Josh Barro, writing for businessinsider.com, recently made the claim that private health insurance is "a government benefit provided through public channels". At the risk of opening myself to political argument, I'd like to address his major point, which is that health insurance is not a typical insurance product,  "designed to turn an individual's risk of loss into a predictable cost". I claim that it he is correct, but by changing the word "individual" to "group", changing the dimensionality of the insurance's target, he would be incorrect. 

Barro argues that sick people's benefits from health insurance exceed their monetized risk, and healthy people pay more than they actually expect to gain. This is a patently wrong conclusion: "broken" homes have "benefits" that exceed their homeowners monetized risk (homeowner's insurance policy), and "well-maintained" homes are on the opposite side of the spectrum. There just isn't a false equivalency here, as Barro explains it. 

So it is a bit striking that Barro is actually able to make his point well, because while, as he first explains it, health insurance is equivalent to homeowner's insurance, he accurately later points out that regulated health insurance:

Buzzwords aside, simple insurance (Barro's homeowner's insurance) is fundamentally a functional agreement which takes present risk and monetizes it over a long period of time: in other words it spreads risk across one dimension. Since my home's risk of, say, termite damage is (largely) has a relatively small spatial covariance with other homes, my contract with an insurer is one-dimensional. I give the insurer money, and they spread my risk over time for a small fee.

In the health sphere, however, my personal risk of illness is considerably tied in to that around me. If you are not convinced, consider the rapid outbreak of swine flue, or peruse the wikipedia entry on epidemic, the common code, the flu, etc, etc... Not only do I encounter risk in terms of random fluctuations in my health over time, but also in the random interactions between myself and others. Health insurance requires a spreading of risk over two-dimensions, across time and between people. 

It is for this reason why, for health insurance to mean anything, it must be imposed on a supra-individual level, such as at the national level.  Because any one person's engagement in a health insurance scheme is an implicit contracted two-dimensional spreading of risk, they are effectively losing their insurance when other individuals fail to enroll. 

Capital-h Health, not health insurance, is what is not a toaster, and so to insure it is to contract in a two-dimensional relationship between individuals and with an insurer. Barro is (almost) right. 

Ever wonder what the most-heard phrase in the world is? A good bet would be:

Which repeats every 15 seconds at at least two locations at (approximately) all of the 50,000 airports in the world (assuming there's a few more moving walks at, for example, LHR than in a small regional Tibetan airport). That gives us a quarter of a billion broadcasts per day, meaning it could probably compete with the Muslim call to prayer, "Mind the gap", or that stupid R.Fancourt roofing song for the most heard or repeated non-spoken phrase. 

Here's a similar one:

Which got me thinking about what exactly this statement meant. One hears it over and over again, to the point where I'd bet most people can form some approximation of it when asked.

Here's the thing, though: there's a difference between the prohibition of "unattended baggage" and "baggage left unattended". Prohibition necessarily requires an actor to be prohibited from performing an action, in this case leaving your bags alone in an airport. The TSA's statement, however, is banning the unattended baggage itself, without reference to who is to be punished!

This is just a good example of the implicit way language can be used to signify meaning. On its grammatical and lexical construction, the TSA's statement is a bit funky because of the misuse of "prohibition".  And this may be by construction, because unattended baggage, by its very nature, has no person to blame for its existence. All interesting things to think about. Just how intelligent is the TSA?

Still, this is the danger of speaking about explicit things (scientific results, for example) using a language which places a considerable weight on implicit meaning. Words mean things that are not immediately self-evident, but data and information is independent of context, which is one of a myriad ways of viewing the reflexive disdain scientists and journalists have for one another. 

Here is a link to a recent paper discussing the negative correlation between African literacy rates in the colonial and post-colonial times and the "slave export intensity" during the pre-colonial era. 

Cherokee Gothic rightly points out that this is an example of economic path-dependency, a concept familiar to mathematicians: for certain quantities, it isn't where you end up, but how you got there. 

Example: if one pegs their net worth at the value of the stock AAPL, and through some sorcery predicts every upturn and downturn in the stock price, liquidating at peaks and converting all cash to stock at local minima, in a year that person would have a considerable amount more money than the person who held the stock fixed, and much more than the person who made the opposite choices. The path taken to the end is what caused the discrepancy in wealth after a year. 

Path dependancy is a familiar trope in political theater and is, depending on the political and philosophical bent of the person, the reason for gender and race gaps in education, poverty, and incarceration.  It can be summarized (thanks to Scott E. Page) in the "old Bostonian jump roping rhyme"

And so this recent paper attempts to gauge the level to which slave export has biased literacy rates in African countries over the proceeding centuries. The answer? According to the abstract: "a negative and signicant relationship between slave
export intensity before the colonial era and literacy rates during the colonial era."

Here's the data used to support that claim, buried in a plot in the supplementary material. This image plots literacy rate against some normalized quantity representing pre-colonial slave exports as a percentage of the extant population. Where's the trend? 

To me (and this is just me), this appears to be a classic case of oversimplification. If the author wrote this paper with the exact opposite conclusion, I would be equally swayed. What causes the four outlier groups in slave export to be there? Why is there seemingly no trend? Why did the author connect two clusters with a line and call it a trend?

Bad science, even in Path Dependancy

If you've been under a rock the last week, former New England Patriots tight end Aaron Hernandez has been indicted for first degree murder. Anyone owning an Aaron Hernandez jersey is allowed to exchange it for any other actual Patriots player's jersey for free, the reasoning being that having NFL fans wearing the jersey of a suspected murderer is not good publicity for the league or the individual team. 

Lemon Laws were initiated in the United States to counteract information asymmetry between the buyer of a product and the seller. Buyers are at an immediate information disadvantage to sellers, since the seller has much more of an understanding of the flaws of the product he is selling. When the product turns out to be worth less than the price the buyer paid based on prior knowledge available to the seller the buyer is defrauded, and depending on the product, has some legal ability to recoup losses.

Now the non-ironic value of a Hernandez jersey is zero, and the average Patriots fan had absolutely know idea how bad of a guy he was, though it appeared to be common knowledge to insiders. His jerseys are lemons. 

The Patriots are enacting an even stronger warranty (maybe a double-lemon law?), since even they didn't know Hernandez was as bad a guy as he turned out to be. So Hernandez is partially to blame for the lemoning of his own jerseys. This is sort of like a car company knowing its cars' third-party manufactured brakes wouldn't last the life of the car, only to find out that, unbeknownst to them, the brakes didn't work at all!

Who should really be responsible for covering the "loss of value of Hernandez jerseys?" (assuming of course people who bought them actually care). Perhaps both the Patriots and Hernandez are equally at fault. I'm not sure how exactly to answer that question. 

At least here it is refreshing to see information asymmetry work in both directions. 

The most interesting short paper I have read in the past few years is the pre-press article by mathematician Kyle Swanson: "Emerging selection bias in large-scale climate change simulations" (behind the Wiley paywall). In it, Swanson shows that the ensemble of models most popularly used for scientists exhibit a selection bias towards accurately capturing some phenomena (say the Arctic sea ice extent). In turn, this has resulted in (for other important phenomena) an intra-model spread which has decreased through the multi-year model refinement process, but whose mean has shifted further away from reality. 

In other words, there is a selection pressure on climate models which seems to be guiding them to converge upon the same result... in the words of evolutionary biology, researchers belief that they are "getting something right" is what is paying for these evolutionary changes. Yet according to Swanson (and a privately held belief of many other researchers) this satisfaction is misplaced. In the paper, Swanson retells Feynman's story of what happened after Robert Milikan miscalculated the charge on the electron. 

(Just to establish a baseline for this criticism, this selection bias is related to model-predicted or model-diagnosed patterns, and cannot be used to criticize climate change as a scientific fact. That the world is warming has been observed, evidenced analytically, and demonstrated in models over a wide range of complexities. Enough about that.)

The picture is a figure from his paper, showing how the model spread of the frequency of anomalously warm and cold months changes from the CMIP3 (~2007) and CMIP5 (~2010) intercomparison projects. This is plotted against "observations" from the reanalysis products HadCRUT and ER. The trend is obvious: the models tend to get cluster together but their mean does not shift towards the actual data. 

For a long time, the rub on climate models was that they had very low precision, model spreads were large and uncertainties were high. The unmentioned benefit for this was in accuracy: real data often fell within error bars of prediction. Now we have increased model precision at the expense of accuracy, and in the hierarchy of model outcomes, accuracy should be placed above precision. Better to be reasonably sure than to be confidently wrong. 

This is troubling: the implication being that modellers are providing a selection pressure for results which is (in a second sense) "unnatural", and climate models are becoming increasingly covariant (perhaps in response to "improvements" in physical parameterizations that are added to the entire ensemble), but are not becoming more skillfull. Not unlike the cartoon rabbit who puts his finger in the dike, only to see a new leak spring forth somewhere else, the rush to improve climate models in certain areas has resulted in even larger problem.