History, the Final Frontier

Fascinatingly, social scientists seemed to be trained in a way that makes them unable to recognize that there even is such a discipline as history. When they say to you, "Of course, history exists," what they mean is that events did happen before today. They have no clue that history is a distinct discipline, with its own unique methods and standards of evidence.

I just received a note from a very bright social scientist that exhibits this befuddlement on the topic of history very nicely. I quote it, with permission:

"I assume the following two points

"1. 'Facts' are theory laden."

Right away, he has gone completely off the tracks. He believes that there are historical facts, around which one devises a theory. What counts as a fact depends somewhat on one's theory, but these facts, while being "theory laden," are separble from one's theory, and could, for instance, confirm or refute it.

But that is all nonsense. The historian's theories are theories about what the facts were. As Collingwood put it, the "fact" that after 200 AD (as I recall -- I may have misremembered the date here) the Roman legions were mostly recruited from outside Italy is not a piece of data with which the historian starts his inquiry, and around which he spins a "theory"; rather, it is the conclusion of an historical inquiry. The historian's theory is that "after 200 AD, the legions were chiefly recruited from outside Italy." His facts are not "theory laden." His theories are propositions about what the facts were.

"2. The synthetic a priori either doesn't exist or is EXTREMELY limited in what it can tell us"

But, of course, historians do not rely on the "synthetic a priori" to do their work. They rely on what Mises called verstehen.

"Given that, how do you do history without rigorous empirical testing?"

Well, of course, historians do "rigorous empirical testing." In fact, it is hard to imagine a discipline more empirical or more rigorous. (Certainly economics, a field in which every practitioner is able to hold tight to her theory come what may, is far less rigorous than history, a field in which questions are regularly and routinely settled to the satisfaction of the entire profession. Two examples: 1) Every single historian I have encountered on the topic agrees that, contrary to popular belief, the educated people of the Middle Ages knew that the earth was round. And they all agree on this, whether they are Christian, atheist, Muslim, or Taoist; while 2) I don't know of a single economist who rejected her previously held theory of the business cycle based on the evidence of the recent downturn. Every single economist I follow contends that the exact same evidence confirms her own theory! So which profession is more "rigorous"?!) This historical testing consists in assembling every bit of evidence available -- military records, town records, soldiers' diaries, inscriptions on monumental stones, the finds of archaeological digs -- and determining what conclusions are consistent with that evidence. But what my interlocutor desires is that historians should be doing regression analysis, or something of the sort, i.e.: "amongst all empires in their late stages of decay, what percentage of troops were, on average, recruited from outside the core of the empire?" What he wants is for history to abandon its own, extremely rigorous empirical methods, and adopt those of some theoretical social science.

Let us note a couple of things about the method he endorses:

1) To follow it, one must have already done one's historical work: compiled a list of empires, determined what was their core region, and found out from where they recruited their troops. One cannot possibly begin to do statistical analysis without first understanding specific situations, since without understanding specifics, one cannot possibly formulate classes into which one can place events, or know which events should go into which classes!

2) Furthermore, this recommended method, whatever it may yield of interest, is absolutely useless for answering historical questions. As Collingwood noted, the job of the historian is much like that of the detective. The detective's job is not to formulate a "theory of the causes of crime," but to find out who committed this particular crime. What my interlocutor believes is that the detective ought to undertake a regression analysis as to what factors are significantly correlated with crimes of the type he is investigating, and if it turns out that being poor, black, male, and between 15 and 25 are significantly correlated, then the detective ought to arrest all those who fit the criteria, and the prosecutor should dole ought a fraction of the sentence amongst the lot of them. Anything else would exhibit unscientific prejudice! That this is a nonsensical way of proceeding can be seen merely by describing it.

Of course, the detective may come into the case with "ideological priors" -- for instance, he may be inclined to arrest a poor, black teenager, based on a study like the one above. But the cure for that is simply more careful attention to the actual evidence: if the only fingerprints at the crime scene are those of an elderly, white widow, and the murder weapon is in her apartment, then the detective must follow the evidence and arrest her, whatever his "priors" were. It should be obvious that more, or more rigorous, statistical studies would be of no help whatsoever in overcoming those priors and actually figuring out who committed this particular crime.