Try and say things that have a reasonable chance of being true...

Adrian Currie writes...

I’m pretty fond of steampunk: a literary and aesthetic genre, maybe best explained as being noir set in Victorian England, but also science fiction. So, worlds populated with dirigibles, steam-powered home conveniences, tophats coupled with goggles, and so on. A particular favourite of mine is Sydney Padua’s 2D Goggles, a comic set in an alternative universe where Ada Lovelace and Charles Babbage build a difference engine and, um, sort of use it to become vigilantes.

Ada Lovelace and Charles Babbage from Padua's pocket universe...

Ada Lovelace and Charles Babbage from Padua's pocket universe...

Steampunk fiction, of course, isn’t intended as *serious* counterfactual history (although Padua’s approach often involves original historical research). But considering how actual histories could have been different is not uncommon among academic historians and scientists. If Archduke Ferdinand hadn’t been shot, would World War 1 have still occurred? If that big rock hadn’t hit us, would the dinosaurs have gone extinct? Many historians are suspicious of counterfactual histories.

Last week Helen Zhao wrote a lovely piece on something historians call the ‘minimal rewrite rule’, which tries to tell the difference between good and bad counterfactual history. The thought is, if you’re going to do counterfactual history, you should only change history in minimal ways. And a while ago, Derek Turner has written on a similar subject, in part proposing a different(?) rule. And back in 2004 Aveizer Tucker spent a bit of time with historical counterfactuals in his book Our Knowledge of the Past. Well, I want in too! Specifically, I want to pick up on Helen’s discussion of the minimal rewrite rule, tie it to some philosophical discussion of the semantics of counterfactuals, and provide a new way of thinking about minimal rewriting and its justification.

But first: a gripe. On occasion, historians (and philosophers of history) will deny that their work concerns anything more than the actual: history is in the business of detailing the what, whys and wherefores of a particular historical trajectory. I get the impression that this is taken to underwrite an argument that says counterfactual history is dodgy simply because of its modal scope. Real history only concerns the actual world, and to go beyond it into might-haves or could-have-beens is folly. I buy that history is not necessarily dealing in law-like explanation, but that doesn’t save you from the modal. This is because explanation is very often a matter of contrasts: to explain something, you tell me why one thing happened as opposed to another. And to do this, you need to know at least which events made a difference to the actual occurrence, as opposed to the contrasts, um, occurring. The historian’s focus on the fragility of historical trajectories is itself deeply modal. A mere ‘chronology’ gives us an ordering of events, but a true history situates those events in not only the world, but in contrast with other possibilities. Further, many plausible notions of causation are closely related to modal notions—particularly interventionist, difference-making and (duh) counterfactual accounts. And causation and explanation are pretty tight-knit. Insofar as history purports to explain, then, modal notions such as contingency, necessity, and counterfactuals are going to matter. And if that isn't convincing, go read Daniel Nolan's paper.

Enough gripe: let’s contrast two historical counterfactuals. First, Sydney Padua’s. In the actual world, Ada Lovelace died young of cancer, while Babbage never in fact built any of his calculating machines. In Padua’s world Ada survives, Charles builds his machine, and they use it to fight crime. So, here’s our counterfactual:

If Lovelace hadn’t died, and Babbage built his machine, then they would have become a vigilante crime-fighting duo.

Second, here’s a more ‘serious’ historical counterfactual (maybe). Ada Lovelace was Lord Byron’s daughter (yes that Lord Byron). Unsurprisingly, Ada’s mother wanted to minimize Byron’s romantic, poetic, influence, and (a bit more surprisingly) prescribed a heavy dose of mathematics to counterbalance her ‘mad, bad, and dangerous to know’ inheritance. In computer science, Lovelace’s main claim to fame is writing an 1842 paper which explained Babbage’s notion of an analytic engine, and included the world’s first specification of a computer program. Presumably Ada’s mother could have picked something other than mathematics, and it’s not mad to suggest that this early training in mathematics was in part responsible for the path Lovelace’s life took. So, here’s another counterfactual:

If Lovelace had been educated in something other than Mathematics, then she wouldn’t have written her 1842 paper.

(by the way, the underlying ideas of both counterfactuals are captured in this 2D Goggles story…)

Ada Lovelace at age four, teetering on the brink of a life of mathematics or poetry...

Ada Lovelace at age four, teetering on the brink of a life of mathematics or poetry...

Okay, so on the face of it the first counterfactual is not on the historian’s ‘take seriously’ list, while the second should at least fare better. How come? Let’s bring in the minimal rewrite rule. Here’s how Helen describes it:

[The minimal rewrite rule] is the requirement that historians (and paleontologists and evolutionary biologists) ought to take as the counterfactual starting points of their narratives only unactualized possibilities implying minimal changes to the states of affairs before them. In other words, it is the rule that historians etc. ought to, as much as possible, avoid rewriting history up to the start of their counterfactual narratives.

As written, there are two ways of understanding the rule (conversations with Helen really helped clarify this). Distinguish between a counterfactual’s antecedent (the bit after the ‘if’) and its consequent (the bit after the ‘then’). On one version of the rule, we restrict the antecedent. The antecedent of the first counterfactual, where Lovelace survives cancer and Babbage builds the difference engine, involves a complex series of changes compared to the actual world. Whereas the antecedent of the second counterfactual only involves a few small—we might say minimal—changes. Just switch Ada’s mother’s solution to a poetic inheritance. On another version of the rule, I should only take seriously counterfactual histories where the step from the antecedent to the consequent is small. For Lovelace and Babbage to become a crime fighting duo, simply changing things such that Babbage gets some extra funding and Lovelace recovers from cancer is not sufficient (Padua has the whole thing take place in an alternative pocket universe with its own rather fabulous physics). However, it looks like changing a few things about Lovelace’s mother (maybe lead her towards philosophy instead of mathematics?) could be sufficient to connect antecedent to consequent in the second counterfactual. So far as I can tell, this ambiguity doesn’t matter enormously, as the reason that such an outlandish antecedent is called for in the first place is in order to ‘get to’ the outlandish consequent. So I’ll note the ambiguity and move on.

However, what does all this ‘small change’ business mean? And why should we think this is a good rule? Well, if you’re anything like me, during the discussion so far a little voice in your head would have been screaming something like ohgoshthislookstoteslikeLewis’counterfactualsemantics. Well, scream no more, little voice: let’s connect the minimal rewrite rule to counterfactual semantics.

Lewis is interested in understanding the truth conditions of counterfactual statements like the two we considered above. Claims about the actual world, like, for instance, someone made a difference engine out of Lego, have relatively unmysterious conditions for truth (philosophers, hush!). You go look at the world, and see if it contains a Lego-constructed difference-engine. But counterfactuals seem odd: we can’t ground the truth of such claims in actuality (it seems), yet discussions about them seem meaningful – that is, they can be true or false. When paleontologists argue about what would have happened if some mass extinction hadn’t occurred, it doesn’t seem like they’re just talking nonsense. But on what basis? David Lewis reckoned that the truth conditions for counterfactuals aren’t so different from the truth conditions of statements about the actual world. It is just that the truth conditions don’t turn on what happens ‘here’, but rather on what happens in some other world. A different possible world. The basic idea is that some counterfactual is true just in case in the closest world where the antecedent is true, the consequent is also true. So, go to the closest—the most similar—world to our own, where in that world Lovelace is introduced to something other than mathematics, and then see whether she writes her paper on Babbage. In doing this, don’t change the world so much – as opposed to rewriting her whole history, just alter Lovelace’s mother at the right moment so she reaches for philosophy or gardening rather than maths. In finding the right world, you want to hold as many things fixed as possible while making the antecedent true. This ensures a kind of ‘closeness’.

The notion of ‘closeness’ is clearly doing a lot of work, just as ‘minimal’ is doing a lot of work in the rewrite rule – perhaps the same kind of work. Just what closeness amounts to is, um, very hard so let’s leave questions about closeness and minimality aside. Further, let’s not worry about the metaphysics of these possible worlds (in my view, whatever story we end up with regarding counterfactuals is going to at least save the success of Lewis’ semantics, and I’m only relying on those features in this post). The question concerning us here is: what is the relationship between Lewisian counterfactual semantics and the minimal rewrite rule?

Right out the gate, it is crucial to see that the two do different work. Lewis’ account is a semantics of counterfactuals. That is, it tells us under what conditions counterfactuals are true or false. By contrast, the minimal rewrite rule is methodological: it tells historians what kinds of counterfactuals to consider. Assuming Lewis has something right about the semantics, why would a rule about which alternative histories are worth considering be so reminiscent of an abstract philosophical view about the truth conditions of counterfactual statements?

Well, notice that our first counterfactual—the crime-fighting one—is almost certainly false, while our second counterfactual might be true. Consider: is it likely that the closest possible world—that is, the one which is the most similar to ours where Lovelace survives cancer and Babbage builds his thinking-engine— is also a world where they become a crime-fighting duo? I doubt it – a lot has to change. Now, it may be, that in the closest possible world where Lovelace learns philosophy, she doesn’t get into computational mathematics. It is at least conceivable that the antecedent would follow.

If we read the minimal rewrite rule as restricting us to minimal antecedent changes – the first counterfactual needs a lot of change to make the consequent true, while the second doesn’t – then we might justify it in appeal to counterfactual semantics. Roughly, the rule just says: try and say things that have a reasonable chance of being true.

I expect that statement of the rule would need a lot of nuancing for me to accept, for instance it’d need to cohere with my love of scientific speculation. I also think you might like the rule even if you think most counterfactuals turn out to be false. But let’s focus on two kinds of defences which are in the spirit of the rule as understood above, which make for a (IMHO) pretty convincing positive case for the rule.

First, disobeying the minimal rewrite rule, when understood in terms of counterfactual semantics, introduces a lot of ambiguity. We can understand ambiguity using sets of worlds. Instead of considering a particular, unique world, we consider a set of worlds in which some of the same events happen. For instance, there are many possible worlds in which Ada Lovelace’s mother didn’t press her into the service of mathematics. The counterfactual above claims that of that set, the one which is closest to our world doesn’t include Lovelace’s publication. In terms of the first counterfactual, consider the intersection of the worlds where Lovelace survives cancer, Babbage builds the difference engine, and Lovelace and Babbage fight crime. I imagine the intersection is going to be more improbable than the antecedents-are-true worlds and the consequent-is-true worlds. Moreover, there are going to be a lot of ways that the intersection might be realized: so much rewriting is required to get things to happen, that notions of ‘closeness’ become pretty indeterminate.

Now, you might say that ‘if p, then possibly q’ is weaker, logically-speaking, than the counterfactuals above – and so will more often be true (and thus more in the spirit of the rule). But counterfactual histories aren’t about what is merely possible, but what would have happened if some other event hadn’t. If counterfactuals involving a lot of rewrite introduce this kind of ambiguity, there is a reasonable complaint that they’re not even doing the job they’re supposed to do. (perhaps, in reference to my question below, this hints at another job?).

Now, onto my second defence. This one involves a more explicitly epistemic component. I’m going to introduce an ugly—but, I think apt—term: cross-world actualism. ‘Actualism’ is a term I’m taking from the philosophy of geology. An actualist in the 19th Century believed that the best way of explaining the history of the earth is via processes which occur now. The present is the key to the past. (it is important to distinguish actualism from catastrophism and uniformitarianism: both are kinds of actualism, but differ on whether those processes must act gradually at a constant rate, as Joyce has deftly explained ). So, by understanding how current glaciers retract and expand, I can extrapolate those processes into the deep past to understand the presence of strange rocks, valleys, and other geological features.

Cross-world actualism is the modal version of actualism. Here, the actual is the key to the possible. Broadly speaking, I think there are two routes here. I’ll call one the robust route. Here, there are certain regularities which hold in the actual world, which also hold in similar counterfactual worlds. The two operate, as it were, according to the same rules. It’s easy to answer what would happen if some events in the past had been different, as we have access to the ways in which such worlds work, and can thus extrapolate. As we shift further and further from the actual, the chances of those worlds behaving similarly to our world get longer and longer. As such, actualism fails in those cases.

There is another way of being a cross-world actualist, which isn’t about regularities, but rather about epistemic constraint. I’ll call this the fragile route. The closer the counterfactual world is to ours, the more we can use how events have unfolded in the actual to constrain what occurs in the counterfactual world. This is in the spirit of Helen’s discussion last week. She argues that if one strays to far from minimal rewrites the ‘modularity’ of a world breaks down. Modular causal systems have components which have relatively regular, understandable consequences when they are altered. When modularity breaks, changes to components produce chaotic results because those components become increasingly interdependent and messy.

I think that Helen may be right that modularity breaks down on large rewrites, but I also think that the fragile route has another face as well, one best captured in terms of cross-world actualism. Above, I presented a possible example of a minimal rewrite: that Lovelace’s mother could have led her to philosophy instead of mathematics, and this would have led to Lovelace not writing her paper. This counterfactual is constrained by a set of questions relevant to (and grounded in) the actual world. We might ask how available philosophy was at the time, or whether it was really Lovelace’s mother who was responsible for sending her towards mathematics. Or, more seditiously, we might be suspicious regarding the neatness of the story in the first place. Is it really true that Ada Lovelace was led to mathematics to cure her of poetry? All of these questions are about the actual world, but constrain how we deal with the antecedent of the counterfactual. We can say the same about the consequent. Could it be that other events in Lovelace’s life overdetermined her getting into mathematics? In this way, particular facts about the actual world serve to constrain our theorizing about possible facts in close possible worlds.

Moreover, the counterfactual conclusion itself – that Lovelace might not have written her paper – itself leads to further questions relevant to the actual world. One which I’d love to know more about concerns influence. We might ask, for instance, had Lovelace not written her paper, would computer science be different? Certainly, people were reading Lovelace as computer science developed (well, at least Alan Turing discussed her in *that* 1950 paper in Mind), but did her approach make a difference to how the science itself turned out? Such questions involving tracing influence and such developments are the very stuff of history. And this stuff comes out clearly in light of the counterfactual. The counterfactual perspective, when keyed to relevantly close worlds—to minimal rewrites—can have a mutually beneficial, dynamic relationship with the actual perspective. It just leads to better history.

So, we have two complementary defences of the minimal rewrite rule, both in the spirit of the reading I think follows from counterfactual semantics (try and say things that have a reasonable chance of being true). First—and this one more-or-less follows from the semantics—the less minimal the rewrites, the more ambiguous the counterfactual claim. Instead of picking out a unique world, or a relatively small set of the worlds, we pick out a large set of worlds. Given the size and diversity of that set, it would be remarkable if the consequent-is-true worlds overlapped significantly with the antecedent-is-true worlds. In such cases, it is an open question whether there is a unique ‘closest’ world. Second—and this involves an epistemic component—counterfactuals with minimal rewrites are much more amenable to ‘cross-world actualism’, both in its robust form, and in its fragile forms. Further, this latter justification also brings with it interesting perspectives and insights into actual history. Considering what could have happened provides insight and research directions into what in fact happened.

Let’s finish with a bold assertion, a note about metaphysics, and an ongoing question.

Bold assertion: I’m so convinced by these justifications that I’m now inclined to think that it is history without counterfactual history that is truly impoverished.

Note: a fair few philosophers of science appear to be pretty dismayed about the central role analytic metaphysics and philosophy of language plays in philosophy at large. Although I don’t think philosophy should have a centre, and so agree in a sense, I reckon Lewis’ account was pretty helpful for me coming to, and articulating, the views in this post. That’s something, right?

Ongoing question: this has largely been a positive story about why ‘minimal-rewrite’ counterfactual history is kosher. I wonder if a different—but still vindicatory—story might be told about ‘non-minimal rewrite’ counterfactual history. Obviously in fiction such a capacity is often crucial (a similar rule applied to fiction would have denied us 2D Goggles, which I take to be a straight-up knock-down objection to such a rule), but I wonder if there are arguments to suggest that much more outlandish counterfactual histories deserve to be taken seriously?

*many thanks to Helen Zhao for comments on a draft, and Ray Briggs for a useful conversation regarding the ideas herein*