Tuesday, June 21, 2011

Caplan Contra Krugman: A Very Clever Post

Paul Krugman has recently been claiming that people on the left understand the views of those on the right much better than people on the right understand the views of those on the left. As Bryan Caplan argues in his response, this partly hinges on the fact that Krugman is comparing left wing academics to right wing polemicists. If instead you compare how well libertarian economists can reproduce Keynesian arguments with how well Keynesian economists can reproduce libertarian arguments, the conclusion might well reverse.

The clever thing about Bryan's response is that he proposes an objective test of the question, a sort of ideological Turing test. 
Put me and five random liberal social science Ph.D.s in a chat room.  Let liberal readers ask questions for an hour, then vote on who isn't really a liberal.  Then put Krugman and five random libertarian social science Ph.D.s in a chat room.  Let libertarian readers ask questions for an hour, then vote on who isn't really a libertarian.  Simple as that.
If provided with sufficient funding, redo the experiment using economists, using political philosophers, replacing libertarians (or liberals) with conservatives. Generate some  actual empirical evidence of who understands whose arguments well enough to defend them, which is arguably important, even essential, evidence of understanding them well enough to be justified in rejecting them.


Anonymous said...

At 5:30 a.m. pst on June 21 the page could not be opened. Reason: server could not be found for the link to Caplan.

dWj said...

I remember hearing a conservative speaker at a University about ten years ago (before his speech, chatting with a smaller, friendlier group) remark that conservatives have an advantage over liberals in debates at colleges because they have prepared responses to the liberal arguments, while the liberals have never heard conservative arguments.

David Friedman said...

Re dWj:

That's been my experience. I suspect one of the reasons my father had such a high reputation as a debater was that he was routinely involved in arguments he had had ten or twenty times--with people each of whom was on his first round.

Of course, my experience is more as a libertarian than a conservative, but there's enough overlap to make me suspect that your point applies to both.

Anonymous said...

Actually Krugman talks about academic economists - Chicago school. And he is right - the typical response from the other side is not "this is wrong for the following logical reasons supported by the following numbers" but "we dont teach this stuff to our students".

David Friedman said...

"the typical response from the other side is"

The response you see them making or the response Krugman claims they make? Do you, for instance, read Becker's comments on the Becker/Posner blog? Did you read Anna Schwartz's old response to Krugman's comments on her work with my father?

Anonymous said...

David: Krugman's statement is not about monetarism, but about the contemporary Chicago school that he says abandoned monetarism.
If you mean Anna Swartz's response (http://tinyurl.com/4xzj665) to Krugman's "Who was Milton Friedman" (http://tinyurl.com/4xh2zxo), Krugman is saying:
Keynesian theory initially prevailed because it did a far better job than classical orthodoxy of making sense of the world around us, and Friedman’s critique of Keynes became so influential largely because he correctly identified Keynesianism’s weak points.
So he clearly is not accusing Friedman of not understanding Keynes :)

Francis said...

Caplan's Turing test would merely determine if Caplan knows liberal arguments better than Krugman knows conservative (or libertarian) arguments. It would not test whether in general, one side understand the other side's arguments better or vice versa.

Anonymous said...

Why does this post make me think of Mike Huben?

Simon said...

In addition to measuring people's behavior, the test could also improve it. When the test exposes a poor grasp of an opponent's arguments, people may get inspired to learn more about the other point of view.

Maybe this way, we would be spared many cheap straw man arguments.

Ricardo Cruz said...

Francis: of course, and Caplan says as much. But merely being able to articulate your opponent's arguments as well as she does, already gives you some insight into how well you understand them, and how well you've thought things through.

Josiah Neeley said...

Actually Krugman talks about academic economists - Chicago school.

Krugman's comment is about liberals and conservatives generally. That's why his example of what conservatives think liberals believe is from a George Will column.

Krugman has previously admitted that he doesn't read conservatives, and it shows.

Josiah Neeley said...

Btw, I would be interested to know which Chicago school economist have responded to a Keynesian argument by saying "we dont teach this stuff to our students."

Power Child said...

I was inspired by this to attempt to make a Turing test of my own on an online politics forum. But soon I realized there's a flaw in Caplan's proposed test that would need to be ironed out or planned around first:

In the original test between a human and a computer, the judge is always a human, so he simply must listen for the human nuances he is familiar with. In the test Caplan proposes, the judges may be liberal, libertarian, or conservative--or something else--and their decision is based on the strength of an argument, not the extent to which it is typical of the side making it.

For example, suppose that a conservative is trying to pretend to be a liberal, but cannot help himself in applying conservative logic to his argument. The question is "Why do you support welfare?"

The actual liberal replies "Because it is simply not right that we live in a society where some people are rich and some are poor."

The conservative pretending to be liberal replies "Because the cost to our society of economic inequality is greater than the cost of forcible economic redistribution."

The judge, who happens to be a liberal, finds the conservative's argument more persuasive, and names him the winner.

But the conservative has not made a typical "liberal" answer. The contest has revealed who is the stronger thinker, not who understands the other side's argument better.

Josiah Neeley said...


I think you misunderstand the proposal. The judges in Caplan's test aren't picking the person who made what they believe to be the most persuasive arguments and declaring him the winner. They are supposed to guess of each participant whether he is really a liberal or a conservative. If a conservative is one of the participants, his goal is not to come up with conservative arguments for liberal positions, but to mimic as best he can what a liberal would say about a given issue. So if a liberal would say that he supports welfare "because it is simply not right that we live in a society where some people are rich and some are poor," then the conservative would want to respond to the question by saying "I favor welfare because it is simply not right that we live in a society where some people are rich and some are poor."

Caplan's prediction (which I share) is that conservatives would do a better job of giving liberal answers to questions than liberals would of giving conservative answers, and the reason for this is that conservatives are more familiar with the liberal position than liberals are with the conservative one.

Power Child said...


I share your and Bryan's predictions as well.

Here's why I got confused. Caplan: "...the ability to pass ideological Turing tests - to state opposing views as clearly and persuasively as their proponents - is a genuine symptom of objectivity and wisdom."

If the conservative defense of welfare is more clear and persuasive than the liberal one, I would think the conservative would want to stick to a conservative defense. But you're right, if he is simply trying to convince the judge that he is a liberal, then he would want to make the typical liberal defense.

I did end up creating my forum thread. In it, I asked 3 yes-or-no political questions and directed repliers to give the opposite answer from the one they believe and then defend it as best they can. I then invite people to browse the thread and see which answers they find most persuasive. In this case, I had to design it that way since many people on the forum are already familiar with each other's political views and there is no anonymous posting. However, I think the results will still be interesting.

Alexandra Thorn said...

There are a lot of glaring problems with Caplan's idea, many of which were mentioned in comments on his post.

One commenter simply links to the wikipedia page "Shibboleth". This is the most obvious problem: people who share an ideology in some sense share a sub-culture, and they are going to have vocabulary differences and other attributes that help them sort in-group from out-group independent of content of argument.

Another big problem is that there's a difference between a good argument and a typical argument. Someone who has truly thought about the issues broadly is likely to be able to present numerous arguments for opposing opinions, including opposing opinions that are not held by the judges in question. That thinker wouldn't do as well on this type of a test as someone who has just focused their efforts on studying the target group in order to attack their views. Their broad thinking and failure to stick to the "talking points" would reveal them as an outsider.

It's an interesting idea, but I don't see how it can be implemented in a way that would accomplish its stated purpose.

Fair disclosure: My political views are decidedly left-of-center. I also don't think that Krugman would do a very good job of correctly modeling the arguments of a conservative *or* libertarian.

David Friedman said...

Re Alexandra's points:

I think I agree that not all well informed left wingers could do a good job of pretending to be right wingers or vice versa. But it does seem to me that the ability to do that is relevant evidence of how well they understand the arguments.

If you have really being paying attention, you ought to know what the most convincing arguments are that intelligent people on the other side are likely to offer, and be able to distinguish them from what you think the best arguments are for their positions.

Of course, it's tempting to tell yourself "I understand the issues much better than they do, so I don't have to pay attention to the bad arguments they make or how they make them, so of course I can't pretend to be one of them." But at that point your basis for believing that you understand both sides is pretty dubious, and you shouldn't expect those who disagree with you to take the claim that you do very seriously.

Alexandra Thorn said...

"If you have really being paying attention, you ought to know what the most convincing arguments are that intelligent people on the other side are likely to offer, and be able to distinguish them from what you think the best arguments are for their positions."

I agree that the ability to correctly model the other side's arguments is partially diagnostic of whether you fully understand what they are saying. But it's a diagnostic that could have false positives or false negatives in either direction. False positives (i.e. appearing to understand when I don't) could arise from pattern matching rhetorical styles without introspection. False negatives could arise from using jargon favored by your own political affiliation, which is bound to have a negative effect on your ability to be heard by the people you're trying to model.

There's also the question of who gets to decide what a "convincing argument" is or who the "intelligent people" are. George W. Bush's public persona (I've heard he's different in private) doesn't tend to meet either of those criteria in the eyes of most of the people I know. But when we look at it empirically, the fact remains that he convinced nearly half of the country to vote for him. Twice.