against. Yet the behavior of a few quadrillion molecules of gas is quite predictable, to the point where in most cases we can predict what they will do with 100% reliability.

There's less people on the roads than a quadrillion, so we can see random fluctuations.
posted by aubilenon at 2:58 PM on December 28, 2008


ikkyu2 and 7segment have it. This is basic probability / statistical theory -- the wikipedia links others have provided are great, and a Stats 101 course at your local college would give you a great foundation if you need to understand it from first principles.
posted by randomstriker at 3:03 PM on December 28, 2008


I kind of restated the four previous answers, so let me address your concern of free will.

As idiopath mentions, the figures you cite indicate aggregate figures, and do not necessarily relate to yourself personally. I (think I) probably have a lower chance of dying in an auto accident than you, because I don't drive. You're free to make any number of decisions about where, how often, how fast, and what type of vehicle you drive, all of which influence your personal chances of dying this way. The figures you cite are (in effect) an average of these factors over all other people on the planet, on which the impact of your personal choices is negligible.

In your first followup, you use the term "collective freedom," but I'm really not sure what you mean by this.
posted by 7segment at 3:05 PM on December 28, 2008


ikkyu2: "the rate of car accidents per mile driven doesn't vary much""

But why?

7segment: " any variations will tend to be "ironed out" across the vastness of this segment of the human population."

Ok but why does it all iron out to between 40,000 and 45,000 per year in the US for example? Why not 85,000? Why not 12,000? I mean think of the complex topography of the road system, weather conditions, the reasons why people drove "that day", the emotional events leading up to the circumstance, whether he had an itch right at the wrong time he had his cell phone in his hand, the history of mechanical servicing of the vehicle, whether the driver had just happened to have her eyes in the right spot at the right time to avoid the accident etc. Aren't you saying that in each of us is this fixed probability potential for death which we can do nothing about no matter what paths we choose or how hard we focus? That's kind of troubling isn't it? Or am I making something out of nothing or misinterpetting something?
posted by zaebiz at 3:10 PM on December 28, 2008


Doesn't that kind of suggest we don't really have free will or choice? We are all just machines...

There's more than just this suggests it. Free will as a concept is kinda bankrupt and/or incoherent, especially if you posit a causally closed universe. How would such a thing even work?

Are we machines? There's a strong chance the answer is "yes", albeit incomprehensibly (to us) complicated ones, and machines that operate under the illusion that "they" are somehow in control.

If you want to explore the philosophical aspects of this, try here.
posted by bonaldi at 3:28 PM on December 28, 2008


See also "Regression towards the mean".

Ok but why does it all iron out to between 40,000 and 45,000 per year in the US for example? Why not 85,000? Why not 12,000?

Because 40K-45K is where the center of the probability curve currently lies. If you change the system state in a fundamental way that number will rise or fall.

Which has happened. Cars used to be built so as to be more dangerous during accidents, and even with a much smaller total population in the US, the yearly fatality rate used to be higher than it is now. (Also because drunken driving used to be much more common and much more tolerated.)
posted by Chocolate Pickle at 3:29 PM on December 28, 2008


If you roll a dice six billion times, you'll get six about 1 billion times. Each time you roll it it may seem random, but actually it's the result of a huge number of tiny, arbitrary events, just like being in the car.

Aren't you saying that in each of us is this fixed probability potential for death which we can do nothing about no matter what paths we choose or how hard we focus?

Er, the chances of death are 100%. The chances of dying at any one moment are certainly under our control. We could, for example, chose not to drive. Or we could be sure not to drive drunk, avoid talking on our cellphones, pay close attention to other drives, and so on. And if everyone did this, the number of accidents would go way down.

The problem is that not everyone is going to do this.

And keep in mind the factors you mentioned: the topology of the roads, weather, people's tories, and so on don't really change that much from year to year.

One thing that has changed recently was the price of gas. When gas went up so much this summer, people actually started to drive less and guess what? there were a lot fewer deaths on the road, because factors change.

But it certainly isn't true that each person has a 'fixed' likelihood of dying on the road independent of their behavior.
posted by delmoi at 3:34 PM on December 28, 2008


Ok but why does it all iron out to between 40,000 and 45,000 per year in the US for example? Why not 85,000? Why not 12,000?

If you drastically altered a variable, then yes, the death toll could spike or plummet significantly.

For example, if suddenly one year America's infrastructure crumbled (due to war, natural disaster, etc.), and its roadways became disastrously dangerous, the death toll would likely spike.

Conversely, if suddenly one year, due to an abrupt oil embargo against the US by foreign nations, drivers were forced to abandon their cars and ride in electric vehicles that traveled shorter distances at lower speeds, the death toll would likely plummet.

But if variables remained relatively stable, so would the resultant death tolls.
posted by terranova at 3:52 PM on December 28, 2008


Post Hoc Ergo Propter Hoc.

You assume a lot of things—even one—cause accidents. That's knowledge we don't have in any capacity. Philosophers have spent their lives trying to figure out what it means for one thing to cause another.
posted by trotter at 3:55 PM on December 28, 2008


What is it about human nature in particular that makes the rate of error leading to death so predictably in the same narrow range?

Sure, sure... people are all different. Each of us is a precious snow flake, yadda yadda...

Look, put down the bong, take What the Bleep Do We Know out of the DVD player, and back away.

In answer to your rather chat-filtery question, it seems that you're over stating the metaphysics of the situation.

People get into cars. Cars go on the road. People have limited options while in this environment. Turn the wheel, hit the gas or break... that's about it. It's also a mostly controlled environment: cars have four wheels, roads are paved, you drive between the lines. Add to the mix other very common issues: we are easily distracted, sometimes drunk, talking on the phone, driving at night, on icy roads during the winter, etc.

The point is that traffic accidents are the result of people making very limited decisions in a mostly controlled environment.

Freewill... pfth!
posted by wfrgms at 3:56 PM on December 28, 2008 [1 favorite]


Another factor to consider is congestion and its impact on speed, which creates a self-regulating check on deaths caused by accidents. Fewer cars = less congestion and therefore higher average speeds, resulting in a relatively small number of accidents but each with a relatively high probability of death. Add cars and congestion rises, causing speed to decrease... giving an end result of more accidents but each with a lower probability of death. Consequently, a big range of cars on the road will result in similar accident totals.

Summer gas prices decreased all key factors--the number of cars, VMT (vehicle miles traveled) and average speed per car, causing dramatic reductions in overall deaths on the road.
posted by carmicha at 4:01 PM on December 28, 2008


Your explanation about human error rates may be the right answer but for some reason I find it unsettling and unsatisfying. In that split second an error was made, the individual driver could have made an alternative decision which would have avoided the event leading to death. Doesn't that kind of suggest we don't really have free will or choice? We are all just machines and most of us are lucky enough to avoid low probability deaths?

When you're talking about the cognitive processes underlying the larger macro-level process you call "driving," how far down do you really expect choice to go? You stop at a red light; maybe you chose to stop and would have run it if you were, say, rushing to the hospital and could see there was no oncoming traffic. But you don't think you chose to recognize the redness of the light (or your experience of the light, or your sense impression of the light, depending on how philosophical you want to get on this). Likewise, you do a great deal of conceptualization in representing that state of affairs as a moment of choice in the first place, but you don't choose whether some of those levels of conceptualization happen or not (or happen correctly). Can you really look at a red circle and not see it as red or circular at this point? And at any rate, one who makes an error and has an accident doesn't seem to be choosing to have the accident, do they? They mistakenly think the light is still green, or forget to look and check at all, and just run right through it without making another choice at that instant. Surely you don't think it's choices all the way down. (And if you do, surely you're wrong.)

If the notion of yourself as a complex machine with lots of lower-level cognitive processes that are not directed by higher-level processes unsettles you, think seriously about the alternative. If I am not a collection of such processes that underlie the possibility of my conscious choices, then there must be some part of me that is free and undetermined by other features of the world. But if so, it will also have to be undetermined by my beliefs, desires and other psychological states since it is supposed to be free of all such determinants and the beliefs, etc., themselves are largely determined by genetics, environmental factors, etc. So if there's some part of me that isn't the sort of "machine" you allude to, then what would it be, and more importantly, why would I think of it as "me"? It determines my actions without the influence of my beliefs, desires, environment, etc. Why wouldn't that strike me as random, or some sort of alien presence behind my actions? I certainly want to identify with my choices, think of them as authentically my own, but that strikes many philosophers as a way of saying that they emerge from a regular, predictable and familiar set of psychological dispositions - the sort of "machinery" you mention. (This is hardly my original idea. See David Hume on free will for a classic and readable source.) This is just to give you a sense of a determinist's or compatibilist's response to the philosophical quandry behind your question. I used to be unnerved by the thought of not having free will till I started to think that the alternative was actually more bizarre.
posted by el_lupino at 4:04 PM on December 28, 2008


Aren't you saying that in each of us is this fixed probability potential for death which we can do nothing about no matter what paths we choose or how hard we focus?

I don't think you can draw that conclusion from the aggregate data. You'd have to have separate groups.

Overly simplified example:

Suppose your aggregate figure is X, and you have equally-sized groups with no overlap. Your groups and their data are as follows: unfocused drivers (X+10), hard-focusing drivers (X-10); drivers who habitually tailgate (X+8), drivers who maintain proper separation (X-8); speeders (X+20), non-speeders (X-20); drivers who do not adjust their driving based on current weather (X+25), drivers who do (X-25); etc.

Each category's good drivers have a lower death toll, but your aggregate figure still balances out to X.
posted by CKmtl at 4:17 PM on December 28, 2008


CKmtl: "Each category's good drivers have a lower death toll, but your aggregate figure still balances out to X."

Ok thanks - yes I can see where I was wrong about the fixed internal probabilities. So the regularity of the statistics doesn't rule out the possibility of our individual ability to change our destiny.

The regularity still troubles me. For example, suppose statistically we could show that given the fairly consistent nature of world affairs, the apathy of the populace, the nature of the domestic political system US troops will be involved in war on average one year out of five. So what's the sense in me protesting and getting upset when it happens provided it fits with the existing probabilities? Ok this is probably now becoming a bit like chatfilter - I guess philosophy is a bit like that anyway. The thread has cleared up a few things in my mind about statistics and the likeliness of regularity in collective unlikeliness. Thanks.
posted by zaebiz at 4:38 PM on December 28, 2008


To address your latest comment, it's possible (and probable) that your protesting (and the protesting of others) is what would make the yearly rate of war involvement one-in-five instead of one-in-four. Get it?

So I guess to answer you in general, your free choice/will affects the outcome to you, but factors into the average in order to keep the average the same predictable number every year. As a whole, humans are very predictable, which is really the basis for society and life as we know it -- we assume all sorts of things like all the people on the road know how to drive and the cashier isn't a maniac, etc. The choices you make on the road mean that you will not die in a tragic accident, but because of statistics and the predictability of human nature (and maybe the limited choices others have mentioned), someone else is basically guaranteed to make choices that will mean they will have an accident, and the aggregate of everyone's choices comes out to the same thing every year.

If I'm understanding what you were asking about "collective" vs "individual" freedom, then I think I've tried to answer that. Individually you have free will, but as a society, all the individual choices add up to a predictable outcome.

Although, onto a very different vein, just because a choice is predictable doesn't mean it's not "free"... are you an Aquarius or something? ;)
posted by thebazilist at 5:08 PM on December 28, 2008


So the regularity of the statistics doesn't rule out the possibility of our individual ability to change our destiny.

With either the assumption of free will, or predestination, the statistics are immaterial to "destiny".

Statistics are an observation, not a prescription. We watch the number of crashes per year, and the number of drivers per year. The probabilities we get out of the statistics are also observations. We note that for every 100,000 drivers, 6 (or whatever) will probably be killed in a car accident. That doesn't mean that out of any 100,000 people you select, 6 will die; it doesn't mean that every year 6 out of every 100,000 people will die. It simply means that, during the time we've been watching, that's the proportion of people who've died.

You're making a mistake in assuming that past incidents determine future incidents. However, without significantly changing an environmental variable, it is probable that we will continue to observe similar aggregate behavior. I can roll a die 6 billion times, and get a one every single time; it isn't likely (in fact, the odds are very bad), but that doesn't mean it's impossible.

Likewise, you cannot take an aggregate probability and apply it haphazardly to an individual. Road death statistics apply to the probability that 6 somebodies per every 100,000 somebodies will die. It does not say that I will die 6 times out of every 100,000 trips; I'm an objectively better driver than most people I know (I know, isn't everybody?). It also doesn't say anything about the 12 year old who steals their parents' car; he's an objectively worse driver than most people I know.

Doesn't that kind of suggest we don't really have free will or choice? We are all just machines and most of us are lucky enough to avoid low probability deaths?

As a(n amateur) student of neurobiology, I'd have to say that's pretty accurate. Who "you" "are" is determined by genetics, conditioning, imprinting, etc. which are all physical processes. You are an individual and unique snowflake... but, you can no more do something outside your personality than a cinderblock can generate electricity. Even when you do something that "isn't like you", you're doing something that your mental machinery is already programmed for. It's arguable that you can "metaprogram" your programming, and change your personality--but, even that has to come from somewhere inside you, no?

When applied to driving, I think there's even less an argument for radical free will. We all have the same interface to the car; the same environment through which we travel; the same rules to obey; frequently the same (or highly similar) training. Human reaction times are all about the same (even race car drivers and soldiers, so the numbers say). It's interesting to note that if you change those variables, you can get different crash rates. I understand that European drivers are less likely to have an accident than American drivers--my guess is that they receive better training.
posted by Netzapper at 5:27 PM on December 28, 2008


Also, you do know that actuaries can tell with horrifying accuracy when you're most likely to die, right?
posted by bonaldi at 5:28 PM on December 28, 2008


Doesn't that kind of suggest we don't really have free will or choice?

Note: I'm going to mostly answer this question, and that will veer me into a side bar. If you feel it's too off-topic, please flag it. I don't see how to discuss this part of your question without going deeper into a discussion of free will.

It's one nail in free-will's coffin. It doesn't (by itself) invalidate free will. If Bob, Jeff and Biff all choose chocolate over vanilla, that might make you suspicious as to whether or not they were acting freely. On the other hand, maybe they have totally free will and just happened to all make the same choice. Of course, this becomes less and less likely as more and more people choose chocolate. Still, given only that choice -- even if a billion people make it -- it's possible that will is constrained in some ways and free in others.

But I don't believe in free will. Like I said, your aggregate-behavior thing is one data point that makes me suspicious. There are many more. For instance, there are recent experiments in which scientists have been able to predict simple choices before people were aware they were making them.

Of course, some people would argue that just because you're unaware of a decision, that doesn't make it not-free. To me, that's crazy talk, but I've heard many people say something like it. I'd say that if freedom means that my unconscious mind is making choices I'm not consciously aware of, that's not a sort of freedom that interests me. (And it's a perversion of the word "freedom.")

Other people talk about how no-free-will implies a deterministic universe and since Quantum Mechanics posits randomness (on sup-microscopic levels), the universe isn't deterministic. To that, I'd say that if my choices are random (or influenced by randomness), that's also not free will as I care about it. Ask someone whether he'd like to choose whether or not to eat cockroaches or to flip a coin to decide.

To me, the biggest problem with free will is lack of causation. What is causing the free choice? If something is causing it, it's not free. (If you're choosing chocolate because I'm holding a gun to your head, you're not freely choosing it.) So a free agent must be uncaused. (An unmoved mover.) I would need some pretty extraordinary evidence to buy that there's a free agent -- something that isn't being buffeted about by deterministic forces or randomness -- inside my head.

On a more nuts-and-bolts level, we know about many external forces that do at least affect choices: one's genes, one's health, one's level of poverty, one's upbringing... none of those constraints prove that there isn't a somewhat free agent in the gaps, but the more we learn about human nature, the more those gaps narrow. Free will seems to be running out of space.

Incidentally, I've heard this bit of nonsense all my life: "If free will doesn't exist, does that mean that we should pardon criminals?"

If free will doesn't exist, we can't freely choose whether or not to pardon criminals, so the question is pointless. If free will doesn't exist, we will or we won't pardon criminals, depending on how causal forces sway us.
posted by grumblebee at 6:11 PM on December 28, 2008 [2 favorites]


So what's the sense in me protesting and getting upset when it happens provided it fits with the existing probabilities?

You're falling pray to the same fallacy. You're essentially saying, "If free will doesn't exist, why should I CHOOSE to protest?" If free will doesn't exist, you won't choose to protest or not to protest. You will do what you're "destined" to do.
posted by grumblebee at 6:14 PM on December 28, 2008


grumblebee: "You're falling pray to the same fallacy. You're essentially saying, "If free will doesn't exist, why should I CHOOSE to protest?" If free will doesn't exist, you won't choose to protest or not to protest. You will do what you're "destined" to do."

The problem is that even if I do have free will (which I am still hanging on to), then the statistics would suggest that collectively, people as a group don't act like they have free will.
posted by zaebiz at 6:46 PM on December 28, 2008


So you're thinking that you're the only one with free will?
posted by grumblebee at 6:52 PM on December 28, 2008


grumblebee: "So you're thinking that you're the only one with free will?"

No I am thinking of a scenario like CKmtl said above. We all have free will - but most of the time we don't exercise it.
posted by zaebiz at 6:55 PM on December 28, 2008


I'm confused. Why do we not exercise it? Do we choose to not freely choose? If so, isn't that a free choice? Or are we in jail but every once in a while the jailer lets us out so that we can make one or two free choices -- and then he locks us up again?
posted by grumblebee at 7:13 PM on December 28, 2008


The problem is that even if I do have free will (which I am still hanging on to), then the statistics would suggest that collectively, people as a group don't act like they have free will.

The statistics don't suggest that (at least, they don't suggest that people aren't acting freely), you're misunderstanding them as others have said.

But even if they did that's not "the problem" -- if the statistics showed that the problem would be that you want to hang on to "free will" despite it being an ultimately bankrupt idea.
posted by bonaldi at 7:18 PM on December 28, 2008


grumblebee: "I'm confused. Why do we not exercise it? Do we choose to not freely choose? If so, isn't that a free choice? Or are we in jail but every once in a while the jailer lets us out so that we can make one or two free choices -- and then he locks us up again?"

(I don't know if these kinds of conversations are allowed in ask.metafilter but presumably a moderator will delete these chatty parts if someone objects. )


Check a few of the above posts by thebazilist, Netzappper and el_lupino see what I mean. Others like wfrgms have an alternative explanation which is still compatible with free will. Basically we run on a sort of subconscious overdrive that doesn't require conscious thought. Most of driving doesn't require an awareness of the events around us and the decisions we make.

You might be right that there is no free will in which case I have no control over the fact that I disagree with you. :)
posted by zaebiz at 7:28 PM on December 28, 2008


We all have free will - but most of the time we don't exercise it.

My example says nothing about a lack of exercising free will; it was only about the misapplication of statistics. The aggregate figure is a way of saying "All things being equal, this is what happened". But all things aren't equal, so applying that figure to careful drivers is misleading.

Say the incidence of food poisoning is 1 in 1000. That doesn't mean that everyone has a 1:1000 chance of getting sick, regardless of what they do. The risk will be higher for people who engage in riskier behaviour (eating mayo that's been left out in the sun, handling food after wiping their ass without having washed their hands, cutting up their salad on an unwashed cutting board that was just used to cut raw chicken) than for those who practice good food safety. The latter are not immune from the risk, as they could buy tainted food at a store or in a restaurant, but their risk is lower.

The difference between the two is not one of free will, necessarily. It's one of wise and foolish actions. Those who drive or cook foolishly can still be freely choosing to do so.
posted by CKmtl at 7:36 PM on December 28, 2008


First off, free will* is a religious or philosophical idea. Its not science and its not statistics. This is what is confusing you. A similar question could be "Explain evolution in terms of creationism." Your question is *ahem* fatally flawed. People who believe in religious free will have no problem with determinist chimps or beetles. Why the exception for humans driving cars?

As far as the why part of your question goes: There is no denying that humans are incredibly predictable animals. This is why electioneering and marketing work so well.

* the term free will is a generic term and there exists many, many definitions.

On preview:

Basically we run on a sort of subconscious overdrive that doesn't require conscious thought. Most of driving doesn't require an awareness of the events around us and the decisions we make.

Why limit this to driving? Human cognition is going to x amount subconscious all the time. My subconscious is posting this message as much as my conscious. Im not thinking 'the M key is next to the N key - move finger." The idea to answer your question came as much from a subconscious desire as a conscious one.

I think there's a simple funcational answer to your question: humans can only perform so well with the task of everyday driving. So many people drive under so many conditions that its easy to predict accidents with historical data.
posted by damn dirty ape at 7:38 PM on December 28, 2008


el_lupino, wfrgms and Netzapper are all pooh-poohing yr idea of "free will", on my reading!

I'm not sure how to get closer to an answer from here. There are two aspects to yr question: statistics that surprise you, and how this affects yr idea of free will.

Other people have dealt with the statistics to show that results like this have no bearing on the compulsion or otherwise of the individual drivers involved. The free will part is a lot more involved, and quickly leads to fairly heavy philosophy, though the wikipedia page isn't an awful place to begin.

Can you write off one or other part? The two really don't fit well together, and you won't understand the statistics by pondering determinism; nor will you gain an argument for incompatibilism from car crash stats.
posted by bonaldi at 7:43 PM on December 28, 2008


Descriptive statistics look at what has occurred and give you the numbers on that. Predictive statistics look at what the variables are like now, what has occurred in the past, and then attempts to predict what will happen in the future. But let's suppose you change a variable--then the whole prediction goes to shit.

Look at the stock market--it's basically a giant exercise in statistical prediction, with enormous numbers of people putting enormous amounts of time and resources into it. There are a couple of key trends that seem to hold true--the whole "average 10-12 percent growth" standard, etc. But despite that, relatively small variables being changed can cause a huge disruption in predictions and screw the whole thing up (see current conditions).

So statistically, maybe your country does go to war every 5 years. Your protesting may not change that fact--or you might find exactly the right message, through brains or chance, that resonates politically and socially. It's unlikely, but statistical predictions--and hell, even descriptive predictions--leave room for chance and error.

Sure, 40-45,000 will die on the roads each year. Did you really think over the Central Limit Theorem link above? Driver error rates are fairly well known, and are the result of our mental and physical limitations. This may not go along with the concept of free will, but just because you will yourself to take flight doesn't mean it's going to happen. You can will something as much as you want, but you can't exceed the physical limitations of your body or the constraints of the universe.

Perhaps humans are predictable as a group--if you followed 538 or even payed attention to advertising, this wouldn't be a surprise--but that doesn't mean that free will doesn't exist. Think of it this way--suppose I offer you a thousand bucks or a kick in the crotch. You're free to choose either, but come on--it's easy to make that decision. Likewise, it's easy to make decisions that make life easier or more pleasant for you, and just because people don't doesn't mean they aren't free to do so.

And those guys and gals who would take the crotch kick? That shows you why predicting human behavior is a little more challenging than you would think.

/Oh, and this is chatfilter.
posted by Benjy at 7:48 PM on December 28, 2008


bonaldi: "el_lupino, wfrgms and Netzapper are all pooh-poohing yr idea of "free will", on my reading!

Depends on how you define free will doesn't it? Do you have free will when you are asleep for example? Anyway I wasn't trying to ascribe my viewpoint to theirs. I was trying to attribute my evolving understanding to elements of their explanations which I believe are compatible with a notion of individual free will.

"I'm not sure how to get closer to an answer from here. There are two aspects to yr question: statistics that surprise you, and how this affects yr idea of free will ... The free will part is a lot more involved, and quickly leads to fairly heavy philosophy"

Agreed.
posted by zaebiz at 7:52 PM on December 28, 2008


Tolstoy devoted a couple hundred of the latter pages of War and Peace to exactly the issue you are wrestling with. Viewed from the collective level, the flow of human history seems as predetermined as, say, the weather. Viewed from the personal level, it appears to be the result of freely willed acts.
posted by ferdydurke at 8:19 PM on December 28, 2008


Ok but why does it all iron out to between 40,000 and 45,000 per year in the US for example? Why not 85,000? Why not 12,000?

Actually the fatality rates in different countries do indeed vary substantially--by something like the amounts you suggest here.

Among the countries listed in the link above, there is a difference (best/worst) of about 4X in fatality rate per capita and of 6X in fatality rate per billion kilometers driven.

So there is indeed a broad range of differences between countries. On the other hand, if you looked at the stats for one country over time, they would be quite consistent.

What that suggests is that "conditions" in each country (and conditions could be quite broadly defined--everything from driving customs and laws to automobile design to road design to topography and weather) determine the overall fatality rate.

If you change conditions you can change the rate.

Of course conditions don't change very quickly so the rate doesn't (usually) change very quickly in any particular place.

More to your point (and this is more of a practical answer to how to deal with the existential angst of "OMG I don't have any free will I'm just a cog in a big machine!!!11!!!" than a real well-thought-out philosophical answer):

On a practical level, random stuff happens. Some of it is bad, some good. Most just is. Call it fate, the working of the universe, or whatever. The point is, it's out of your control. You can accept that certain things will happen that are out of your control, or not, but either way they're still out of your control. (I suggest accepting it as opposed to the alternative. And some people even learn to revel in it . . . )

What you can change are conditions.

So if you (for instance) want to change your odds of being injured or killed in a car crash, buy a safer model car, take a defensive driving course, make a habit of paying attention when you're driving (as opposed to putting on your makeup while you text on your cell phone), lobby the DOT for safer roads, lobby law enforcement to put more of their energy into activities more likely to actually reduce accident rates (rather than simply line the city's coffers), etc., etc., etc.

All of that changes your odds of avoiding traffic collisions (and the odds of other people around you as well--a double payoff).

But of course--because of the problem of random stuff happening--none of that guarantees your safety.

You can do ALL of the above (and more). And still some drunk driver plows into you while you're sitting in the parking lot at Wendys and you're dead.

In short: Change what you can and don't worry about what you can't.

Whether there is free will or not, that is what you actually can do. So don't worry about it--just do it.
posted by flug at 8:26 PM on December 28, 2008


Of related note--

As others have pointed out above, there are quite a lot of things that are unpredictable at the individual level yet quite predictable at the collective level.

But the reverse is also true: a number of phenomena are quite predictable at the individual level yet completely unpredictable at the collective level.
posted by flug at 8:36 PM on December 28, 2008


Can someone explain why despite the fact that every one of these incidents occurs at a unique time and place, each involves a complex history of events and personal decisions leading to its very unlikely outcome, that the death toll on the roads year-on-year is so predictable?

Because, as others have mentioned, the range of possible events and decisions is relatively small.

We like to go around pretending that we have infinite choices, or that we even make use of any but a tiny fraction of the choices we do have, but that's not really how humans live. Humans are in fact extremely predictable creatures.

If you as an individual want to make sure that you're not one of the deaths on the road, you can make certain choices that will lower your chances tremendously (to 0% if you never go on the road) but in doing so you will wander far afield of the culture you live in.
posted by tkolar at 10:44 PM on December 28, 2008


"suppose statistically we could show that given the fairly consistent nature of world affairs, the apathy of the populace, the nature of the domestic political system US troops will be involved in war on average one year out of five. So what's the sense in me protesting and getting upset when it happens provided it fits with the existing probabilities?"

Because the people who protest are part of it - they're part of the balance of the system, and so they are part of why the US goes to war once every five years, instead of, say, once every three years.

It's the same as voting. Realistically, if you do nothing (or if you do something), it's not going to make any difference, unless a lot of others match your actions. But your actions, and theirs, are likely motivated by similar things. Hence the pentagon tries to control the information that Americans can obtain about the war (because the nature of the information available to you and others can sway you and others like you, potentially producing a tide of opposition), likewise politicians buy advertisements that can sway you and people like you.


Figure this out and you have your answer:
If you roll a dice once, the roll is unique and individual and if I try to predict the total, my margin of error will be about a hundred percent. I have very little chance of predicting what this individual dice will do.

But if you roll the dice a thousand times, even though each roll is STILL unique and individual, I can predict the total with such unnerving accuracy as to shock you - to within a fraction of a percent.
And I can do this every time. It's really quite simple.

What it boils down to, is that for something with various outcomes, the extent to which they are predictable depends on a ratio between options and sample size. A big enough sample size will make even mind-bogglingly unconnected outcomes settle down.

Part of the problem may be that the amount of person-minutes spent driving in the USA is astronomical - there is no way to fully grasp how vast it is, therefore you cannot comprehend that it is orders of magnitude more vast than the vast individuality of crash scenarios, and thus capable of evening out the wildcards. Cold math could do it, but math isn't your problem, it's intuition. And your intuition is broken here because the values are vaster than what the human mind can grapple with.

For the same reason, a lot of people intuitively reject evolution, because the mechanisms do not work with timescales and sample sizes that can be humanly comprehended.

Every now and then while driving, you might have a close call. Every now and then you might take your eyes off the road to adjust the radio. Given a thousand occurances of each, it's going to be relatively consistant how often they coincide and you're ill-prepared for the close call because your eyes are not on the road. Given a thousand occurances of people having a close call while not aware of it because their eyes are off the road, it's going to be fairly consistent whether a crash results. And so on.
posted by -harlequin- at 12:24 AM on December 29, 2008


I'd say that if freedom means that my unconscious mind is making choices I'm not consciously aware of, that's not a sort of freedom that interests me. (And it's a perversion of the word "freedom.")

Amen to that.

math isn't your problem, it's intuition. And your intuition is broken here because

all intuition is based wholly and solely on prior experience, and until you spend a lot of time playing with math, your math-related intuition is just going to be poor.

Intuition is of no use at all without training; in fact, the main point of training (as distinct from education) is to improve intuition within selected fields.

Take chess, for example. Put a grand master up against a chess novice and the grand master will not need to do conscious positional analysis or think three moves ahead; the grand master will be able to play completely intuitively and will still wipe the floor with the novice, even if the novice is in all other respects smarter than the master.

What is it about human nature in particular that makes the rate of error leading to death so predictably in the same narrow range?

Human nature is a complete red herring within the context of this question. The natures of individual drivers has almost no bearing on the predictability of the aggregate death rate, which rests squarely on the consistency of previously observed aggregate death rates and is quite unremarkable to people with training in statistics.

Teasing apart the various factors contributing to that death rate is in general going to be difficult. The most reliable way is just going to be implementing likely-seeming policies (like mandatory seat belts, drink-driving laws, road redesign, car safety standards) and observing what actually happens.

Attempting to predict the aggregate road death rate by doing thought experiments involving the free or otherwise will of individual drivers will get you precisely nowhere. You may have heard the saying that "the plural of anecdote is not data": same thing. Best you can do is stuff like "if we cut back the overhanging trees on the approaches to this intersection, that will make it easier to see traffic approaching on the side road, and that ought to make crashes less likely here next year".
posted by flabdablet at 5:00 AM on December 29, 2008


I'd say that if freedom means that my unconscious mind is making choices I'm not consciously aware of, that's not a sort of freedom that interests me. (And it's a perversion of the word "freedom.")

That's an interesting comment, because one thing it is clear we CAN do, is train our unconscious mind (and our habits, and our reflexes) to react in certain ways.

Be open to the idea that "free will" might be most powerfully exercised in ways far different than what you (or anyone) might think at first blush.
posted by flug at 4:39 PM on December 29, 2008


And of course the unconscious mind can be far more powerful than we expect.
posted by bonaldi at 11:22 AM on December 30, 2008


« Older Just watched a very disturbing...   |  I recently read a novel which ... Newer »
This thread is closed to new comments.