Thread by @Yascha_Mounk

My tweet yesterday may have sounded like a generic objection to a "left bias" in academia. What I actually hoped to express was a set of more specific concerns with fashionable forms of history.

So let me explain my thinking in a bit more detail.

[Thread]

https://twitter.com/Yascha_Mounk/status/1131927845641437186

As I've written in the past, many other forms of inquiry have deep problems of their own.

A focus on high-n stats, for example, risks defining some political questions out of existence, and may have contributed to our failure to predict the rise of Trump.
chronicle.com/article/How-Po…

But to their credit, statistical studies do usually grapple with two big challenges: confirmation bias and case selection.

One way of doing this is to spell out a clear null hypothesis and ensure you're not "selecting cases on the dependent variable."

That's why I was so keen to construct a comprehensive database of populist governments. Are populists really more dangerous to democracies, or is that an artefact of a few cases that are particularly salient to me?

(Sadly, the answer is: they are.)

theatlantic.com/ideas/archive/…

A lot of history does not have the same inbuilt checks. And as it has become more proudly political, one mode of operation has become common:

1) I believe A.

2) I'll focus on aspects of history that imply A.

Most of this scholarship is painstaking and historically correct in the way that Naomi Wolf's isn't.

But there are three primary concerns I have that apply to both:

- Case Selection
- Genetic Fallacy
- Politics

Case Selection:

Let's say you're an historian who thinks authoritarianism is good for growth. What do you study? Singapore!

Go write an accurate, well-documented history about Singapore's impressive economic performance under the semi-authoritarian leadership of Lee Kuan Yew.

Would this convince me that authoritarianism is good for growth? No, because plenty of cases show the contrary. On the whole, the evidence suggests dictatorships are not good for growth.

But this requires us to look at cases with countervailing implications with equal attention.

Quantitative social scientists--for all of their flaws (see above!)--do this as a matter of course: A lot of their work consists in constructing comprehensive databases, and being very aware of what their databases might exclude.

Do historians do something analogous to a sufficient degree?

I have serious doubts. Honestly ask yourself this question:

When did you last see a doctoral student who was working on a topic whose implications seemed to run counter to the core political beliefs of their advisor?

(To be clear, I'm *not* saying that high-n statistics is the only way around the problem. But I do think that historians need to take the problem of case selection and confirmation bias much more seriously than they do.)

Genetic Fallacy:

A lot of contemporary historical scholarship implicitly or explicitly takes the form of the genetic fallacy. It is, by now, a well-honed recipe:

i) I dislike X.

ii) I'll show that some variant of X has bad historical connotations or has led to unjust outcomes.

This can be valuable in the way Raymond Geuss has argued: While it does not show that X is in fact bad, it does show that X is potentially dangerous.

We need to check whether the bad effects of X are still happening and what the bad effects of abandoning X might be.

But this is precisely the point at which historians usually bow out:

"Empirical studies on X's systematic effects today or normative arguments about whether it is good are not part of my job description!"

Fair. Academia entails the division of labor. We need different disciplines.

But this does rather elide the fact that the very reason why people tend to study X, and why work on particular Xs attracts so much attention, is because of the half-spoken implication that X is bad.

Politics:

I have a BA and an MPhil in intellectual history from Cambridge, and some of my PhD work at Harvard was also in the history of political thought.

(Hold on, there's a pay-off to this obnoxious credentialing set-up in the next tweet.)

The assumption behind everything I ever learned in history was that past communities of thinkers were deeply shaped by their political assumptions.

But somehow implying that the same is likely true today is supposed to make me a fellow-traveler of the anti-intellectual right?

Anybody who has set foot in a history department in the past years must know that there is robust disagreement about plenty of important things, but also robust agreement about plenty of other important things.

So we'd be crazy not to interrogate that consensus, and the way it shapes our work, self-critically.

(I say "we" because the same is surely also true in other departments.)

In one exchange I had yesterday, someone basically said:

Oh, but we're completely aware of that! Any trained historian has read a lot about what the Whiggish interpretation of history got wrong!

But in practice, the critique of Whig history has mostly led to a determination to critique rather than praise, the present.

In my mind, though, it's a much deeper warning about confirmation bias and the inevitable blindspots of all serious intellectual communities.

I'm sure that there's plenty to pick apart in my methodological concerns (and, yes, that I have unexamined assumptions of my own).

But if my critique makes you think that I'm your political enemy, then you are proving the very point you are pretending to contest.

Finally, here's why I haven't named particular scholars:

Twitter isn't the right medium to pick an individual's work apart. And the problem is systematic, affecting the collective wisdom of subfields, rather than an individual one, proving any particular book wrong.

But clearly my critique hit a nerve, and is in need of elaboration. So I'll try to write a bigger essay on the pitfalls of history, and its current role in public discourse, sometime in the year or so.

[End.]