Previous Month | RSS/XML | Current | Next Month

December 28th, 2005 (Permalink)

The Limits of Expertise

Louis Menand has a fascinating review in the New Yorker of a new book that touches upon a neglected issue for appeals to expert opinion: what are the limits of expertise? There are always people who claim expertise in some area, and some even make money giving advice about it. However, in some areas anyone's opinion is as good as anyone else's―or as bad. Not so surprisingly, one such area appears to be political prognostication. One interesting result reported in the book is that experts actually make worse predictions than nonexperts as a result of overconfidence, which is a fault more common among the highly-educated.

In addition to overconfidence, Menand discusses other cognitive biases which affect experts as much as nonexperts:

The future will always be full of surprises, and so we should be skeptical of those who claim to predict it, whether they read tea leaves or history. I won't predict it, but I hope that you will read the whole thing and follow Menand's advice to think for yourself!

Source: Louis Menand, "Everybody's an Expert", New Yorker, 11/28/2005

Update (1/8): Carl "The Numbers Guy" Bialik's latest column also discusses this book, which is by the psychologist Philip Tetlock. Menand's review receives an interesting mention:

The New Yorker's review of [Tetlock's] book surveyed the grim state of expert political predictions and concluded by advising readers, "Think for yourself." Prof. Tetlock isn't sure he agrees with that advice. He pointed out an exercise he conducted in the course of his research, in which he gave Berkeley undergraduates brief reports from Facts on File about political hot spots, then asked them to make forecasts. Their predictions―based on far less background knowledge than his pundits called upon―were the worst he encountered…. "Unassisted human intuition is a bomb here," Prof. Tetlock told me.

This seems to verify Pope's famous dictum that "a little learning is a dangerous thing", since the students did worse than the experts, who were no better than dart-throwing chimps! So, the students would have been better off to stay ignorant and just make random guesses.

As advice, this leaves us in a no-win position. If the experts are incompetent at predicting the future, what else are we to do but to think for ourselves? We can try to avoid predicting the future as much as possible, but as citizens in a democratic society we often have to estimate probabilities of future political events in order to decide how to vote. However, we can at least learn to practise intellectual humility in the face of the uncertainty of the future: to estimate probabilities hesitantly, cautiously, and to willingly change them in the fact of contrary evidence.

Source: Carl Bialik, "Evaluating Political Pundits", The Numbers Guy, 11/28/2005

December 27th, 2005 (Permalink)

A Prediction Puzzle

Which of the following events is most likely to happen next year? (To see the correct answer, click on the event which you think is most probable.)

  1. President Bush will cut spending for homeland security.
  2. President Bush will cut spending for homeland security and open up parts of the Alaskan wildlife refuge to oil drilling.

December 21st, 2005 (Permalink)

Bush's Weak Analogy

The New York Times has recently reported that the National Security Agency has conducted warrantless eavesdropping on communications into and out of the United States. President Bush has criticized the Times' reporting by comparing it to an earlier incident in which the Washington Times reported that Osama bin Laden used satellite phones, which supposedly led to bin Laden's changing his way of communicating.

I'm no expert in such matters, but Daniel Benjamin makes a convincing case that the two Times cases are significantly different, so that an analogy between them is weak. Specifically, the N. Y. Times' reporting the mere fact that warrantless eavesdropping has taken place does not compromise any intelligence sources or methods.

Also in Slate, Jack Shafer is skeptical that the Washington Times story is actually to blame for bin Laden's ceasing to use satellite phones. It may be a case of post hoc reasoning, in that bin Laden stopped using such phones shortly after the story, though there had been previous news articles that had mentioned his use of the phones.


December 17th, 2005 (Permalink)

Letter to the Editor

George Will's article "Free Speech Under Siege" (Dec. 5) reminds me that most conservatives have either lost touch with reality or have become so buried in the right's false rhetoric that they actually believe the propaganda they peddle. He claims liberals have a "program of extending government supervision of life." It is what the right has been saying for years, and it is purely misleading. It was the conservatives who tried to decide what was best for Terri Schiavo. It is conservatives who want to tell a woman what to do with her body. It is conservatives who want to tell us whom we can and can't fall in love with. Then they confuse liberal programs that help people with big government. Will tries to claim that campaign-finance reform―supported by Democrats and Republicans alike―is an attack on free speech. Campaign-finance reform is the last, best hope we have to limit career politicians. Attacks on free speech are more about calling people anti-American who don't agree with the war in Iraq, or limiting the press coverage of fallen soldiers' coming home in coffins.
Glenn DiTomaso
Brockton, Mass.

DiTomaso is right that many conservatives, though not all, aim to extend the power of government in some ways. However, this doesn't tend to show that many liberals don't also aim to extend the power of government as well, though in different ways. That is why tu quoque arguments, like this letter, are red herrings―that is, logically irrelevant distractions―and, thus, fallacious. Will's article may well be wrong, but this letter doesn't begin to show that it is.

Source: "Mail Call", Newsweek

December 14th, 2005 (Permalink)

Q&A: Types and Tokens

An anonymous reader wrote to inquire about type/token ambiguity. A type is a category of thing, that is, a collection of things that have something in common, for example: book. A token is a particular thing of some type, for example: the book Nineteen Eighty-Four by George Orwell.

Type/token ambiguity comes about because a word can sometimes mean both a type and a token, and it can be unclear in which sense it is used. For instance, I own two copies of Orwell's 1984. Are these the same book or two books? If I were counting the total number of "books" in my library for purposes of estimating how much shelf space they require, I would count them as two books. However, if I were counting the "books" in my library in order to catalog them, I would count them as two copies of one book. Since one is not equal to two, this shows that the word "book" is used in two different senses: in the first case, I am counting tokens of books; in the second case, I am counting types of books.

In other words, 1984 is both a token of the type "book", and it is also a type of book, of which individual copies of 1984 are tokens. A particular copy of 1984, such as the one currently on my shelf, is both a token of the type "1984" and of the type "book". Thus, "book" is ambiguous between a physical object made of paper and ink, and an abstract object made of words and sentences.

If this sounds confusing, it's because it is. Hence, type/token ambiguity is at least a logical boobytrap, which could cause someone to commit a fallacy of ambiguity. I'm not sure that such fallacious arguments are common―though some philosophers have thought them common among other philosophers―so I hesitate to speak of type/token ambiguity as a fallacy.

Nonetheless, an attempt to avoid type/token ambiguity is the reason for some of the unusual technical vocabulary used in the Fallacy Files. For instance, the word "fallacy" itself suffers from type/token ambiguity. I use the term "fallacy" strictly in the type sense, that is, a fallacy is a common type of bad argument. However, the word is frequently applied to particular fallacious arguments: "that argument is a fallacy". Most logicians use the word in both senses.

Another important logical word which is affected by type/token ambiguity is the adjective "valid". "Valid" is used by most logicians to apply both to types of arguments―such as Modus Ponens―and to particular arguments―such as tokens of Modus Ponens. As with most types of ambiguity, context usually clears up which sense is meant, so that fallacies are seldom committed. However, in order to avoid any possible confusion, I limit the term "valid" to token arguments, and use the unusual word "validating" for types of argument.

Using such technical vocabulary may be like wearing both a belt and suspenders, but it is incumbent upon a work on fallacies to avoid even the appearance of ambiguity.

December 6th, 2005 (Permalink)

The Missing Link

I used to criticize the "Bushisms" feature in Slate because it was a regular source of contextomies, the online magazine never linked to the sources of its quotes so that readers could examine the context, and it seldom if ever published corrections.

Eventually, about a year and a half ago, after being tirelessly hounded by Eugene Volokh, and also criticized by the well-respected Spinsanity, Slate began to include links to the sources of its "Bushisms". At the same time, the contextomies stopped. Now, as Volokh points out, Slate is back with another contextomized "Bushism", and the link to the context has disappeared at the same time! Is this a coincidence, or does Slate want to make it difficult for its readers to catch it quoting out of context?



December 3rd, 2005 (Permalink)

Headline Update

Here's a sad update to the headline from last month: "Pandas married with city-wide ceremony in Thailand":

Panda's wife files case against him

Divorce so soon?

Via: James Taranto, "Best of the Web Today", 12/2/2005

December 2nd, 2005 (Permalink)

Don't Name that Fallacy!

Eugene Volokh has posted a weblog entry concerning what he calls "the Reverse Mussolini Fallacy". I missed this the first time around, but according to an earlier entry the fallacy is:

The argument that, because Mussolini (supposedly) made the trains run on time, it's wrong to make the trains run on time. Of course, it's never put quite that starkly (since then it's visibly absurd)… Even a stopped clock is right twice a day.

This was prompted by an example cited in an earlier entry by David Bernstein:

[George] Soros believes that a "supremacist ideology" guides this White House. He hears echoes in its rhetoric of his childhood in occupied Hungary. "When I hear Bush say, 'You're either with us or against us,' it reminds me of the Germans." It conjures up memories, he said, of Nazi slogans on the walls, Der Feind Hort mit ("The enemy is listening"). "My experiences under Nazi and Soviet rule have sensitized me," he said in a soft Hungarian accent.

Volokh criticized this rhetoric as follows:

True, the Nazis and the Soviets spoke out against foreign threats, rallied their people with patriotism, and warned about espionage. They also ate ham and cheese (or in the Soviets' case, would have liked to eat ham and cheese, if they could get it). But this doesn't mean that foreign threats or espionage should be ignored, patriotism disdained, or ham and cheese eschewed. The argument that "The bad guys did it" (or "It conjures up memories [of the bad guys doing it]") is not by itself much of an argument.

This is a good description of the fallacy of Guilt by Association, combined with an equally good explanation of why it is fallacious, based on a not so good example (not so good because it's unclear what Soros is arguing). Now, "guilt by association" isn't the best name for this fallacy, since the phrase is often used in a much broader, vaguer way: judging someone guilty based on bad friends or associates―which is a bad thing to do, but not a logical fallacy.

The "Mussolini Fallacy", according to Volokh, is approving of Mussolini based upon his supposedly making the trains run on time. Again, this isn't a logical fallacy, though one would have to have rather strange priorities to think that way. Presumably, having punctual trains is a good thing, but at the most this shows that Mussolini did one good thing, which is a poor reason to approve of him given the many bad things he did. Guilt by association is the "reverse" of this because, instead of approving of Mussolini because he made the trains punctual, one disapproves of punctual trains because Mussolini was such a bad guy.

However, "reverse Mussolini fallacy" is a poor name for this fallacy. Hitler and the Nazis are much more frequently mentioned in guilt by association arguments than Mussolini and the Fascists―as is evidenced by the Soros example―which is why there is the subfallacy of playing the Hitler Card. Moreover, in order to understand what it is the reverse of, one must know what the "Mussolini fallacy" is. For these reasons, I'll stick with "guilt by association", despite its faults.


Reader Response (12/7):

John Congdon sent the following comments:

You said that "it's unclear what Soros is arguing." I would agree that the relevance of memories of "Der Feind Hort mit" aren't necessarily germane, but the point of Bernstein's paragraph seems pretty clear: Here, at least, Soros is drawing a specific parallel between Bush and the Germans for their (allegedly) shared inclination to bifurcation of political loyalty ("You're either with us or against us"), as well as (perhaps) implying suppression of dissent, a hallmark of totalitarian regimes of whatever political or religious stripe. He is not drawing a parallel between Bush and the Germans for speaking out against foreign threats, rallying with patriotism, or warning about espionage (although this last is, as you point out, unclear), let alone for eating ham and cheese.

Regardless of whether Soros is correct or incorrect in this assessment or guided or misguided by his emotions and memories, Volokh seems to have fallen into what we might call the "reverse-Hitler card." Here, an argument is assumed to be fallacious if the Nazis are merely mentioned, even if (as in this case) a specific comparison of specific behaviors is being made.

An extreme example of this might be:

X says, "Hitler hated Jews, and so does the KKK."
This is playing the Hitler Card.
Therefore, it is wrong to draw a parallel between Hitler hating the Jews and the KKK hating Jews.

In fact, Volokh is committing the classic Straw Man fallacy, in that he criticizes Soros (albeit not completely without reason) for invoking the Germans and the Soviets speaking out against foreign threats, rallying with patriotism, and warning about espionage, when Soros is actually concerned (at least here, and rightly or wrongly) with an absolutist bifurcation into "us" and "them".

If I understand you correctly, you think that Soros was drawing an analogy between Nazi Germany and the contemporary United States, with the conclusion, I presume, that Bush is becoming as bad as Hitler. This is speculation, of course, but it would explain why Soros spent millions of dollars attempting to have Bush defeated for re-election.

This is, indeed, an alternative interpretation which occurred to me, and why I said that it was unclear what Soros was arguing. If this is in fact what Soros meant, then Volokh's guilt by association interpretation is incorrect, and you are right that he attacked a straw man.

However, it's worth pointing out that this alternative argument is equally fallacious, but commits a different fallacy: weak analogy. As long as I recall, some people have always argued that the U.S. is on its way to becoming a Nazi-style dictatorship, and so far they have always been wrong. The trouble with such arguments is that they rely on cherry-picked similarities between the U.S. and Nazi Germany, and ignore the many important disanalogies. Another Hitler has always been right around the corner, and the country, thankfully, has never turned that corner.

There is still another fallacy perhaps lurking in Soros' remarks which I leave as an exercise for the reader.

December 1st, 2005 (Permalink)

Name that Fallacy!

The headline for a press release from the National Retail Federation…read, "Blockbuster Black Friday Weekend Sees Sales Near $28 Billion."

Here's how the NRF's polling company, BIGResearch LLC, arrived at that estimate: The company has gathered an online panel of consumers who answer regular surveys about their buying habits, elections and other matters. The company emailed panelists on the Monday or Tuesday before Thanksgiving to advise them that a survey was coming over the weekend. Then a second email went out late Thanksgiving night, saying the survey was open. It stayed open until late Saturday night. The survey asked consumers a series of questions about their weekend shopping activity and season-long plans. The key question for the group's estimate was, "How much did you spend on holiday shopping?"

…[E]ven though the group's press release was remarkably precise―it claimed the average expenditure was $302.81―the survey didn't ask respondents for a specific dollar amount, but rather to choose one of 16 dollar ranges that best fit their spending. I asked Phil Rist, BIGResearch's vice president of strategy, how the company converted the range figures to hard numbers, but he told me the techniques were proprietary.

Barry Ritholtz, chief market strategist for the New York-based investment bank Maxim Group, called the NRF's findings "false and unsupported assumptions" in an email to investors Monday, and said the survey methodology was ill-equipped to measure total spending. He added, "Investors who rely on it as investment advice are sternly warned they do so at their own risk."

Source: Carl Bialik, "Numbers Guy: Holiday Sales Numbers Don't Add Up", Wall Street Journal, 11/30/2005


Solution to the Prediction Puzzle:

  1. Congratulations, you selected the correct answer! You are in the minority, as most people select the second answer. Perhaps you realize that the probability of a conjunction of events is never greater than, and usually less than, the probability of either conjunct.
  2. Congratulations, you are normal! Like most people, you selected the wrong answer. To see that the answer you selected is wrong, consider that however unlikely it is that Bush will decrease spending on homeland security, it has to be even more unlikely that he will both decrease such spending and open up the Alaskan refuge to drilling. This is true because of a fundamental principle of probability: the likeliness of two events occurring is never higher than that of one of them occurring. However, psychologists have discovered that most people―even those who are knowledgeable about statistics―make judgments of probability that violate this principle.


Previous Month | RSS/XML | Current | Next Month