"Most Admired"—Interpreting Statistics

I recently came across a post online by someone who argued that the stories about Obama's loss of popularity must be bogus, since a poll had shown him to be the most admired man (and Hillary Clinton the most admired woman) in the country. I pointed out in a comment that "most admired" might mean admired by ten percent of the population if nobody else got above nine.

It turns out that I was correct, although the number is a little higher than my example.

The open-ended poll released Monday found 16 percent of people named Obama as the most-admired man, while 15 percent named Clinton as the most-admired woman. Both scores, however, have dropped significantly since 2012.  

In 2012, 30 percent of people named Obama as the most-admired man, while 21 percent named Clinton the most-admired woman. 

(From The Hill)

One implication is that the post's argument was strikingly wrong, since Obama's popularity, by this measure, is only half what it was a year ago. The more interesting point is the risk of arguing from factoids without thinking carefully about what they actually mean.

A Chinese Double Standard

No one thinks of enshrining Hitler nor labels Jewish anger an overreaction when Auschwitz is brought up.

(From a Chinese media source criticizing a visit by the Japanese prime minister to a shrine honoring Japanese war dead)

China does, however, continue to honor Mao, who killed more Chinese than any Japanese war criminal, probably more than all of them put together. Also more people than Hitler killed.

For details, see Rummel's figures on Mao's democide.

More on RSA/NSA

In a recent post, I mentioned the report that RSA accepted a ten million dollar payment from the National Security Agency in exchange for deliberately making a vulnerable pseudorandom number generator the default choice for their encryption software. A recent Wired article by Matt Blaze goes into detail on the technology and its implications. It's worth reading the whole piece—the following is a sketch.

DUAL_EC_DRBG is an algorithm for generating pseudorandom numbers—pseudo because the process is deterministic, so if you know all of the inputs to the  generator you can predict the output. Blaze writes:

One of its parameters, called “Q” in the standard, turns out to have the property that if it is chosen in a certain way, whoever selected it can have a secret backdoor that allows them to reverse the algorithm and discover the seed. (This property of Q appears to have first been noted by Daniel Brown in 2006.) And a fixed value of Q is specified in the standard, with no explanation of how it was selected. That this could provide the NSA with an effective backdoor to predict DUAL_EC_DRBG’s output was observed in a talk at the 2007 CRYPTO conference by Dan Shumow and Niels Ferguson of Microsoft.

In other words, the value of Q could have been chosen in a way that, along with additional information, let those who chose it deduce what number came out of the random number generator, hence predict the key the software using it would generate, hence decrypt messages encrypted with that key. The process only works in one direction—knowing the value of Q doesn't let you deduce the information needed to use it as a back door to decrypt. Knowing how Q was generated does.

Which means, assuming the obvious conjectures are correct, that what the NSA was embedding in RSA software was a master key. Using it the NSA could decrypt information encrypted using numbers generated by DUAL_EC_DRBG. Other people could use that master key only if they were able to get from NSA the information on how the value of Q used in the standard had been generated. 

A very clever idea. Assuming Blaze is correct, quite a lot of the cryptographic infrastructure generated during the nine years when DUAL_EC_DRBG was the default algorithm in RSA encryption software is insecure against the NSA. Also  against anyone else who somehow obtains the information on how Q was generated.

I should  add that RSA has denied the charges but offered no explanation of why they made that particular PNG the default in their software and kept it the default long after security professionals had pointed out its weakness. Nor has RSA denied or explained the purported ten million dollar payment from NSA. Their denial amounts to "trust us, we didn't do it."

Sharia: Constitutional (1). Fiqh: Constitutional (2)

Public discussions of traditional Islamic law, by both supporters and critics, routinely refer to the law as Shari'a. As best I can tell, that is a mistake. Islamic law as it was interpreted by legal scholars and applied in the traditional court system is fiqh. Shari'a is what fiqh would be if the scholars always got it right, something they have never claimed to do. Think of it as law in the mind of God.

The same distinction exists in American constitutional law, even though we confusingly use the same word for both concepts. If a lower level court rules a law unconstitutional, it is making a statement not about the Constitution but about the views of the Supreme Court. Roughly speaking, "this law is unconstitutional" is the prediction that, if the case gets to the Supreme Court, the law will be struck down. 

But consider the same question from the standpoint of a Supreme Court justice. His vote is not a prediction of how he and his colleagues will vote, it is a statement about how they should vote. If he ends up in the minority, the implication is not that he is wrong, only that his view failed to prevail. While his vote may, under the principle of stare decisis, be influenced by past votes of past justices, what determines it is his view of what the Constitution says and how it ought to be interpreted.

A constitutional scholar is likely to use "constitutional" in both senses. In some contexts, the question he is interested in is how the Supreme Court can be expected to rule. In others, it is how it ought to rule.

Think of "constitutional" in the Justice's sense as "constitutional (1)," in the lower court's sense as "constitutional (2)." Shari'a in traditional Islamic law corresponds to constitutional (1) in our system. Fiqh corresponds to constitutional (2).

Evaluating Controversial Claims

Someone makes a scientific claim, typically as an argument for some policy. Examples would be current claims with regard to global warming, claims fifty years ago about the consequences of population growth, claims early in the Obama administration about the need for large deficits to bring down unemployment. There are at least four different ways in which an interested observer can decide whether or not to believe the claim.

1. Partisanship. If you support the policy, believe the claim. If you don't, don't. This is probably the most common approach.

2. Evaluate the arguments for yourself. This is  the most entertaining and educational approach but no more reliable than the first—and likely to give the same answer. There is always controversy about the claim among people better equipped to evaluate it than the random observer, although one side or both may try to deny it. In the case of global warming, the relevant claim is not merely that temperatures are going up, or that the reason is human activity, or that they can be expected to go up by enough to cause serious net costs, but all of those plus the additional claim that there are ways of reducing the increase that are worth their cost. To evaluate all of that you need a reasonably expert knowledge of climatology, statistics, ecology, economics, and probably two or three other fields I have not thought of. Since you don't have all that, you end up believing whichever arguments you want to believe.

3. The argument from authority. You try to figure out what the consensus of the people who are experts is or what some authoritative source of information says. An outsider trying to figure out what professionals in a field believe is at risk of overvaluing whatever position has the most support from public sources of information, such as the mass media, or has done the best job of getting its supporters onto the committees of scientific organizations that put out public statements. And even if he could figure it out for one field, that isn't sufficient. Again taking the global warming case, it is not enough to know what the consensus of the climatologists is, even if you can separate the facts from the puffery on that subject. Climatologists are not economists, so  could be correct about the expected temperature increase and wrong about the magnitude or even the sign of its consequences. Economists are not ecologists, so might show the costs they are looking at to be insignificant while missing the effects of climate change on other species. I discussed problems with this approach at greater length in an earlier post.

4. Prediction. Once such a controversy has been going for a while, partisans have a track record. If they have made confident predictions that turned out to be wrong, that is good evidence that they are either dishonest or arguing from an incorrect theory. Figuring out whether the arguments for a theory are right or wrong is much harder than finding out what that theory predicted. Sometimes all the latter takes is  a book or article by its supporters written a few years back.

The clearest case is the population hysteria of the 1960's. Paul Ehrlich's Population Bomb, published in 1968, confidently predicted mass famine in the third world over the next decade, with hundreds of millions of people starving to death. Not only did it not happen, the real world moved in the opposite direction, with calorie consumption per capita in the third world going up, not down. That is very strong evidence that Ehrlich can not be trusted. It is somewhat weaker evidence that the movement of which he was part, whose members generally took him and his arguments seriously, can not be trusted.

Weaker examples apply to my other two cases. Early in Obama's first term, the Administration offered predictions of what the unemployment rate would be without a stimulus and how much lower it would be with the stimulus that the Administration wanted and got. Actual unemployment rates for the next several years were higher with the stimulus than predicted without it. That does not tell us whether the stimulus was a good policy. It is possible, as its supporters argued after the fact, that things were simply worse than they thought. But it is good evidence that predictions made by Administration economists can not be trusted, that either they were deliberately fudging the results or were using models much less reliable than they claimed.

For the case of global warming, we have the IPCC's repeated overpredictions of global temperatures, hurricane rates that are strikingly lower, not strikingly higher, this year than the average, and a number of other predictions to which the real world has failed to conform. Again, that does not show that the underlying argument is wrong. It does show that the people and models that have been generating the mistaken predictions cannot be trusted.

Which is about the most that the interested outside observer can hope to learn.

---

P.S. Not surprisingly, a lot of the comments on this post focus on the particular case of global warming rather than the general argument. One commenter provides a link to an article by Chris Landsea on the effect of global warming on the frequency and strength of hurricanes. One conclusion is that global warming by the end of the century might result in a slight increase in the strength of hurricanes and might also result in a substantial decrease in their frequency. 

Another is that the historical record provides no support for the claim that hurricanes have been becoming more frequent or stronger over the past century or so. The reasons that hurricane damage has trended up is that the total value of property in coastal counties has increased. The reason that the number of recorded hurricanes has trended up is probably the large improvement in our ability to detect hurricanes, in particular ones that fail to make land. For details see the article.

Is RSA Liable to Damages for Fraud?

According to a recent news story, the National Security Agency paid RSA, a major provider of encryption software, ten million dollars to promote a version of its software which the NSA could crack. Assuming the story is true, it would seem to put RSA at risk of being sued for deliberately selling its customers a product it knew to be defective.

"But several (RSA employees) said that RSA also was misled by government officials, who portrayed the formula as a secure technological advance."

It is hard to see why NSA would have to pay RSA to use more secure technology, hence equally hard to believe the claim.

It should also put RSA at risk of prosecution for criminal fraud, fraud being both a crime and a tort. But not much risk, since criminal prosecution is controlled by the government, the same organization a different branch of which, assuming the story is correct, bribed RSA to sell a deliberately crippled product.

Science Has Almost Caught Up with Science Fiction

Beyond This Horizon, an early Heinlein novel, describes a technology for what I like to describe as libertarian eugenics, a way in which a couple can choose, among the children they could have, which one they do have. The central trick is a way of separately analyzing the genes of sperm and egg before combining them.

The obvious problem is how to analyze a cell without damaging it. Heinlein's ingenious solution takes advantage of the fact that egg and sperm each contain half of the full set of genes of the cell from which it is derived—and the process that produces an egg or sperm also produces other bodies with the rest of genes. Analyse an ordinary cell, analyse the extra body, subtract, and you now know what genes are in egg or sperm.

According to a news story published yesterday, someone is finally attempting a version of Heinlein's idea, looking at the polar body that contains the genes that are not in the egg in order to deduce the genes that are. At least, that is what the story seems to be describing, although there are not enough to details to be certain.

Heinlein published the novel, and the idea, in 1942.

Deducing the Past: The Maya as an Example

One of my students is doing a paper on the Maya for my Legal Systems Very Different from Ours seminar, and I recently spent some time following up her references and whatever else I could find online. The most interesting conclusion about the institutions was the evidence for parallel evolution, the fact that some were strikingly similar to institutions from unrelated Old World societies. But more interesting, to me, was looking at the sources of information. How does one find out things about a past civilization and how much confidence can one have in the results? In particular, how does one find out things about a civilization that was destroyed five hundred years ago so thoroughly that knowledge of its writing system was completely lost, making all written evidence from before the Spanish conquest unreadable.

It turns out that there are about six different sources of information on the Maya. Each has serious limitations, no one by itself is adequate to establish very much about them, but the combination may give us a fairly accurate picture of what Maya civilization was like. Here is my list:

1. Modern Archeology.

The advantage is that one can dig up ruins, artifacts, other physical remains of a civilization and date them. Physical objects, unlike written texts or oral tradition, can't lie or be mistaken.

The disadvantage is the problem of interpreting what you find—which may well depend in part on what you expect to find. As Chesterton pointed out, future archaeologists might conclude that the 19th century English believed the dead could smell things, as shown by the evidence of flowers in grave sites.

2: The oral traditions and current practices of the descendants of the Maya civilization.

The advantage of that source of information is that there are lots of people who are bilingual in one of the Maya languages and a modern language, so anthropologists who interview them can avoid the problem of making sense of an ancient language and an extinct system of writing.

The  disadvantage is that we do not know how much of what current Maya believe about events in the distant past is true, nor to what degree current institutions preserve the institutions of the distant past.

3. A book written in Spanish by a 16th century Spanish Bishop describing his observations shortly after the conquest.

The advantage is that it is written in a language we can read, using a writing system we can read, based on first hand observation.

The disadvantages are, first, that it is first hand observation by a single observer of a society very different from his own, and second that the observer had serious biases that may well have affected what he observed and recorded.

4. Surviving texts written shortly after the conquest, using the Latin alphabet to write in a Maya language:

The advantages are that those texts were written by people with first hand experience of the Maya civilization, using a writing system we can reliably read, in a language that, while it has surely changed somewhat over the centuries since, still exists.

The disadvantages are, first, that there are very few such texts and they do not tell us nearly as much as we would like to know, and second that much of what they tell us may not be true. Maya civilization dates back several thousand years. The period that modern scholars view as its height ended about six centuries before the Spanish arrived. The texts tell us what some sixteenth century Maya believed about their own past—but consider how accurate a picture one would get of fifteenth century Europe based on a book written by a random modern European.

5. Surviving codices, written in a Maya language using the Maya script.

The advantage is that they were written by people about their own civilization, back when it was still a going concern.

The disadvantages are, first, that there are very few of them, almost all having been deliberately destroyed by the Spanish, second that they are written using a writing system knowledge of which was lost more than four hundred years ago, third that they were written for their authors' purposes not ours. That implies not only that they will not tell us all the things we want to know but also that some of what they tell us may not be true, may, for instance, be propaganda in some ancient political or religious controversy. Consider the picture one would get of the European past from a single surviving copy of Mein Kampf or the Malleus Maleficarum.

The problem of the writing system, while serious, is  less serious now than it was fifty years ago, since scholars now believe they have cracked the code, figured out almost all of the relation between the written symbols and the spoken language. For a detailed account, see the Wikipedia article.

It turns out, assuming the current view is correct, that in the Maya script (as, oddly enough, in Japanese) a single symbol can represent either a word or a syllable. This raises a potential problem of interpretation. Once the syllabic system has been worked out, that plus the surviving version of the language makes it possible to read text represented by its sound. But reading text represented by ideographs requires one to first figure out, possibly by context, what word each symbol stands for. The result might well depend on what the translator expects to find. I do not know enough about the subject to guess how serious that problem is in practice.

6. Surviving inscriptions.

The advantage is that these provide first hand information scattered through both time and space, giving us a picture of Maya civilization not only in its final years but throughout its history.

The disadvantages are, first, whatever problems remain with interpreting the writing system and languages, second the fact that inscriptions on buildings and monuments record what those responsible for inscribing them wanted said, which, as in the previous case, may be neither what we want to know nor reliably true. It would be nice if we had something analogous to the Mesopotamian clay tablets, a large collection of more or less random written material, but we don't.

My conclusion is, first, that we have grounds for a fairly extensive description of Maya civilization, and second that I would have to invest a very large amount of time and effort to form a confident opinion of how good those grounds are, how confident we can be that the description is accurate.

Will the Obamacare Mess Hurt Hillary?

The current and continuing problems with Obamacare have badly hurt his ratings and generated a flow of gleeful commentary by his critics. Obama, however, is in his second term. One possible interpretation of his many times repeated claim that anyone who liked his present coverage could keep it, a claim he knew would be proved false, was that it was an endgame strategy based on his knowledge that he would never have to face the voters again. That is not true for the congressional democrats who supported Obamacare, some of whom may well be defeated next year as a result. But what about the election after that? At the moment it looks as though Hillary Clinton will be the Democratic candidate. If so, how will she be affected?

Three different answers occur to me:

1. She won't be hurt. Her involvement with the administration was only relevant to foreign affairs, so she cannot be blamed for domestic failure.

2. She will be hurt. Obama's prize program was Obamacare. Hillary's, a few administration's back, was Hillarycare. While her supporters will argue that the two programs were very different, they were similar enough that support for the one will be seen as support for the other.

3. She will be helped. By the time she is running, the American health care system will be in a state of catastrophic failure. She will offer an alternative, probably along the single payer lines supported by many on the left.

My feeling is that the third alternative is the least likely, but I wish I could be more confident of that.

My Mac Lied To Me

My daughter, explaining why she was reluctant to upgrade her OS if not necessary, complained that a previous upgrade resulted in the computer reporting a program as damaged when the only thing wrong with it was that it was not from the Apple Store or an identified developer. A web search for the text of the message eventually provided her the explanation—to run the program she had to change her security settings to permit applications downloaded from anywhere to run. Prior to upgrading the system (to OSX 10.8.5) the program had run with no message, so she concluded that the upgrade had either added a new "feature" or altered the security setting without telling her it was doing so.

My initial reaction was skepticism. While I could understand Apple warning a user against a program not from a trusted source, describing such a program as damaged is a lie. The appropriate response would be a warning that the current security setting will not allow the program to run because it is not from a trusted source. That would allow a user to decide whether or not to change the security setting. The result of reporting the program as damaged could well be to cause the user to waste quite a lot of time trying to locate an undamaged copy.

I have now checked her report, and it was correct. My desktop is also running 10.8.5. I set my Security & Privacy settings to "Allow applications downloaded from Mac App Store and identified developers." I went to the site for maptool, the program with which she had encountered the problem, and downloaded maptool-1.3.b89.dmg. When I double clicked on the maptool program on the disk image, I got a message:

"MapTool-1.3.b89" is damaged and can't be opened. You should eject the disk image."

I then reset my security settings to "Allow applications downloaded from: Anywhere" and double clicked on the program. I got a warning that it had been downloaded from the internet, clicked the "open" button, and it opened.

Apple is in fact deliberately lying to its users.

For Cold Weather: A Recipe

A few days ago, my wife and daughter and I had mulligatawny soup at a local Indian restaurant. All three of us liked it. One of my current projects is getting our chest freezer empty enough so that we can defrost it. Included in its contents were several large containers of frozen chicken broth, produced as a side effect of a slow cooked Chinese chicken recipe I am fond of making.

Also, it has been cold out, and the web is a good source of recipes.

This is what I came up with, based mostly on this recipe scaled down to fit in a blender:

Mulligatawny Soup

1/2 cup  lentils                                     

7 cups chicken broth, divided

4-1 inch chunks fresh ginger, peeled                           
1 medium tomato

2 medium carrots,                              

1 celery rib

1 1/2 T butter                                                   
1T olive oil

1 T garam masala                                       
1/2 T ground cumin

1/2 T ground coriander                                   
1/2 t ground turmeric

1 medium to small onion, coarsely chopped                                  

4 garlic cloves, pressed

3T  flour
1 t tomato paste                                           
~1/2 T salt

~1/2 t pepper

Simmer lentils in 3 cups of chicken broth for about 30 minutes, until soft.

Combine 4 cups of broth, chopped carrots, celery, tomatoes and ginger in a blender or food processor. Blend and set aside.

Melt butter, add olive oil in a sauce pan and sauté spices, onion and garlic, stirring frequently so as not to burn, for about 5 minutes. Stir in flour, cook about 1 minute. Add tomato paste, stir. Add the contents of the blender, bring to a boil, cover and simmer for about 20 minutes.

Put the contents of the saucepan plus most of the cooked lentils back in your blender or food processor and puree until smooth. Return to pot along with the remaining lentils and the broth they were cooked in. Season with salt and pepper to taste, bring back to a simmer, turn off, serve.

A tasty spicy soup for cold days.

I should probably add that what I used for pureeing was my new vita-mix, a very high powered blender. My guess is that a food processor would work, but I haven't tried it.

Symbolic Thrift

American law schools, including the one I teach at, currently face serious budget problems due to declining enrollment. The obvious response is to try to cut expenditures. A particularly visible example, at least in our case, is abandoning the practice of serving catered food at faculty meetings and similar events. 

My guess is that the total amount saved is a tiny fraction of the budget, but I think that fraction understates the effectiveness of the change for two different reasons, both in some sense symbolic. The first is that it is hard to persuade other people that they should be careful to hold down expenditures if you are not doing so yourself. Catered meals are a visible extravagance provided mostly for the benefit of the faculty—who, to a considerable extent, run the school. Abandoning them is a way of signalling staff members that they too should be willing to make do on less money, even if it makes life a little harder and less pleasant for them.

The second reason is one that I intuit better than I can explain; it has something to do with the different feel of different human organizations. Consider at one extreme a loving family where every member takes it for granted that he ought to take account of the welfare of the other members in his decisions. Consider at the other extreme a bureaucratic organization, public or private, where the individual concern is not with the consequences of his acts but with the paper trail, his ability to prove to the satisfaction of his superiors that he has done what he should do, whether or not it is true.

Most organizations lie somewhere between those two extremes, depending in part on their size—it is easier to know and care about four other people than four thousand. Most organizations, however large and bureaucratic, make some attempt to take advantage of the family level feelings in order to motivate their members to act in the interest of the organization and the other members, but large formal organizations are less likely to succeed than small informal ones. How successful they are depends in part on how much the organization feels like a family, how much like a bureaucracy.

Which is one reason why, when my school stopped serving lunch at faculty meetings, I started bringing chocolate chip cookies.

Electric Blue Book: An Open Source Proposal

Some thirty years ago, when I was a professor at Tulane Business School, we had a project to develop software for taking exams on computers. It never got completed, but such software now exists, developed later by other people. Our main design objective was to make taking exams easier for students and grading them easier for professors. The main design objective of the current software appears, as best I can tell, to be preventing students from cheating, understandable but depressing.

In a post some years back I discussed the ideas for our software, to be titled "Electric Bluebook," and offered one possible solution to the cheating problem. An alternative  would be to combine our design features with something similar to what the currents software does, mechanisms to lock the user out of both the internet and his own hard drive while taking the exam. How difficult that would be I do not know.

As best I can tell from using the current commercial software, it has none of the features I want, aside from making life easier for students who find typing on a computer easier than writing on paper and solving the problem of reading students' handwriting. A quick look at the producer's web page did not suggest otherwise, but I might be missing something. There are open source programs already in existence for taking exams online, but I have not found any equivalent of Electric Bluebook, software to make taking ordinary exams easier for students and grading them easier for professors. If a reader knows of such software—you can get a picture of what I want from my earlier post—by all means let me know.

The purpose of this post is to see if anyone out there would like to start an open source project to revive Electric Bluebook.

Libertarian Arguments for Income Redistribution

Matt Zwolinski has recently posted some possible arguments in favor of a guaranteed basic income or something similar. While the position is not one popular with libertarians—the only other example that occurs to me is In Our Hands by Charles Murray—it does raise some interesting questions. 

Matt offers three different arguments. The first is that a guaranteed income or something similar would be an improvement on our present system of welfare. That is probably true, especially if you imagine it replacing not only welfare but all policies, such as the farm program, that are defended as helping poor people. The problem, as Matt appears to realize, is that if a guaranteed minimum income is introduced it will almost certainly be an addition to, not a substitute for, current programs.

His third argument is that a guaranteed income is a good thing for reasons that libertarians, among others, should recognize. One version of that is to point out that private charity faces a public good problem, hence that we are on net better off if government taxes us to provide the charity that each of us wants provided but would prefer that other people pay for. This is not a particularly libertarian argument, but it is  essentially the same as one that many libertarians accept in the context of national defense. 

One problem with the argument here is that we do not have any way of setting up mechanisms for income transfer that can only work in the way we would want them to. Once those mechanisms exist, individuals will try to game or alter them in order to be transferred to rather than from. That will impose real costs—resources spent gaming existing rules and lobbying to change them. And we may end up, as we often have in the past, with transfers that go up the income ladder rather than down or in all directions at once.

The most interesting part of Matt's essay, and the most libertarian part, is the second argument. As he points out, the existing state of the world is in part a result of past rights violations. Land claims in libertarian theory may be based on a series of voluntary transfers beginning with the person who first mixed his labor with the land, but many land claims in the real world run back to an initial seizure by force. Similarly, claims to other forms of wealth must be justified, in libertarian moral theory, by a chain of voluntary transactions back to a first creator. In at least some cases that chain is interrupted by involuntary transactions. Consider a house built by slave labor. Is the legitimate owner the person with the present title to it or the heir of the slaves forced to build it, or is it perhaps partly the legitimate property of one and partly of the other? What about property in other forms inherited through a chain that leads back to a slave holding or slave trading ancestor who owed, but never paid, compensation to his victims?

Most libertarians would recognize this as a legitimate problem, although many might point at the practical difficulty of establishing just ownership in such cases as justifying some sort of statute of limitations with regard to wrongs in the distant past. Matt's alternative, suggested by a passage he quotes from Nozick, is to argue that the descendants of those who gained by past rights violations are on average better off than the descendants of those who lost, hence redistribution from richer to poorer in the form of a guaranteed minimum income represents an approximate rectification for past injustice.

While the argument suggests that transfers from richer to poorer might do a better job of rectification of past injustices than random transfers, it does not imply that such transfers do a better job than doing nothing, that they on net reduce injustice rather than increasing it. Some present wealth may be due to causes that are, from the standpoint of libertarian moral theory, unjust, but not all. If I justly owe you forty cents, taking a dollar from me and giving it to you makes the resulting distribution less just, not more. Unless most inequalities are inherited from past rights violations, a claim I think few libertarians would support, the logic of the argument breaks down.

A brief digression is needed here to distinguish between wealth due to past rights violations and wealth inherited from such violations.  Suppose I am one of two doctors in a town. Someone murders the other. The result is to increase demand for my services, hence my income. I have benefited by the murder but I have not violated any rights, so do not owe the victim's heirs any compensation. The current distribution of income is due to all sorts of events in the past, some of them unjust, but it does not follow that everyone who is better off as a result of past history, even past rights violations, owes a debt to everyone who is worse off as a result.

A further problem with Matt's argument is that, even if you believe that a guaranteed basic income reduces net injustice, it is hard to argue that it is the best rule of thumb for the purpose. Consider the case of Afro-Americans. Almost nobody whose ancestors immigrated to the U.S. after the Civil War is the heir of benefits created by violation of the rights of their ancestors by his ancestors. On the other hand, the ancestors of present-day Afro-Americans were enslaved by Africans to be sold to European slave traders. The present inhabitants of Africa, at least sizable parts of it, are more likely than the present inhabitants of North America to be descendants of  people who owe, and did not pay, reparation to slaves and their descendants. 

It follows that Matt's second argument implies that the (very poor) present inhabitants of Africa owe compensation to the (relatively rich) present American blacks. I do not think Matt would accept that argument, whether or not he could rebut it. If so, he does not really believe in his second argument.

Readers sufficiently interested in these issues my want to look at the draft of one of the new chapters for the third edition of The Machinery of Freedom which discusses some related arguments for redistribution.

Lying for a Good Cause

While waiting for a dentist's appointment this morning I took a look at a Time magazine from last month and was mildly irritated by its attempt to defend Obamacare. The author described a deliberate lie about people being able to keep their insurance if they wanted to as the administration being insufficiently clear—I do not have the magazine with me so can't offer an exact quote. And he echoed the Administration talking point that represented all existing policies that did not cover everything the ACA requires, including contraception, as worthless junk that people only bought because they were desperate for insurance. Which left me wondering about the author of that particular piece of partisan puffery disguised as news commentary. My guess is that, dosed with truth serum or in a sufficiently private conversation with a trusted friend, he would admit that the Administration's claim was a deliberate lie but justify it on the grounds that it was necessary in order to get a good law passed.

It is not an absurd position. "The end justifies the means" is usually stated as if it were obviously false, but it is not clear that it is. Most of us would be, I think should be, willing to do things we would usually disapprove of in order to achieve a sufficiently good result. To what degree ends do or do not justify means is, as it happens, one of the subthemes of my second novel, where an antagonist who is not a villain behaves very badly to my protagonists for good, from his standpoint sufficiently good, reasons.

It occurred to me to wonder if the author of the Time piece or others with similar views would accept the same argument applied to a previous instance and a different President, if they would agree that, while the facts it was based on might be mistaken, the moral reasoning was correct.

Imagine that you are President Bush and that you believe the following:

1. Saddam Hussein is a murderous tyrant whose people would be far better off without him.

2. If he is overthrown by the U.S., his government can be replaced by a reasonably free and democratic one which will serve as a model to convert other dictatorships in the region into free and democratic societies.

3. Points 1 and 2 will not be sufficient to persuade the American people to support an invasion of Iraq. They would, however, support such an invasion if they believed that Hussein was producing weapons of mass destruction.

4. While it is possible that Hussein is producing weapons of mass destruction, the evidence is very strong that he is not.

Would you be justified in pretending to have good evidence of WMD's in order to get sufficient public support to make possible a U.S. invasion of Iraq?

The logic is the same as in the case of Obamacare—lying to the public in order to make possible policies you consider highly desirable.

In both cases, the argument hinges on factual beliefs. Point 2 above turned out to be strikingly false. Obamacare will, I think, turn out to have been a very serious mistake. But the question I am asking is not whether the beliefs were correct but whether the moral argument is. If Bush believed points 1 to 4, was he justified, in terms of those beliefs, in lying to the American people? If Obama believed that the ACA would greatly improve American health care, was he? 

And, perhaps most interesting, would people who answered "yes" to the second question be willing to give the same answer to the first?

Legal Insurance as a Commitment Strategy: A Solution to the Patent Troll Problem

One intellectual property issue that has gotten a lot of attention lately is the problem of non-practicing entities, referred to by their critics as patent trolls—firms that buy up an inventory of patents not in order to practice them but in order to sue other firms for infringing them. Patents protect ideas and ideas have fuzzy boundaries. That makes possible a strategy of suing on weak cases not because you expect to win but in order to be paid to drop the suit. It may make sense for the targeted firm to settle for an amount less than the cost of litigating the case  if it wins and considerably less than the cost if it loses. A large firm that knows it could be targeted multiple times might go to court in order to deter future suits, but that makes less sense for a small firm, especially one that will be bankrupted if it loses and badly hurt even if it wins.

The purpose of this post is to suggest a way in which small firms could protect themselves against such suits and ask why it is not in common use. The solution I am proposing is litigation insurance to cover most or all of the cost of defending against such a suit. While most insurance costs more than it is worth from an actuarial point of view, this kind of insurance might well be worth more than it costs.

Ordinary insurance costs its customers more than it pays them for two reasons. The obvious one is that, in addition to paying out claims, an insurance company has to rent offices, hire claims adjustors and salesmen, bear a variety of additional expenses. The less obvious reason is that insurance reduces the incentive of the insured to take costly precautions against whatever the insurance covers. If my house has no fire insurance, it is worth it to me to take any precaution—installing  a sprinkler system, say—whose expected payoff in reducing the chance that the house will burn down is greater than its cost. If I am insured for 90% of the loss, it only pays to take those precautions whose benefit is at least ten times their cost. The inefficiency of worthwhile precautions not taken reflects what economists refer to as moral hazard. The benefit that balances those costs is the fact that the money the insurance company pays comes when something has happened, such as my house burning down, that makes money especially valuable to me.

Litigation insurance has those costs as well but, for a firm likely to be targeted by a patent troll, it also has a large potential benefit: Deterrence. Insurance which covers litigation cost but not out of court settlement makes settling less of a temptation, increasing the chance that the suit will go to court. The more likely it is that the plaintiff will have to litigate, the less profitable it is to bring the suit in the first place. Insofar as the effect of the insurance is to keep you from being sued, it provides a benefit that costs the insurer nothing. 

I have a (currently unpublished) article that uses what I have learned by looking at feud societies such as saga period Iceland to explain why non-practicing entities present a special problem for patent litigation and borrows an idea from the legal system of Periclean Athens to suggest a legal solution. The private solution I am suggesting here is in part inspired by yet another legal system I have written on, privately enforced criminal law in 18th century England. Part of what made that system work was the existence of associations for the prosecution of felons, the members of which paid in advance into a fund used to pay the cost of prosecuting anyone who committed a felony against one of them. The real function of such associations, I believe, was not insurance but deterrence: The membership list of the association was published in the local newspaper for the felons to read, in order that they would know who not to target.

Which leaves me with a puzzle: Why isn't the equivalent approach used against patent trolls—or if it is, why doesn't it work?

The Third Edition of the Machinery of Freedom

I have just finished drafts of the  chapters for the new edition and webbed them for comment.

The Rush to Judgement

A friend or acquaintance comes to you with a story of how badly he has been mistreated by someone—his employer, his girlfriend, a store, an airline. He expects you to agree with his complaint, take his side, despite the fact that you have not heard the other side of the argument and so, unless you happen to have some other source of information, have no way of knowing whether his side is correct. Your honest response would be to point that out—at which point he will get mad at you too.

Seen from a sufficiently cynical point of view the pattern makes sense. Agreeing with him makes him your ally, allies are useful, and the target of his attack is far away and, with any luck, will never know you have sided against him, her or it. Agreeing is also stupid if it does not occur to you that you have heard only one side of the story or if you have not yet learned how dangerous it is to reach conclusions on that basis,  dishonest if you have.

I was reminded of this particular recurrent irritation by recent news stories about a waitress who claimed to have been stiffed by a couple she served, given a note criticizing her (lesbian) life style in lieu of a tip. Her original account did not identify the couple, but it provided sufficient information for them to identify themselves—at which point they provided what looks like convincing evidence that she was lying, including the visa charge for their dinner, tip included. The most recent story I have seen includes comments by friends and former colleagues of the waitress reporting a history of minor lies designed to provoke sympathy on the basis of invented stories.

What struck me was not the behavior of the waitress but the behavior of the large number of people who took her side, including reporters who took the waitress's initial story as gospel, reporting it as something that happened, not as something someone claimed happened, despite no evidence beyond a digital image of what purported to be the check with note and without tip. Judging at least by reports, thousands of people on Facebook condemned the supposed behavior of the couple—with no evidence beyond the news stories—and many sent donations to the purported victim. 

Their behavior was stupid and unjust. The behavior of the reporters was also professional incompetence. 

One question about the story that nobody else seems to have commented on occurred to me. All of the reports describe the waitress as an ex-marine. She is also described as 22 years old, and the most recent story mentions "a day care center where she once worked." The minimum age of enlistment for the marines is 17. The usual terms of enlistment are for three to five years of active service. Marine corps training requires an additional three months. It is not impossible that someone could have enlisted at 17 on the shortest terms, left the corps at 20 and by 22 have worked first at a day care center and then at a restaurant, but the timing is sufficiently tight to be at least mildly suspicious, especially when combined with evidence that the person in question is a habitual liar. 

It would be nice to know if any of the reporters checked with the marine corps to make sure that "ex-marine" was not another fabrication.

Obama, Silicon Valley, and Learning by Testing

There were PhDs working as low paid data managers during Obama’s ’08 campaign and top product managers developing interactive during ’12 campaign. There are many talented developers/product managers/data modelers who would take a pay cut to work on something they believe in. Especially for those with enough life experience to know how important the Affordable Care Act is, even if it’s not an ideal solution.

The quote is from a comment on a very interesting essay about the failure of the Healthcare.gov website.  Part of the essay's point is the danger, in IT projects and elsewhere, of a particular approach to doing large projects:

The preferred method for implementing large technology projects in Washington is to write the plans up front, break them into increasingly detailed specifications, then build what the specifications call for. It’s often called the waterfall method, because on a timeline the project cascades from planning, at the top left of the chart, down to implementation, on the bottom right.

Like all organizational models, waterfall is mainly a theory of collaboration. By putting the most serious planning at the beginning, with subsequent work derived from the plan, the waterfall method amounts to a pledge by all parties not to learn anything while doing the actual work. 

It occurred to me that the comment, combined with that point, raised an issue that had probably not occurred to the commenter. The Silicon Valley people who worked to reelect Obama were acting on their view of Obama and his policies. The arguments of the essay imply that they ought to be willing to revise that view and alter their political activities accordingly as further evidence comes in.

If, as many sources seem to suggest, Obama did not realize that healthcare.gov was not going to work, and if the reason he did not realize it was that he had created a culture around him in which people did not feel free to pass on bad news to their boss, then he is not, and was not, competent to be President. If, as Obama himself implied in contrasting the failure of healthcare.gov to the success of the IT efforts of his reelection campaign, government is very bad at doing this sort of thing, that is at least some evidence that the ACA was a mistake, likely to make health care worse rather than better.

I wonder how willing his supporters in Silicon Valley will be to apply the "test and revise accordingly" approach to their own political views.

The Second Amendment in the 21st Century

A recent facebook post pointed me at an entertaining video in favor of gun control. The point of the video, surely correct, is that mass shootings were a lot less practical with 18th century firearms than with modern firearms. Its conclusion: "Guns have changed. Shouldn't our gun laws?"

There are two problems with the argument. The first is that gun laws have changed quite a lot over the past two hundred plus years. The second is that, while mass shootings get a lot of publicity, they represent only a tiny fraction of all killings.

There is, I think, a better argument to be made for the effect of technological change on the argument for the right to bear arms. As I interpret the Second Amendment, it was intended as a solution to a problem that worried eighteenth century political thinkers, the problem of the professional army. As had been demonstrated in the previous century, a professional army could beat an army of amateurs. As was also demonstrated, a professional army could seize power. Oliver Cromwell and the New Model Army won the first English Civil War for parliament and then won the second English Civil War for itself, with the result that Cromwell spent the rest of his life as the military dictator of England.

The Second Amendment, as I interpret it, was intended to solve that problem by combining a small professional army with an enormous amateur militia. In time of war, the size of the militia would make up for its limited competence. In time of peace, if the military tried to seize power or if the government supported by the military became too oppressive, the professionals would be outnumbered a thousand to one by the amateurs. It was an ingenious kludge.

It depended, however, on a world where the weapons possessed by ordinary people for their own purposes, mostly hunting, were as effective as the weapons possessed by the military. We are no longer in such a world. The gap between military weapons and civilian weapons is very much larger now than then. One result is that the disorganized militia, the population in general, no longer plays any role in military defense. Another is that, if there ever was a military coup in the U.S., ordinary civilians would be much less able to oppose it with force than they would have been two hundred years ago.

Civil conflict in a modern developed society is much more likely to be carried on with information than with guns—a government that wants to oppress its population does it by controlling what people say and know. It follows, in my view, that the modern equivalent of the Second Amendment, the legal rule needed to make it possible for the population to resist the government, has nothing to do with firearms. The 21st century version would be a rule forbidding government regulation of encryption. A government that has no way of knowing what who is saying to whom lacks the most powerful weapons for winning an information war.

There remains a strong argument for the right to bear arms, different from but related to its original function. People who are unable to protect themselves are dependent for protection on the police. The more dependent people are on the police, the more willing they are to tolerate, even support, increased police power. Hence disarming the population makes possible increased levels of government power and the misuse thereof, although for a somewhat different reason than in the 18th century.

Which is an argument against restrictions on the private ownership of firearms.

The Killer App for Google Glass

I can remember large amounts of poetry, but people's names, faces and the  information associated with them are a different matter. For the most part, I successfully conceal my handicap by a policy of never using names if I can help it, but once in a while the tactic fails. I still remember, as perhaps my most embarrassing moment, recommending Larry White's work on free banking to someone who looked vaguely familiar—and turned out to be Larry White. 

Help, however, is on the way. I first encountered the solution to my problem in Double Star, a very good novel by Robert Heinlein. It will be made possible, in a higher tech version, by Google glass. The solution is the Farley File, named after FDR's campaign manager. 

A politician such as Roosevelt meets lots of people over the course of his career. For each of them the meeting is an event to be remembered and retold. It is much less memorable to the politician, who cannot possibly remember the details of ten thousand meetings. He can, however, create the illusion of doing so by maintaining a card file with information on everyone he has ever met: The name of the man's wife, how many children he has, his dog, the joke he told, all the things the politician would have remembered if the meeting had been equally important to him. It is the job of one of the politician's assistants to make sure that, any time anyone comes to see him, he gets thirty seconds to look over the card.

My  version will use more advanced technology, courtesy of Google glass or one of its future competitors. When I subvocalize the key word "Farley," the software identifies the person I am looking at, shows me his name (that alone would be worth the price) and, next to it, whatever facts about him I have in my personal database. A second trigger, if invoked, runs a quick search of the web for additional information.

I am told that Google itself has a rule against building face recognition into glassware, so my Farley file software may not appear in the immediate future. But it is the killer app, and someone will build it.

What Should Replace Obamacare

A recent post on the Forbes site offers a convincing explanation of what was wrong with the current system of health insurance before Obama, hence what both it and Obamacare ought to be replaced by. Its central point is that what we call medical insurance is in part actual insurance, protection against low probability/high cost risks, in part prepayment of ordinary medical expenditures. The reason insurance policies take that form, also the reason that most of them are provided by the employer and so not portable, is that employer provided health insurance is bought with pre-tax dollars, ordinary medical care with after tax dollars. 

One result is that individual consumers have little incentive to be careful shoppers for health care services, since for the most part they are not the ones paying for them. A second is that insurance companies, in order to provide a substitute for careful shopping by customers, require a lot of paperwork from providers, driving up their costs. Costs are also driven up by state regulations that require insurance companies to cover things that the customers might prefer not to pay to have covered—the same problem that Obamacare produces on a national scale. In my state, California, for example, health insurance must cover acupuncture, and in Connecticut it must cover hair prosthesis.

One implication is that tax law should be changed to put employer provided insurance, privately purchased insurance and payments for uninsured medical expenditures on the same footing. To get the economics right, all should be treated as ordinary consumption expenditures. From the standpoint of the relevant politics, however, what the Republicans ought to propose is to make all three tax deductible, at least up to the level of what most people now pay. It's a lot easier to sell a tax cut than a tax increase.

A second implication is that insurance companies should be allowed to sell policies interstate. That would eventually eliminate inefficient regulatory requirements, since state insurance regulators would have to compete with each other to provide regulations that generated the policies consumers wanted to buy. In this case as in many others, competition is a good thing.

A well written and informative article by someone I am pretty sure I interacted with online many years ago. It's a small world.

Multitasking or Parallel vs Serial Thinking

It is useful to know what one is good at, but also what one is bad at. 

The example I am thinking of is multitasking, doing and thinking about several things at once. The first clear evidence of my inability to do it well appeared decades ago in the context of my medieval hobby, which included combat with medieval weapons done as a sport. I was much worse at melee combat—one group of fighters against another—than at single combat. In single combat I only had to focus on the opponent I was fighting. In melee, I had to be, or at least should have been, simultaneously keeping track of everyone else near me. And I wasn't.

The same problem showed up much later in the context of World of Warcraft. Group combat there, a raid with a group of from five to forty people, requires the player to keep track of what he is doing, what other people in the group are saying—in the form of typed messages on the screen—and other things going on around him. I focused on what I was doing and frequently missed important things other people were saying. Interestingly enough, that was less of a problem if the group was using software that permitted voice communication, so that one kind of information was coming in mostly through my ears, another through my eyes. 

It is not just that paying attention to multiple things is hard. My daughter, playing the same game, can not only pay attention to everything in the game, she can also conduct one or two independent conversations, in typed text, while doing so. Pretty clearly, it is a real difference in abilities, whether innate or learned I do not know.

Thinking about it, it occurred to me that I had observed the same pattern in an entirely different context, the difference between how I think and how Richard Epstein, a friend and past colleague, thinks. I usually describe the difference as my thinking in series, Richard in parallel. It shows up when he is sketching the argument for some conclusion. 

A implies B. B implies C. C ...

At which point I demonstrate that B doesn't really imply C, that there is a hole in the argument. That is no problem for Richard, who promptly points out that A also implies B', a somewhat different proposition than B, which implies C', from which he can eventually work his way back to D, or perhaps E or F, and so to the conclusion that the original line of argument was intended to establish. Pretty clearly, he is running a network of multiple lines of argument in his head and only has to find some set of links in the network that gets him where he is going. I am focusing on running a single line of argument. Hence parallel vs series.

Nonconforming ≠ Substandard

President Obama may have secured a measure of political relief for himself by allowing substandard insurance policies to be renewed for another year.

… 

 

“The president told me that if I like my health insurance, I could keep it. And that shouldn’t have an expiration date,” said Crusco, who has been covered under a nonconforming plan that did not cover maternity care. That fit her needs because, she says, she doesn’t plan to have more children.”

One of the things that irritates me about news coverage of the Obamacare mess is the willingness of many in the media to describe plans that do not fit the requirements of the ACA as "substandard."

The two quotes above, both from the same news story, nicely illustrate the rhetorical trick. A plan that does not cover maternity care is nonconforming, since it does not conform to the ACA requirement that all insurance plans provide maternity benefits.  It is substandard for someone who does not plan to have children, possibly a man or an elderly woman who is unable to have children, only if one assumes that the standard of what all plans ought to cover for everyone is determined by what Congress wrote into the act, which, as the example shows, is crazy. It should not take more than about thirty seconds of thought for a fair minded journalist to realize that at least some plans that do not fit the ACA's requirements are what their purchasers do and should want. 

Which suggests that quite a lot of journalists are either incapable of thought or engaged in deliberately biased reporting.

More on Selective Enforcement as Legislation

My previous post raised the question in the context of Obama's apparent intent to unilaterally modify his healthcare legislation. But it is an interesting problem more generally. The theory of our system is that the legislature makes laws and the executive enforces them. But laws cannot, in practice, be perfectly enforced, so the executive is necessarily making the decision about what resources to allocate to enforcing what laws. Where can or should one draw the line between that decision and using selective enforcement to rewrite the law?

This is at least the third time that Obama has offered to do it. The first was when, during his first campaign, he said that under his administration federal marijuana law would not be enforced against people using medical marijuana in conformance with state law—a promise that he promptly broke. The second was when he announced that certain categories of illegal immigrants would not be prosecuted. Revising Obamacare is the third.

Imagine the following scenario at the state level. The governor of California proposes a bill to tax cars that burn gasoline but not electric cars. The bill fails to pass. He responds by announcing that he has instructed the state police that they should enforce speed limits strictly against gasoline powered cars but only stop electric cars if they are going at least twenty miles an hour over the speed limit. 

I am not a constitutional scholar and do not know whether there are legal limits to executive power that would prevent such a tactic. It is legal to selectively tax gasoline powered cars. It is legal for the police to devote their limited resources to catching some speeders but not others, for instance by patrolling highways where they believe speeding is a particularly serious problem. Is it legal to accomplish the substance of the former under the form of something like the latter? 

More generally, what are the limits of such an approach? The executive is not entitled to enforce a law the legislature has not passed. But is it entitled to selectively enforce one that the legislature has passed in order to achieve the effect of one that it has not passed?

Comments from those who know more about constitutional law, state and federal, than I do are welcome.

Selective Enforcement as Legislation

The poet Ibn Harma performed before the caliph, and so delighted was the Prince of the Muslims that he asked the poet to name his reward.

"The reward that I want from the Prince of the Muslims is that he send instructions to his officials in the city of Medina commanding that when I am found dead drunk upon the pavement and brought in by the city guard, I be released from the penalty prescribed for that offense."

"That is God's law, not mine," the Caliph replied. "I cannot change it. Name another reward."

"There is nothing else I desire from the Prince of the Muslims."

The Caliph thought a moment, then sent instructions to his officials in Medina commanding that if Ibn Harma was found drunk and brought in for punishment, he should receive sixty strokes of the lash as the law commanded. But whoever brought him in should receive eighty.

It is one of my favorite medieval Islamic law and economics stories. In theory, Islamic law is not made by the sovereign but deduced by legal scholars from the Koran and the Hadith, traditions of what Mohammed and his companions did and said. The Caliph accordingly could not change the law against drunkenness. He could not even change the punishment, since it is a Hadd offense, one with a fixed punishment deduced from the religious sources. He could, however, repeal it de facto although not de jure by changing the incentive to enforce it.

I was reminded of this by today's news. President Obama is attempting to forestall congressional efforts to alter the Obamacare legislation by doing it himself without appeal to congressional authorization. Presumably the theory is that, since he is in charge of the executive branch and the executive branch is in charge of enforcing the law, he can simply announce that the part of the law forbidding insurance companies from continuing to offer plans that do not meet the requirements of the new law will not be enforced, at least as far as existing customers of such plans are concerned.

This raises two questions. One is a question of constitutional law, whether what he is doing is in law, as it obviously is in fact, a violation of the division of powers between the legislative and executive branches. The other is a political question. Arguably, the political effects of the present mess will have at least partly died down over the next year. Is Obama making things worse for his party rather than better by pushing the next failure, the result of good risks choosing to keep existing plans and leaving the plans sold through the market to the bad risks, making them very expensive, to just before the next election?

How to Run a Restaurant

My previous three posts were serious proposals for changes that I thought worth making. This one is more nearly a puzzle. In the other cases I can see plausible reasons why the changes might not have occurred even if I am right in thinking them desirable. In this case, I take the nonexistence of what I propose as pretty strong evidence that I am missing something, that it is for some reason a considerably less good idea than I think.

When I sit down in a restaurant, I am consuming two different things—the food produced and the use of seat, table, heating or air conditioning, the part of the restaurant I occupy and the services it provides. I can choose to eat lots of expensive food fast, in which case I consume lots of the first and little of the second. Alternatively, I could order something inexpensive, perhaps a bowl of soup, and linger over it for an hour. 

Since the restaurant charges only for the food, I have no direct pecuniary incentive to economize on my consumption of space. Since the price of the food has to cover the cost of both food and space, I have too strong an incentive to economize on my consumption of food. If desert costs the restaurant a dollar to produce, is priced at three dollars, and is worth two dollars to me, I don't buy it—a net loss to me plus the restaurant of a dollar in potential surplus. 

The obvious solution to these inefficiencies is to price food and space separately. When I sit down, a clock at the table starts running. When I leave, my bill includes a certain amount per minute for the time, plus the cost of what I ordered. If I want to spend two hours chatting with a friend over tea and scones, I can do it without worrying about angry looks from the waiter—and pay for it. My total bill should average out about the same, since the combined bill still has to cover the same costs. But now the separate cost of sitting and of eating is being billed separately, giving me the right incentive with regard to each.

The puzzle is why no restaurant, so far as I know, is run that way. Some have crude approximations, such as a cover charge. But why not simply price food and space separately, just as rental cars sometimes price use of the car, mileage, and gas separately?

How to Admit Students to College

Colleges base their admission decisions on a variety of different criteria. One of them is how well the student can write. At present they have two ways of measuring that, neither of which is worth much.

One way is by the SAT writing exam. The problem is that consistent grading across a large number of students requires something close to machine grading, human graders checking the essay against a simple and objective set of criteria. That might tell you how well the student has trained for the test but it is not very good evidence of how well the student can write.

The other way is by having a prospective student send in an essay for the admission people at the college to evaluate. However good a job they do of evaluating the essay, they have no way of knowing who wrote it. The applicant may have written it entirely himself, he may have written it himself and had it gone over by someone more expert in writing, he may have hired someone to write it for him. I have no inside knowledge, but given how important college admissions have become I would be astonished if no such market exists.

There is a simple solution. Many applicants visit a college before applying. As part of the process, put the applicant in a room with a computer and a list of topics and give him an hour to write an essay. If the applicant is not going to visit the college, perhaps there is an alumnus living near him who would be willing to provide the computer and monitor the writing. If multiple colleges want applicants to write essays under controlled conditions, it should be in the interest of someone, perhaps the organization that now administers SAT exams, to arrange suitable facilities in cities scattered across the country.

It seems like an obvious idea and I do not know why, so far as I can tell, it has not yet happened.

How to Buy a House

If you go to a real estate agent in search of a house, there are two questions you are likely to be asked. One is how much you want to spend. The other is what sort of house you are looking for—how large, how many rooms of what sort, in what location. 

It is in the agent's interest to ask the first question; since his commission is a percentage of the sales price, he would like to sell you the most expensive house you are willing to buy. It is not clear that it is in your interest to answer it. Even if you can afford a two hundred thousand dollar house, you might prefer one that fits your requirements a little less well and costs substantially less. If you tell the agent that you are willing to spend two hundred thousand dollars, he may decide not to show you any house that will sell for much less than that.

The second question raises another problem—what you want to buy depends on what it costs. You would prefer a house with a bedroom for you and your wife, a bedroom for each child, and an extra room for a home office—but if an extra bedroom increases the price of the house by too much, you could put your desk at one end of your bedroom or persuade two children to share a room. In order to give a sensible answer to the question, you need a price list, a description of how much more you can expect to pay for a larger house, one with more bedrooms, one in a better location. If you had such a list, you could figure out about what sort of house you wanted to buy and what it should cost and ask the agent to select houses to show you accordingly.

So far as I know, no such price lists are currently available—but there is no good reason why they shouldn't be. Realtors have access to extensive information on houses that have sold. In any area with a sufficiently lively real estate market, it should be possible to use conventional statistics to work out from that data about how much more a house costs with one more bedroom, all else held constant, or with an additional hundred square feet of area, or with a larger yard, or in a better school district. My proposal is that somebody, perhaps the existing multi-lister service that provides the data to the realtors, should do so. The realtor could then provide his customer with a price list, the customer could decide about what sort of house he wanted, and the realtor could proceed to find the houses that came closest to fitting what his customer wanted.

I should  add that it is almost twenty years since I was last in the market for a house. If what I have just proposed is, at this point, common practice in the industry, perhaps one of my readers with more up to date information can tell me.