As historians it seems like we’re always cautioned against saying “what if” but the truth is that doing so is a necessary part of history and to pretend otherwise is simply irresponsible.
Last week in the Guardian, Richard Evans issued a scathing attack against the use of “counterfactual history.” He rightly noted that the centenary of World War I has driven some highly speculative fantasies about what might have been had some elements of the war unfolded differently. While Evans’s critiques of these particular applications of counterfactual history might be well grounded, the expansion of his comments to take in the whole of counterfactual reasoning as a historical method is misguided: either betraying a misunderstanding of the nature of counterfactuals as a philosophical concept, or attacking the method as a proxy for attacking the politicized motivations of some of the what-ifs of WW1.
Throughout his essay, Evans equates the asking of “what-if” or the consideration of counterfactuals with completely and rampantly making things up: Calling them “speculations [that] are of course unprovable” and products of “fantasising [that] threatens to overwhelm our perceptions of what really happened in the past, pushing aside our attempts to explain it.” But this begs the question of what it means to call something an historical “explanation” without considering the range of possible alternatives.
Tacit Counterfactuals in Claims of Historical Cause
Almost any historical account consists of more than simply relating “what really happened.” Historians are charged not merely with reporting unfiltered facts about the past (leaving aside the question of biases in even acquiring historical facts.) We’re expected to make estimations of importance. Which facts matter? When we relate an account of an event, we make choices of what to include. These choices reflect decisions about which events count as part of an explanation and which don’t. To say that something matters, or doesn’t, is to make a counterfactual claim.
In my history of the events leading to the Scopes trial, I discuss the 1924 reelection of Tennessee Governor Austin Peay, but I did not mention the 1924 presidential election. Even though both of these were events that took place shortly before the 1925 Scopes trial, and even though both of those elections are things that “really happened” one of them was involved in a chain of events that contributed to the trial. From the point of view, one of those events is part of an explanation and the other is an extraneous fact that also happened.
Any time we make an interpretation about relevance, anytime we make a claim that one fact caused or contributed to a subsequent fact, we’re tacitly making use of counterfactuals. If I’m saying that the reelection of Peay contributed to the Scopes trial, I’m logically making a claim that if Peay had not been reelected, then the events of the Scopes trial would have unfolded differently, if at all. If I’m saying that the presidential election had no bearing on the trial, then I’m logically saying that the Scopes trial would have happened even if John W. Davis (the Democrat) had won in 1924 instead of Calvin Coolidge. Even if they’re not expressed as a what-if, counterfactuals are unavoidable. Logically, every single time that we make a positive causal claim, every time we say “A caused B,” we’re logically also saying that “if not for A, then B wouldn’t happen.” There might be some other things that could also cause B, but if that’s true then we’re not likely to identify A as the cause (or as part of a cause.)
If we believe that history includes an account of cause and effect, then we are tacitly making use of counterfactuals. Evans is right that the popular use of counterfactuals tends to make the causes in history appear to be more singular. The Scopes trial doesn’t happen without Peay’s reelection (without it, no one is proposing an education reform bill that prompts an antievolution bill in protest) but there were plenty of other things that had to occur in addition to Peay’s reelection for the Scopes trial to occur. The fact that we recognize that history is complex and multi-causal doesn’t make counterfactuals disappear, it just makes them more complex. In choosing what facts to include in a narrative—whether it’s social statistics, cultural practices, the presence of certain ideas in newspapers and books, economic data, or the singular decisions of a political leader—those choices imply counterfactuals that legitimates their relevance.
Philosophical Counterfactuals and History
The real problem with Evans’s argument, one that it perhaps shares with some some overindulgent what-if history games, is that it treats counterfactuals as any consideration beyond “what really happened.” By relegating them all to the outrageous, fantastical, and speculative; this portrayal loses sight of what counterfactual conditionals actually are. Consider the following conditional statement:
If Coolidge is elected president in 1924, then there is a Scopes evolution trial in 1925.
We can look to the historical facts to test this and discover that the statement is true in the real world of what actually happened. Coolidge was elected, and a Scopes trial took place. But does it follow from this that the election of Coolidge qualifies as a cause of the Scopes trial, or is the truth of the two events simply a coincidence?
For estimations of cause and effect, we need to consider counterfactual conditionals.
If Coolidge had not been elected president in 1924, then there would not have been a Scopes trial in 1925.
The counterfactual condition invites us to try to envision a logically coherent world in which the first statement is false. There could be many scenarios in which Coolidge is not elected in 1924.
- The Democrat Davis wins.
- The third-party Progressive candidate Robert M. La Follette wins.
- Coolidge dies and is replaced on the ballot by another Republican, who wins.
- Canada conquers and invades the United States and the presidential election is cancelled.
- Space aliens invade Tennessee and destroy the human population of the State. La Follette wins the election.
- Space aliens invade the earth and the human race is destroyed. Consequently there is no election.
All of these are scenarios in which Coolidge is not elected president in 1924 (you might come up with others.) It seems pretty obvious that in some of these (such as #5), the Scopes trial also would not occur. It also seems fairly likely that in some of these (such as #2), the trial would continue unimpeded.
One of the most influential ways of thinking about evaluating the truth of counterfactual conditionals considers the idea of ‘close’ or ‘near’ alternatives. Probably the best known explication of this is David Lewis’s discussion of “possible worlds.” A counterfactual conditional statement is considered true if the statement is true for the closest worlds in which the hypothetical is true. In other words, in the closest possible worlds in which Coolidge is not elected, is it the case that the Scopes trial doesn’t happen?
(There’s debate among philosophers over the details of this expression: what Lewis has meant by closeness, and whether comparative closeness can be evaluated—the details of which I chose to leave aside for now. There’s also debate over using the idea of “possible worlds” as the way to think about interpreting counterfactual conditionals. But for the question of applying this to historical method, I’m going to use it for the time being.) The important concern for history is one that’s generally agreed upon by philosophers: the consideration of counterfactuals is an inherently conservative process, in which one tries to consider the least speculative fanciful worlds that are closest to the actual one. Unlike what Evans describes, it’s not counterfactual history to simply make up anything that doesn’t happen.
Counterfactuals are about considering the worlds that are nearest. In historical counterfactuals, this is a role for historical judgement. Based on available evidence, I’d suggest that the possible worlds in which Davis is elected are nearer ones than the possible worlds in which aliens destroy the human race. (If, after alien contact is made and alien military archives become available for research, I might have to revise this estimation.) If Davis had become president is there any reason to suggest that anything local in Tennessee would have changed? Historical evidence can be used to argue the importance of this. If it’s the case that in the nearest worlds, (Davis elected president, but with the minimal changes necessary to effect that change) that the Scopes trial does take place, then the counterfactual conditional is false.
The conditional statement
If Coolidge is elected president in 1924, then there is a Scopes evolution trial in 1925.
is true, but
If Coolidge were not been elected president in 1924, then there would not be a Scopes trial in 1925.
is false.
It’s the falsity of the counterfactual conditional that allows us to dismiss the presidential election as part of a cause or explanation of the Scopes trial.
For the sake of completeness, let me say that—to my best interpretation of the available evidence— the closest worlds in which Peay wasn’t reelected are those in which the Republican candidate won (not those in which Tennessee is ruled by a Canadian military authority or destroyed by aliens). In those nearby worlds, a Republican Governor of Tennessee would not have proposed a General Education bill that would have prompted the Antievolution bill to be proposed in backlash. In effect this is a claim that these worlds are in fact closer than ones in which Peay is reelected and does not propose a General Education Bill.
Counterfactuals are not simply license to make up absurd premises and and imagine a fantasy world. And they’re not invitations to find the most obscure antecedent event and claim everything deterministically follows the way a proverbial “want of a nail” causes kingdoms to be won or lost. Counterfactuals are used in history to explain why things happen; they’re used conservatively to assess the impact of one event on subsequent events. Evans picks out some particularly outlandish examples, where people chose such extreme hypotheticals that it’s almost impossible to make a real case for what possible worlds are nearer to another. Evans blithely suggests, that if you’re going to fantasize about an alternative world in which Archduke Ferdinand wasn’t assassinated, then he can equally imagine one where “Franz Ferdinand might have fallen victim to another assassin’s bullet, or died in a hunting accident.” In the egalitarianism of everyone-can-make-stuff-up we’re either left with an infinitude of incomprehensible and bizarre worlds, or we must restrain ourself only to the actual world. Is Evans’s hunting accident a legitimately nearby alternative to consider, or is this an attempt to depict all counterfactuals as equally baseless speculations by using a convoluted one to simply preserve his argument?
Evans is certainly right that some people use explicit counterfactuals to focus on singular events and individuals—and that is problematic. But counterfactuals don’t have to result in a “kings-and-battles” view of the past. The presupposition is that it’s easiest to think of nearby worlds in which a single person’s decision is changed, rather than one in which social conditions (which influence people’s choices) are different, and it’s outlandish to thing that individual decisions aren’t shaped by social conditions. It might very well be that a world in which social conditions haven’t changed but in which leaders make slightly different decisions turns out to be fraught with inconsistencies. (It might be possible to imagine a world in which Governor Peay did not push a General Education Bill, but that would perhaps be inconsistent with the idea that Tennessee voters would have reelected him, so it’s not a near possible world at all.)
I almost agree with the statement that: “the problem with counterfactuals is that they almost always treat individual human actors – generals or politicians, in the main – as completely unfettered by these larger forces.” Evans is probably right that many historians err when considering the differences between a possible world in which Ferdinand is assassinated and one where he’s not. But that’s a constraint on what it means for a world to be possible (not internally contradictory), not an argument against possible worlds. It might be harder to speculate what might have been the political reality of a nearby world in which Germany’s unemployment rate was 2% higher. But that’s a rationale for having historians recognize the use of counterfactuals, so that they are clearer in such argumentation. It’s not a case for us to teach students that anytime they encounter a counterfactual phrasing they should immediately dismiss the work. The problem Evans identifies is nota problem with counterfactuals, it’s a problem with historians who use them badly. Unfortunately the blanket condemnation of counterfactuals doesn’t actually promote clearer historical thinking.
This matters particularly because this consideration of nearby counterfactuals is not only relevant to cause and effect in historical argumentation, it’s also relevant to cause and effect explanations in science, and Evans argument seems to augment the pseudoscientific presumption that all alternative explanations deserve equal weight and consideration. Given some of his statements about the history of science, that might not be unintentional.
Counterfactuals in History, Science, and the History of Science
Scientific theories that make statements about causes and effects have a similar logic. Perhaps the most clear case of this is in the scientific analysis of the history of nature, such as the evolutionary life history of Earth. In a complexly interacting ecological system, with selection occurring at individual levels, group levels, and even levels interior to single organisms, it’s very difficult to say any single factor caused a given evolutionary effect. But comparative data and scientific models help refine the idea of what factors have a robust impact on a specific development. Perhaps the most famous version of this argument is Stephen Jay Gould’s analogy of “replaying life’s tape.” Gould suggests that evolutionary history is so contingent that even the slightest alteration can cause radical differences to the unfolding of life. One argument against this view is the case made for evolutionary convergence (such as that put forward by Simon Conway Morris) which suggests that certain evolutionary outcomes are strongly likely because of long-scale properties about Earth and life, not the concatenation of highly divergent variables. Even if life’s tape replays differently at the individual level, general trends like the development of optical vision, the development of higher intelligence, the development of interior skeletal structures are—not predestined—but so strongly favored that they occur even in different evolutionary lineages. Conway Morris’s examples of counterfactuals in political history may suffer from the kinds of defects that Evans castigates, but he makes the case that this is a useful part of reasoning in the sciences.
Those who argue that naturally occurring evolution cannot give a causal account of the diversity and complexity of life (this seems to be a fundamental tenet of modern Intelligent Design) seem to make use of anti-counterfactualism like Evans’s in their arguments. In describing evolution as purely random chance, they reduce all alternatives to equally random likelihood or nonlikelihood. That presumption begs the question of whether any evolutionary explanation can possibly be robust enough to be an explanation of cause.
The concept of “irreducible complexity” that Michael Behe put forward in Darwin’s Black Box in 1996 could be interpreted this way. Behe points to structures that require, for example, a number of genes to be produced and claims that these could not have evolved because there would be no benefit to a structure (or possibly harm) at intermediate stages. This amounts to a claim that the nearby worlds in which a cell has only 39 genes of a 40-gene structure is not a possible world, and that the nearest worlds are therefore ones in which an intelligence intervenes in creation. In his updated version of the book, Behe derides the work evolutionary biologists had done to show the possibility of such worlds: “Call them wishful speculations or call them plausible scenarios—both just mean a lack or real answers.” In conflating wishful speculation and plausible scenario and juxtaposing them both against a supposed “what really happened,” Behe’s engaged in the same kind of distortion of counterfactuals that Evans has done by ignoring nearness, ignoring conservatism, and ignoring the requirement that accounts of cause require consideration of more than the instances that actually happen.
[Note: I may want to further develop this line of reasoning about the tacit use of claims about counterfactuals in arguments against naturalistic sufficiency. —comments especially appreciated]
The type of anti-counterfactualism that Evans advocates is consistent with that put forward by proponents of intelligent design (like Behe.) So perhaps it is no surprise that Evans has also supported the claims ID advocates make about the history of science. One of their biggest historical claims about evolution—used to show the purported immorality of Darwinism—is the Darwin-to-Hitler myth promulgated by Richard Weikart. Weikart, a senior fellow of the Discovery Institute, claims that Darwin and Darwin’s ideas were directly responsible for the rise of Hitler and Nazism. This is a claim that’s been addressed by Peter Bowler’s counterfactual argument—that a coherent and nearby alternative world without Darwin would still have seen the rise of the Nazi regime, its racial and moral commitments, and attempts to justify them scientifically.
This is not the place to fully address the historical debate over the relationship between Darwin’s ideas and Naziism, but I do want to observe that Evans is frequently cited in Weikart’s book. In fact, just after a lengthy citation from Evans, Weikart writes in the last paragraph of his book:
without Darwinism, especially in its social Darwinist and eugenics permutations, neither Hitler nor his Nazi followers would have had the necessary scientific underpinnings to convince themselves and their collaborators that one of the world’s greatest atrocities was morally praiseworthy.
Weikart uses Evans to justify a counterfactual claim. The conclusion of the entire book is this counterfactual claim. If there had not been social Darwinism, then Nazis wouldn’t have claimed scientific justification for the Holocaust. How did Evans respond to this apparently illicit use of counterfactual argument? By giving praise on the back cover of Weikart’s book.