In the excellent and rich paper, “So close no matter how far: counterfactuals in history of science and the inevitability/contingency controversy” (2020), Luca Tambolo discusses counterfactuals in the historiography of science. Tambolo argues that

“In the case of general history, it is often possible to imagine a consequent dramatically different from actual history, and yet plausible; in the case of history of science, imagining outcomes far removed from the results of actual science seems more complicated” (p. 2012).

The argument is, in its core, simple and elegant. A historical counterfactual need to have a plausible antecedent and the consequent must follow from plausible principles concerning how the world works. Given that the actual science is our source of knowledge of how the world works, the consequent must be derived by using the results of actual science. This use of the results of actual science affects how a counterfactual narrative can develop. “In order to be plausible, the outcomes of counterfactual histories need to be appropriately continuous with [the] results [of actual science], which provide the yardstick for the assessment of the plausibility of counterfactuals” (2113). “What [–] we view as a plausible alternative to actual history of science is influenced by our currently accepted knowledge” (2123).

Tambolo discusses many studies from the existing counterfactual historiographies of science and points out that they all build a narrative which converges towards the actual results of science. I read Tambolo as making the following claims (or explanatory argument): 1. There exists a regularity in the historiography of science: (Plausible) counterfactual narratives converge towards the actual results of science, and 2. This convergence is dependent on the central role that actual results of science play in the building of plausible counterfactual narratives. 3. The results of actual science play a central role in the building of counterfactual narratives because each step in a counterfactual narrative is restricted by plausibility considerations and these considerations are based on what we know about how the world works, i.e. on the actual science.

This argument seems to have a very important consequence that Tambolo points out: Counterfactual historiographies of science seem to be somewhat unable to tell how science could have been differently. For example, I have argued (Virmajoki 2018) that the contingency of (a feature of) science depends on how plausible a counterfactual scenario where we have a different (version) of science is. The more plausible the scenario, the more contingent science. Given Tambolo’s analysis, my definition of contingency could force all historiographical inquiry to concede that science is inevitable. Given that it is difficult to come by with plausible scenarios where science is differently, science is judged to be inevitable. The existing scientific results guarantee their own inevitability through the backdoor.

I would like to make some comments on the basis of the considerations above.

First of all, it is not exactly a surprise if science turns out to be inevitable or if we are not in a position to judge it as contingent. Given that we trust what mature science says, we know how certain aspects of the world work.  For example, given a experimental setup and given that electrons have the charge 1.602176634×10⁻¹⁹ C, of course a well-performed experiment by Millikan provided a value near to the actual value. This line of reasoning can continue even deeper. It is possible to argue that, given the results of science, we can formulate demands that all laws of nature must satisfy and argue that “the set of demands [–] are nearly inevitable and non-negotiable” (Scardigli et al. 2019).

Secondly, the inevitability follows only if we assume, in addition to the power of our theories to capture how the world behaves, that adequate methods were used. To provide an obvious illustration, had the Aristotelian tradition persisted another millennium, there would not have been an oil-drop experiment in the early 20^th century. However, counterfactual scenarios like this do not have a plausible antecedent or at least the antecedent is benign from the perspective of the contingency of science, as the scholars say. A more interesting counterfactual could concern Ehrenhaft’s experiments. Ehrenhaft was a contemporary of Millikan and deduced a different value of electric charge from different experimental setup. What could have happened, had Ehrenhaft been the only one who worked on electric charge? One could argue that the errors in Ehrenhaft procedures (the errors we now think they had) would have been discovered sooner or later and the real charge would have been found. Be that as it may, the inevitability of science does not follow from the fact that world behaves in accordance with our current theories. We need to add “given that certain type of inquiry existed”.[1]

Third, the last point also means that it is at least possible to write a counterfactual history where different methods were used in the past and tell what would have happened in that history. In such history, we use our knowledge of the causal structure of the world and nature of different methods to tell where those methods would have led us. I say “possible” because it is difficult to imagine what could have happened in such cases. As Tambolo and others have pointed out, imaging different science is hard – and it is hard even if we are able to assume that we know how the world works. Be that as it may, it seems possible to have a counterfactual scenario where there is a different science. In that scenario, the science is, from our perspective, false and based on inadequate methods. Moreover, we could easily conjecture that the science which was a product of unsound methodology would not have been successful and therefore it would not have been philosophically interesting alternative to the present science. The philosophical question of contingency of science is usually framed in terms of equally successful but fundamentally different science. Be that as it may, there is no guarantee of convergence of counterfactual narratives on the actual science. It seems that the convergence is a consequence of additional assumptions we make about which counterfactual scenarios are important.

In the fourth place, we have different degrees of trust in different theories of science. Even if are never able to escape our existing knowledge in counterfactual scenario-building, we can assign different degrees of credibility on different counterfactual scenarios depending on our trust in the theories they are based on. Even if all (important) counterfactual scenarios of historiography of science converge to the present science and even if this indicates that we have to judge science as inevitable, we can still have different levels of confidence towards the adequacy of these judgments. In this context, our judgments about the contingency of science would track our degree of confidence towards existing theories. There would not be independent scenarios which tell how science could have been differently but only inevitability-indicating scenarios which we do not completely trust.

In the fifth place, we can formulate “meta-contingency” claims on the basis of the philosophical theories of epistemology of science. We could use philosophical arguments to determine how much we should trust a particular theory of science. For example, the notorious pessimistic meta-induction says that there have been successful but false theories in the past and therefore our successful theories may turn out to be false. Given the pessimistic meta-induction, we should have some skepticism even towards the most successful science. This indicates that our science-based scenarios should also be treated with a pinch of skepticism. Surely, the issue of contingency of science does not remain an independent issue on the meta-level but blends with old-fashioned epistemological questions. Someone could see this as a trivialization of the whole issue. I do not. I think we need a systematic analysis of the connections between different philosophical positions even if we cannot find an Archimedean point from which to tell which positions to ultimately accept.

In the sixth place, we need to see what the consequences of the convergence of counterfactual narratives are from the point of view of the estimating of the futures of science. To begin, we can notice that if historical counterfactuals converge towards the present science, then it is difficult to use those counterfactuals to tell how science could be differently now or in the future. Moreover, it seems that future-scenarios of science have the same tendency to converge (or remain close to) actual science. Even if we assume that something changes in how science is practiced in the future, our dependence on our current knowledge would probably lead us to infer that science will remain essentially similar as it is now (at least if we are not careful and reflective enough). However, there are some differences between the counterfactual scenarios of the past and scenarios of the future.

(I) The “put-up-or-shut-up” inevitabilism is based on the idea that as long as fundamentally different and equally successful science is not created, there is no way to tell whether we could have a different science. A standard response to this has been that, given the actual history and present, there simply have not been enough resources to produce an alternative science. However, the future use of our resources is an open question. It seems that we should be more skeptical about the science remaining as it is in the future than we are about the counterfactual scenarios where science converged to the actual outcome. For example, one obvious difference can be found in how much the resources can be allowed to change in plausible scenarios. In the 1700^th, it was impossible to spend billions of euro on research, for example. Of course, it might be that not much depends on the resources, but this independence should be established. Until it is, the amount of resources can make a difference between the logic of past and future scenarios, at least in principle.

(II) Sketching an alternative science is usually not enough to tell whether it could in fact have been an equally successful alternative to the actual science. However, sketching alternative future sciences is epistemically more merciful practice, as we do not expect all our future scenarios to actualize. Yet, the practice of sketching future science is indispensable part of the current science, as all novel hypotheses suggest how things could be understood in the future. In fact, the ruthlessness we have towards the sketches of alternatives for the present science is an interesting phenomenon in the philosophy of science. Surely, scientific hypothesis will be accepted only after they have been tested while counterfactual sketches cannot be tested (in the same way). However, the counterfactual scenarios of the history of science also serve different functions than scientific theories and hypothesis. It is important that we test the theories on which we base our estimates of COVID-19 pandemic, for example. On the other hand, counterfactual scenarios serve a different purpose. They are supposed to provide fresh perspectives on the actual science. One could argue that this could be taken better into account in the epistemic demands of philosophy of science.

(III) Meta-contingency should be taken into account when estimating the future of science. Different philosophical theories of science draw different pictures of the amount of change in science (and the causes of that change). It might not be possible to always sketch what an alternative future science could exactly look like, but we could be able to estimate, on the basis of philosophical theories, what kind of changes to expect. In such situation, we are not able to formulate detailed future scenarios, but we would nevertheless be able to “[challenge] conventional thinking in order to reframe perceptions and change the mindsets of those within organizations” (Wright et al. 2013) which is one of the most important functions of scenario-building. I have discussed this kind of approach in a previous post.

Finally, in the seventh place, we need to ask why the history-of-science-counterfactuals seem to converge towards the actual state of affairs, but other historical counterfactuals do not. In both cases, we need to apply our actual knowledge to the counterfactual past. First, notice that general historical counterfactuals do not, in fact, diverge from what science says. There are no plausible counterfactuals of the form “had X been the case, we would have Y” where Y violates the laws of nature. The difference between the history-of-science counterfactual and other counterfactuals seems to concern the flexibility and repetitiveness of certain causal processes – or at least our conceptions of them. In the case of science, we think that the uniform structure of the world and its repetitive effect on the human cognitive system shapes the beliefs as time passes. In the case of other histories, we more easily think that situations are unique and once the actual effect was not produced, there would not come similar opportunity. Had Hitler not been in power, there would not have been a war in 1939. And given the changes after the counterfactual ’39, a war would have had different characters (armies would have been differently prepared etc.) and maybe a different outcome. On the other hand, had Millikan not measured the charge of electrons, someone else would have worked with identical electrons in the future.

It all comes down to this: I think that above mentioned difference in our perceptions of the possible convergences of different histories is based on the implicit assumption that scientist and their beliefs are intimately connected to the world and causally interact with it. (Elsewhere, I have described this as the “pull of the nature intuition”). Assume that we accept scientific theories. Assume also that we accept that scientist do not interact with the world. Under these assumptions, it is completely possible to write a counterfactual history of science where scientists do not end up having similar beliefs as we do. In fact, it is difficult to understand how they could be able to describe how the world works without interacting with it (we of course think that our science is able to describe the structure of the world because it has interacted with the structure). Only if we accept that scientific theories describe the structure of the world (to some ontological depth) and that scientist interact with that structure, we are able to conclude that counterfactual histories converge to the actual state. The interaction of scientists with the repetitive structure of the world determines the (presumed) convergence.

I am not sure if the idea of intimate interaction of scientists with the world and the explanatory role of the interaction is compatible with philosophical antirealism. Here I might disagree with Tambolo (2020, p. 2123) and myself (Virmajoki 2019, 6.7). Needless to say, the issue requires further attention.

References:

Scardigli F., ’t Hooft G., Severino E., Coda P. (2019) Free Will in the Theory of Everything. In: Determinism and Free Will. Springer, Cham. https://doi.org/10.1007/978-3-030-05505-9_2

Tambolo, Luca (2020). So close no matter how far: counterfactuals in history of science and the inevitability/contingency controversy. Synthese 197 (5):2111-2141.

Virmajoki, Veli (2019). Cementing Science.

Virmajoki, Veli (2018). ”Could Science Be Interestingly Different?”. Journal of the Philosophy of History 12 (2)

---

[1] However, Scardigli et al. cited above argue that “We note that the actual laws of physics known to hold in our universe are quite close to what we have constructed purely by mental considerations. Of course, the author admits that this will be attributed to hindsight, but we claim that a super intelligent entity could perhaps have ‘guessed’ nature’s laws of physics from such first principles. Tis would be important to know, since this would encourage us to use similar guesses to figure out how the remaining physical laws, not yet known to us today, might also be guessed.” (p. 36)