We Argue in Metaphors

and we disagree over whether they fit

Hindu mystic: Life is like a fountain

Skeptic: Life is like a fountain?

Hindu mystic (sheepishly): Well. . .Isn’t it?

My parents loved this joke when they first heard it, about 60 years ago. For me, it illustrates that we speak in metaphors all the time, so that what we mean is less than perfectly clear.

“Bitcoin is like a chain letter.” Well. . . Isn’t it?

Heather Cox Richardson writes,

Trump’s lies are not errors. They are part of a well-documented strategy to overturn democracy by using modern media to create a false political world. Voters begin to base their political decisions on that fake image, rather than on reality, and are manipulated into giving up control of their government to an authoritarian. 

…In his autobiography Mein Kampf, or “My Struggle,” Adolf Hitler wrote that people were more likely to believe a giant lie than a little one because they were willing to tell small lies in their own lives but “would be ashamed to resort to large-scale falsehoods.” Since they could not conceive of telling “colossal untruths…they would not believe that others could have the impudence to distort the truth so infamously.”

A woman recommended Richardson’s essay to me, because it reinforced her opinion that Trump is like Hitler. I am not persuaded.

All models are false. Some are useful.

“Trump is Hitler” is a metaphor¹. Metaphors are neither true nor false. They either provide a good fit or they don’t, and our opinions on that may differ. The cliche goes, “All models are false. Some are useful.”

In physics, the metaphors of Newtonian mechanics are useful in many contexts. In other contexts, they break down. The metaphor of “light behaves like a wave” can be useful, but so can the metaphor “light behaves like a particle.” The metaphor that “electrons orbit around the nucleus” is useful but can be misleading.

Economists use mathematics to create a sense of certainty about their theories. But back in 1966 Martin Bronfenbrenner exposed the problem with this. He argued that any mathematical model requires an “applicability theorem” that connects the symbols in the model to some real world phenomena.²

For example, the “law of comparative advantage” states that free trade benefits everyone. It is a metaphor that applies reliably when there are exactly two goods exchanged between exactly two producers. The real world is more complex, and the metaphor does not strictly apply. How well it applies is debatable. I happen to believe that the metaphor is still very useful, in that specialization and trade are the central phenomena in economic activity and prosperity.

Metaphors and Arguments

Given that we think in terms of metaphors, and metaphors are neither strictly true nor strictly false, how can we settle arguments?

I think that the best approach is to try to agree on the implications of a metaphor. With the woman who claimed that Trump is Hitler, I proposed that we could bet on the implications of applying that metaphor.

• If Trump wins, will Congress pass an “enabling act” abdicating its Constitutional role and making him a dictator?

• Will Trump arrange to have his rivals killed?

• Will Trump deprive an ethnic group of its rights?

The woman demurred. For her, it is sufficient that Trump cannot accept that he lost in 2020 (just as Hitler could not accept that Germany lost WWI) and that he acts as if he can lie with impunity. For her, those characteristics show that the Hitler metaphor applies.

The Billiard Player Metaphor

In economics, Milton Friedman famously articulated a metaphor to defend the use of assumptions. Suppose we are observing an expert billiard player, and we assume that he uses the laws of physics to aim the cue ball. That assumption is not true. But it may very well be useful in predicting how he will strike the cue ball.

Friedman said that one can think of the assumptions that economists use as metaphors. We treat the consumer as if he were using fancy calculus to solve the problem of what to buy. He argued that if our predictions about market behavior ended up correct, then this metaphor of the calculating consumer was useful.

The Caplan Test

If you and I disagree about the applicability of a metaphor, but we can agree on its implications, then we can formulate a bet based on those implications. Bryan Caplan advocates for betting along these lines as a way to test the sincerity of a belief.

The problem is that we rarely agree on the implications of a metaphor. For example, consider the metaphor “inflation is too much money chasing too few goods.” What are the implications of applying this metaphor? Obviously, we do not actually see dollar bills running through shopping centers, chasing stuff.

One implication might be that we can observe changes in the rate of inflation following similar changes in the money supply. This creates a challenge of measuring inflation and also measuring the money supply. When we do this, the connection is often disappointingly loose.

Hard-core monetarists annoy me by playing games with how they look at the relationship between money and inflation. They differ in their preferred definition of money. They experiment with different structures for the lag³ between a change in the money supply and a change in inflation.

Scott Sumner uses nominal GDP as his preferred measure of “money.” To me, that seems like cheating.⁴ It is using the dependent variable as a measure of the independent variable.

I believe that “too much money chasing too few goods” is a useful metaphor. It’s just that the economy is complex and we have many methods of paying for goods and services. We cannot be certain that a particular measure of the money supply will enable us to reliably predict short-term fluctuations in the rate of inflation.

Doesn’t it imply that…?

I have seen many people get annoyed by the metaphor “Israel is an Apartheid State,” most recently articulated by Ta-Nehisi Coates. I have seen a lot of ad hominem attacks on Coates, and relatively few attempts to evaluate how well the metaphor applies.

Supporters of Israel could ask about the metaphor:

• Doesn’t it imply that Arab Israelis could never run for office? Yet some of them serve in parliament, and a few years ago there were some Arab Israelis in the cabinet.

• Doesn’t it imply that Arab Israelis would be legally forbidden from living among Jewish Israelis? That is not the case.

• Doesn’t it imply that a Jewish minority would rule over an Arab majority?

Critics of Israel could point to the West Bank, where the metaphor seems more applicable.

Legitimate Differences of Opinion

I think that we tend to over-estimate the legitimacy of our own beliefs and under-estimate the legitimacy of the beliefs of those who disagree. Yes, we can point to ways in which our opponents seem to misread the facts. But most of the time the basis of the disagreement is over whether or not a particular metaphor applies.

We can try to get an opponent to agree to a Caplan test concerning the implication of a metaphor. But that may not be sufficient to resolve a disagreement. In that case, we have to admit that the difference of opinion is going to remain unsettled.

substacks referenced above: @

October 6, 2024

1. In English class back in the day, I was taught that “Trump is like Hitler” is a simile, because it uses “like” or “as.” It’s been decades since I’ve heard anyone use the word simile, as distinct from metaphor. “Trump is like Hitler” just gets called a metaphor.

2. Bronfenbrenner’s essay appeared in The Structure of Economic Science, edited by Sherman Krupp.

3. Milton Friedman famously claimed that there were “long and variable lags” between changes in growth of the money supply and changes in inflation. Then came the “rational expectations revolution” of the 1970s, in which it was argued that anticipations of future money growth would affect inflation. Monetarist critic Franco Modigliani sneered that this made the relationship between money and inflation one of “long and variable leads.”

4. Actually, I think he would prefer to use market expectations of future growth in nominal GDP. This is a somewhat less circular but still not satisfying to me.