Now, I’m quite sympathetic to the poorest of the poor being taken care of, but I had no idea until this week that “being taken care of” in this particular case means even better treatment than the treatment those who are paying to support Medicaid receive.

Nothing like a good dose of reality to reaffirm my disdain for the State!

Considering evidence is an important role in rationality, Bayesian analysis, and in the Less Wrong sequences.  This EWOT post brought to mind this older post by Eliezer (which, low and behold, was already open in my browser when I went to go look for it).

The type of anarcho-capitalism schilling I see on EWOT and other blogs reminds me of the Salem witch hunt imagery that Eliezer evokes:

no matter what the accused witch said or did, it was held a proof against her.

Just in case the parallel isn’t obvious, in the forementioned blog, high quality of care is used to support the anti-gov’t position on public health care (in this case, because the high quality- and therefore high cost – of care isn’t fair to those truly footing the bill). However, I’ve seen the opposite result – low quality of care – also being used to support the anti-gov position.

This practice surely isn’t restricted to this issue (or anarcho-capitalists!). The job of a pundit is to use their narrative to explain the outcome of any event and so, you get multiple hypotheses explaining the same event and no hypothesis seems to have the power to exclude any another one.

But the main problem is that each observer has their one-size-fits-all glove and they’re looking for hands to fit it to…and any hand will do. Any evidence they encounter confirms the theory because the theory can be retrofit onto the evidence.

This is why careful reasoning is so important.

If an observer sees high quality of care for medicare patients as evidence that you should have disdain for the state, shouldn’t low quality of care be evidence that one shouldn’t have disdain?

The statement can be broken down as such:

prior: high disdain for the state

hypothesis: if public health insurance provides better care for those who are not paying for it than for those who are, then one should have disdain for the state

evidence: high quality care for medicare patient

outcome: high disdain for the state

This is a perfectly legitimate libertarian theory and how the original evidence calculation should have been formulated. The outcome matched the prior and therefore did not shift it much.

However, the hypothesis was, in actuality, retrofit to the evidence.

In a true probability calculation, if the outcome had been different (say, for example, medicare refused to cover the ultrasound imaging), the hypothesis would have been smashed. It would have been evidence that people receiving public health care do not get higher quality care than those who are paying for it, and we could not support our prior (disdain for the state). Clearly, if public health care sucks, the state is doing something right (keeping costs low for taxpayers, perhaps).

I’m going to generalize here and say that this is not what most libertarians (or any other political-identifying group) would have done. They would have simply found some other hypothesis that supports their prior (if the government isn’t providing high quality care, then disdain for the state, for example).

This a particular problem when you don’t allow yourself to expect counterevidence. In a practical sense, if no evidence you encounter has the ability to shift your beliefs, then you’re not coming up with testable hypothesis.

To make things worse, complete stories, ones with more details tend to be more convincing. We thus not only gravitate to stories, but stories with embellishments. Are there better (and more likely to be correct) ways of satisfying our curiosity?

Andrew Gelman gives us a bit of a hint:

Many years ago, Don Rubin convinced me that it’s a lot easier to think about “effects of causes” than “causes of effects.” For example, why did my cat die? Because she ran into the street, because a car was going too fast, because the driver wasn’t paying attention, because a bird distracted the cat, because the rain stopped so the cat went outside, etc. When you look at it this way, the question of “why” is pretty meaningless.

Similarly, if you ask a question such as, What caused World War 1, the best sort of answers can take the form of potential-outcomes analyses. I don’t think it makes sense to expect any sort of true causal answer here.

Last week I wrote that people think that their decisions influence the outcome of those decisions. This is especially true with buying insurance. Buying insurance, for most people, on even risky things is a tacit admission of fallibility. Auto insurance is mandated, health insurance is used often and life insurance is about the family. But we simply do not insure against some of life’s biggest risk (and, since I’m reading Taleb, we’re not even talking about Black Swans here, which you can’t really insure against anyway by definition).

But I wonder what other incentive signals might be at play. Is there really no market for microinsurance, is government signals creating too many moral hazards to make people view this type of insurance as a necessary cost?

Prospect theory tells us that people are very risk averse and overweight the likelihood of extreme events occurring. Why then do we not insure against those events?

Perhaps I am correct in postulating that we think that deciding not insuring ourselves against these risks is effectively protecting us against them (or at the very least, what we ignore cannot bother us). It’s a frightening thought, and explains catastrophes like Katrina or 9/11 are apparently made worse by poor disaster management strategies.

Zach muses:

At an Overcoming Bias meetup group I went to a couple weeks ago, several people claimed that learning how to think rationally has helped them get ahead in life. My instinct is that this is probably true, but it doesn’t seem to me that rationality enthusiasts are statistically better off than irrational folk.

First: we all make rational and irrational decisions on a daily basis. That is, although we often so not always behave ideally to maximize expected utility, a lot of what we do is good enough. Second, I would call rationality enthusiasts only “aspiring rationalists”—there’s no guarantee that they’ve actually achieved what they claim to value, though I like to give people the benefit of the doubt.

I can see how Zach might have seen the people at the meetup as overemphasizing the benefits of studying rationality. But they wouldn’t be too different from fitness enthusiasts who go on about how running, swimming, or weightlifting have improved their lives.

Overstated benefits? Likely.

But there are areas where understanding and counteracting failures of rationality don’t experience diminishing returns as sharply. On a larger scale almost anything with high payoff and sufficient complexity qualifies. War, politics, global catastrophic risks are all examples.

photo by mescon on Flickr

On an individual basis, I have an easy example: clinical depression.

Cognitive therapy, which has since been incorporated into more comprehensive Cognitive Behavioral Therapy is centered around the idea that depression always manifests as dysfunctional thought. The approach is to stop depressive thought. This is regardless of the etiology: effectively counteract the manifestation of dysfunctional thought, and you are effectively treating the symptoms of depression. Hinder the occurrence of the dysfunctional thoughts, and you limit the onset of depression.

A key part to this is the idea that depressive thoughts are the result of errors of thinking. Sound familiar? For a sad or frustrating occurrence to form become a depression instead of temporary grief or anger, depression-prone people must succumb to a false piece of reasoning about whatever is bothering them.

David Burns, M.D., psychiatrist and important figure in the development of Cognitive Therapy

Many cognitive distortions are also logical fallacies; related links are suggested in parentheses.

1. All-or-nothing thinking – Thinking of things in absolute terms, like “always”, “every” or “never”. Few aspects of human behavior are so absolute. (See false dilemma.)
2. Overgeneralization – Taking isolated cases and using them to make wide generalizations. (See hasty generalization.)
3. Mental filter – Focusing exclusively on certain, usually negative or upsetting, aspects of something while ignoring the rest. For example, focusing on a tiny imperfection in a piece of otherwise useful clothing. (See misleading vividness.)
4. Disqualifying the positive – Continually “shooting down” positive experiences for arbitrary, ad hoc reasons. (See special pleading.)
5. Jumping to conclusions – Assuming something negative where there is no evidence to support it. Two specific subtypes are also identified:
   • Mind reading – Assuming the intentions of others.
   • Fortune telling – Predicting how things will turn before they happen. (See slippery slope.)
6. Magnification and minimization – Inappropriately understating or exaggerating the way people or situations truly are. Often the positive characteristics of other people are exaggerated and negative characteristics are understated. There is one subtype of magnification:
   • Catastrophizing – Focusing on the worst possible outcome, however unlikely, or thinking that a situation is unbearable or impossible when it is really just uncomfortable.
7. Emotional reasoning – Making decisions and arguments based on how you feel rather than objective reality. (See appeal to consequences.)
8. Making should statements – Concentrating on what you think “should” or ought to be rather than the actual situation you are faced with, or having rigid rules which you think should always apply no matter what the circumstances are. Albert Ellis termed this “Musturbation”. (See wishful thinking.)
9. Labeling and mislabeling – Explaining behaviors or events, merely by naming them; related to overgeneralization. Rather than describing the specific behavior, you assign a label to someone or yourself that puts them in absolute and unalterable terms. Mislabeling involves describing an event with language that is highly colored and emotionally loaded.

Personalization (or attribution) – Assuming you or others directly caused things when that may not have been the case. (See illusion of control.) When applied to others, blame is an example.

(from Wikipedia)

Another form of therapy based on careful examination of cognitive processes is (incidentally named) Rational emotive behavior therapy.

As a formerly severely depressed person, learning successful strategies for dealing with my own thoughts has been the primary vehicle to recovery.

But, in the end, I agree with Zach’s general perspective: learning about rationality (and its failures) does not greatly or immediately make us capable of making better decisions. It’s general knowledge, after all. Applying rationality to the complex situations we face still remains the more important undertaking.

Of course, I’m personally interested in understanding how I can use rational thinking to get ahead, in personal and professional life, but I haven’t seemed to be able to inductively reason how I might go about this. Any ideas?

New York City, MTA Bus

The first major issue with this poster is the choice of the word “risk.”  Slips, trips and falls are potentially injuring but is the action itself risky? Walking or running out of a bus is risky because there exists the potential to trip, but once you’ve already started to trip no new risks are being taken because, presumably, you don’t choose to follow through with the trip – it just happens. The poster should have said “running is risky because you may slip, trip or fall” or maybe “slip trip or fall is dangerous.” But slipping can’t be said to be risk-taking, at least not how I understand the word.

I think this picture is self explanatory about why reference frames are important. The other glaring error is the quantitative element: that a slip, trip or fall is twice as risky. Twice as risky as what, exactly? If you’ve already tripping, the risk that you will continue to fall is fairly high – probably more than 2. So I don’t think that’s what they mean.

Most probably, slip trips and falls cause bus-related injuries two times more frequently than some other incident – though we can only imagine what that might be.

Since they also qualify the statement with sometimes – when exactly should I be worried about slipping and tripping? What are the risks when these conditions don’t apply?

What bothers me so much about this poster is that they don’t provide reference, nor do they explain under what conditions are slip, trips and falls more likely (only that they sometimes are). Reading that poster, should I be nervous about suddenly and randomly slipping out of my seat – because, sometimes, that’s twice as likely (twice as likely as gravity reversing and I get stuck to the ceiling, maybe?)?

This poster basically tells me nothing because they don’t frame it around a stable reference or provide relation – I don’t know how likely I am to trip at any given moment. Realistically, I realize that the poster is only trying to make passengers aware of a common problem in bus safety, but it would have taken about an two extra seconds to realize the statement is nonsense and to fix it.

Now, every time I see it on the bus (thankfully not every day) I am forced stare at the poster in frustration.

I don’t think its a stretch to assume that the man had no meteorological basis for making this claim, yet we hear these sorts of statements all the time. Why do people commonly assume that there are physical relationships between past and future events? Or, even worse, that past events can somehow influence the outcome of future events even when lacking a direct causal element?

I’m currently working my way through some of David Hume’s philosophy. He seems to focus on one important notion; that causal relationships, even down to apparent the physical laws of the universe, are influenced by our perceptions. In other words, we assume, despite lack of evidence, that a falling object will conform to Newtonian laws of motion everywhere in the universe. We make inferences about effects without completely understanding the causes because all previous experience has given rise to these expectations – and so we want all future experiences to conform as well. Furthermore, because these experiences must be filtered through human consciousness, events (and our expectation about future events, causes and effects) are inherently biased.

This notion can be extended one step further, by discussing the statement from the guy on the radio.

Not only do people seem to have askeptical expectations about causal relationships (by assuming a deterministic worldview and ignoring the effect of subtle variations on outcomes) and too brazenly draw inferences about the world, but there is also the expectation that the human brain can bend outcomes at will.

Radio guy wills there to be a mild winter and has some expectation that this will occur – despite the lack of data that links weather patterns in this way. Sports fans often wear their lucky jersey or sit in a specific way, in order to meta-influence the game.

Hume was only half way there.

Not only do our minds distort our expectations for cause to preceed a predicted effect, often these expectations are based nothing more than hope, desire and observation biases. Once these biases are set up, we think that our [incorrect] expectations can have an affect on actual outcomes.

Harvard Law Professor Cass Sunstein leaning suggestively on a doorframe

I spent most of the afternoon yesterday obsessing about Cass Sunstein, whose views on just about everything from free speech to animal rights confuse the heck out of me on first glance.

In general, that means it’s time to learn more.

Here’s what happened: a friend of mine on Facebook, yelling about other things, was gracious enough to share this The New York Post :

Advance copies of Sunstein’s new book, “On Rumors: How Falsehoods Spread, Why We Believe Them, What Can Be Done,” have gone out to reviewers ahead of its September publication date, but considering the prominence with which Sunstein is about to be endowed, his worrying views are fair game now. Sunstein is President Obama’s choice to head the White House Office of Information and Regulatory Affairs. It’s the bland titles that should scare you the most.

[...]

Sunstein was appointed, no doubt, off the success of “Nudge,” his previous book, which suggests that government ought to gently force people to be better human beings.

Fascinating! I haven’t read Nudge, either (this is what happens when one’s book budget goes down the drain) but from what I know, Sunstein seems to enspouse a modified form of libertarian paternalism.

The Post column has actually made me more interested in Sunstein’s work. Considering the press Pop SocSci authors like Malcolm Gladwell get, it seems to be not very well understood outside of law and behavioral economics.

I guess I should stop ignoring the guy.

Most brazen part of this: I have a copy of Cass Sunstein‘s 2007 book Worst Case Scenarios sitting right in front of me.

The beer is gone, but the book remains unfinished.

The cafeteria in the hospital where I work, has a half a dozen cashiers. The people queue up with their food and are directed to the next empty cashier by a dedicated line management director (who, for some reason, wears a doctor’s white coat as his uniform). This caused me to think about why the hospital – the catering services department in this case – felt that this type of signaling was necessary.

If there was ever a case where the traditional economic assumption of rationality is probably correct, it’s here. Consumers, who quite obviously don’t like to wait on long lines, have a very clear goal: to maximize their utility function of food consumption. They want to reduce the amount of time on line and maximize the time they can relax on their lunch break, socialize or get back to work if they need to. Information here is quite clear. The cashiers are set up in such a way that scanning the available cashiers will reveal the next available one. There are ropes clearly marking where individuals should stand, single file, to wait for a cashier. So why are they wasting resources on hiring a line manager, who’s sole apparent purpose is to tell the next person in line which cashier to go to?

Assuming that my employers are themselves rational and have compared the cost/benefit analysis of smooth operation with and without a line manager, I must assume that without the ‘central planning’ of the line, there would be enough disorder that it hurts diners’ experiences and therefore profits.

What this brings to mind is something fellow FS blogger, Robert Simione, said to me the other day about the economic assumption of rationality:

…[E]conomists label “rational agent” as someone who wishes to maximize their monetary gains (or whatever you wish to call their pool of money).  I was thinking about where this assumption comes from, and… about how all through history, philosophers since Aristotle stated that everyone’s goal is to maximize their happiness. Then economists took this idea and consider the monetary value of various happiness costs, and they say: “People trying to maximize happiness is like people trying to maximize the use of their money.” So now I wonder, what if this idea, that “rational” people always try to maximize their general happiness, what if this idea isn’t true? …I think the Efficient Intentions Hypothesis is not always true, and I think that you are right, that we may not be particularly good at finding a path to optimal happiness.

This is an astute analysis that I want to take even further. Even if people are rational, are good at – or don’t even bother trying – maximizing their happiness (which doesn’t appear to be the case), individualist goals under self motivated behavior to reach some new happiness potential does not necessarily produce the most efficient outcomes in aggregate group behaviors.

Its like a story I picked up from Tyler Cowen from Marginal Revolution (via Jason Kottke and other sources). Individuals may choose the shortest route to their destination when driving, but this doesn’t necessarily produce the best outcomes for anybody. Everyone wants to take the shortest routes, but the resulting traffic slows everyone down (the solution, by the way, is to intelligently close certain roads). My cousin-in-law, who is a traffic engineer for the NYC-NJ Port authority, pointed out to me that this is why the NYC’s plan to cut off traffic in parts of Broadway in Times Square and Herald Square was such a good idea. Instead of being stuck in traffic, drivers must take multiple alternative routes through midtown Manhattan, which effectively relieves congestion (and leaves more room for pedestrians).

This is an analogous, though not exactly parallel, story to my line management problem. We have the problem of self interested behavior creating chaos, rather than the smooth order predicted by the invisible hand. But, unlike driving, lunch-goers can see all of their options and select the fastest route accordingly. In this case, picking the cashier with the shortest line prevents congestion at the other registers. It’s not as if consumers are picking out one register just because its the closest or has the most attractive cashier (though this could still be a problem). If people are acting rationally and have no problem acting upon their preferences, I would predict that no line manager should be necessary.

It remains a problem to know exactly how rational people are and if they have enough information to pick a route that maximizes their utility to their expectations. Automobile operators, particularly ones that are well informed about pure route distances but not traffic patterns, may not be entirely rational or have enough information to produce beneficial outcomes. I don’t believe this is true for a lunch line.

In economics, these problem exists, as well as the ever important fact that the central planner has even less information than the markets. So while the aggregate result of self-interested, irrational behavior may be chaos, the outcome may still be preferable than a planned economy under technocratic rule.

to just accept the widely accepted fact that air consists of collection of different gases you need “faith” in that science can really resolve and fully explain this kind of things

I see what he’s getting at and agree to some extent.  The general public accepts 99.9% of the basic scientific principles at face value (just intuitively) without questioning them.  But is this faith akin to religious faith?

We accept scientific principles not because we have some deep religious trust in our scientist (maybe some of the big guns have become an exception) but because the public knows, on some level, the type of rigid attitude scientific method implies.

Its a shame, but we, for the most part, no longer teach children about classic science experiments in school or even much at the university level.  I believe this lack is partly to blame for the confusion between trust and faith when it comes to science.

You cannot easily verify this fact by yourself and so you must in the end trust people that have taken to their mission to find these things out.

We can and do trust scientists to do their jobs, but this trust is not akin to faith. To have faith in the bible means you can’t check the references or repeat the experiment.  Even our most basic scientific principles, however, have been published, largely in a public forum.  Want to read about Newton’s or Maxwell’s early physics experiments?  Go to the library and you’ll see all their data and how they intepreted it.  The same goes for any branch of natural or social sciences.

In fact, here is a monthly blog carnival, The Giant’s Shoulders, dedicated to reconnecting with our scientific past.  Look through some of the posts and see how some of our most basic scientific principles used to be falsifiable, hot button issues, needed to be proven with rigor.

jaskaw, in his original post, makes this general point as well, but seems to imply that this type of trust is rational (in contrast to religious faith).  That I’m not so sure about.

Is it rational to trust scientists – who we know have political or personal agendas, who have been known to lie, publish fradulent data, use their connections to stretch the limits of high standard and peer-reviewed publications for personal gain?  We have a real scientific community, and then there’s the one the media presents us with – and it’s not always friendly – that stretches already tenuous results.  The media promises us miracle cures and cancer drugs every other week, but real progress is much slower.

Maybe people are irrational for trusting scientists so much.  99.9% of us will reach for the pill bottle without even a literature review, trusting the work of corporations and government scientists to do the proper vetting.  This is despite the fact that the media plays up reports of ineffective or dangerous drugs, because bad news sells.

Despite all of this, we approach science with a trusting, almost religious-like, blind eye.  We trust in the scientific process and the painfully slow scientific method somehow gives us great results in the end.  But, if we were truly rational about it, maybe we would reject science altogether?  And I say this as a scientist with a first hand account of what the scientific process is all about.