Identifying Future-Proof Science

(Monograph with Oxford University Press, forthcoming 2021/22)



Is science getting at the truth? Sceptics claim that, given all the ‘scientific revolutions’ in the history of science, we should expect further major changes in scientific thinking in the future. They ask: given the history of science, wouldn’t it be naïve to think that current scientific theories reveal ‘the truth’, and will never be discarded in favour of other theories? After all, previous scientists thought so, and they were wrong. Through a combination of historical investigation and philosophical analysis, my work over the past ten years has defended science against such potentially dangerous scepticism, arguing that we should be confident that, by and large, contemporary science does reveal the truth (or at least something very close). My journal articles and blog posts on this topic have laid the foundation for this book, which is being written during my 2019-21 research leave owing to a British Academy Mid-Career Fellowship.

Extended synopsis

At the heart of the book lies a sceptical historical challenge: scientists of the past have sometimes thought that their theories were ‘future-proof’ – meaning that they would never be replaced – and then those theories were indeed replaced. Classic examples, well-worn in the literature, include the caloric theory of heat, the phlogiston theory of combustion, and the nineteenth-century aether theory of light. But my 2012 AHRC project ‘Evaluating Scientific Realism: A New Generation of Historical Case Studies’ showed that there are many further examples. My paper ‘A Confrontation of Convergent Realism’ (Philosophy of Science, 2013) starts with a list of twenty relevant (at least potentially relevant) historical examples. It turns out that there are many theories in the history of science which enjoyed significant success, sufficient to convince many scientists of the day that the theory was true (or at least approximately true), and then it later turned out that the theory was rejected in favour of some other, quite different theory.

These examples from the history of science are a serious challenge, because they seem to show that scientists can be fooled by the evidence. That is, they seem to show that scientists can think that the evidence for some theory is so strong that the theory has to be (more or less) true, when in fact the theory is radically false (and will later be rejected, replaced by a significantly different theory). And if scientists were fooled in this way in the past, then it seems quite possible that they could be fooled in this way in the present. This leads to the thought that we should be sceptical when we hear contemporary scientists say “There is now overwhelming evidence that XYZ” and “We now know that XYZ is true, based on the evidence”. Such statements might even – worryingly – include claims about anthropogenic climate change, the health benefits/risks of vaccines, the origin and nature of a new virus, and so on. Needless to say, scepticism about such things can have serious consequences (e.g. HIV/AIDS denialism in South Africa 1999-2008, and damaging hesitations when it comes to taking action on climate change). Tellingly, even some of the prominent defenders of science against such scepticism allow that the history of science shows that “scientific truths are perishable” (Oreskes 2019, p. 50).

My approach is to undermine the historical cases used to support such scepticism. For nine years now (and over two AHRC projects, in 2012 and 2014-18) I have been tackling historical challenges head-on, explaining why they should not undermine our confidence in many contemporary scientific ideas. I am now in a position to draw on that prior research, tackle some remaining historical challenges, and distil the most important historical lessons for judging the evidential status of contemporary scientific theories, bringing everything together in one long argument. The result is a monograph, completed during the course of my British Academy Fellowship 2019-21, entitled Identifying Future-Proof Science.

The book makes its case via scientific case studies, some historical and others contemporary. Historical case studies come from three different scientific fields – biology, geology, and physics – where scientific evidence was (with the benefit of hindsight) at least partially misleading. The chapters detail the cases and present responses to the challenges they pose. The lessons of these cases are then applied to contemporary cases, again considering different scientific fields. It is argued that many contemporary scientific claims are indeed future-proof, but that there are special challenges when it comes to certain corners of non-classical theoretical physics (notwithstanding the remarkable experimental success of theories such as quantum mechanics). The book ultimately argues that the historical basis for a strong form of scepticism about science crumbles upon close inspection, but at the same time it offers a concrete proposal for the circumstances in which we really should be sceptical.

The chapters of the monograph run as follows:

1. What is future-proof science?

2. The historical challenge to future-proof science: the debate so far

3. Meckel’s successful prediction of gill slits: a case of misleading evidence?

4. The Tiktaalik ‘missing link’ novel predictive success and the evidence for evolution

5. The judgement of the scientific community: lessons from continental drift

6. Fundamental physics and the special vulnerability to underdetermination

7. Do we know how the dinosaurs died?

8. Scientific knowledge in a pandemic

9. Core argument, objections, replies, and outlook

Chapter Outlines

Chapter 1. What is future-proof science?

This chapter introduces the subject matter, provides an initial framework for discussion, and offers a brief characterisation of key concepts (to be gradually enriched as the book progresses). An indicative list of 30 examples of ‘future-proof science’ is put forward, including singular scientific facts as well as more involved bodies of thought, or theories.

Chapter 2. The historical challenge to future-proof science: the debate so far

This chapter engages with the literature on contemporary scientific scepticism. First of all it is argued that some of the scholars who describe themselves as ‘sceptics’, or ‘anti-realists’, or ‘instrumentalists’, actually allow for some (even many) examples of ‘future-proof science’. This sounds backward, but that is only because of a confusing use of labels and problematic terminology. Most importantly, whilst the realism debate concerns our epistemic stance vis-à-vis ‘unobservables’, this book has no such focus, and in fact most of the proffered examples of ‘future-proof science’ concern observables (broadly construed). Thus many of the antirealist’s usual arguments – designed as they are to doubt our knowledge of unobservables – are not relevant. Some arguments still are relevant, since they equally affect scientific claims about observables and unobservables, including (a variation on) Laudan’s pessimistic induction, Stanford’s problem of unconceived alternatives, and what Frost-Arnold (2019) describes as ‘the problem of misleading evidence’. However, since the focus is now on observables, many antirealists will want to defend against these arguments just as much as realists, and the distinction between ‘realists’ and ‘antirealists’ breaks down. Thus this book is not a stance in the ‘scientific realism debate’.

              An important concept here and throughout the book is the concept of evidence. It is natural to appeal to evidence when one wishes to defend a scientific idea as ‘future-proof’. Most important of all is the challenge of judging the overall weight of evidence for a scientific claim, and assessing the idea that sometimes the weight of evidence crosses a threshold such that scepticism is no longer reasonable. Alternative approaches to judging evidence are critically analysed (a discussion to be filled in as the case studies progress in subsequent chapters).

Chapter 3. Meckel’s successful prediction of gill slits: a case of misleading evidence?

Chapter 3 concerns the evidential significance of J. F. Meckel’s 1811 novel predictive success concerning the existence of gill slits in the mammalian (including human) embryo. It is argued that this successful prediction, whilst prima facie impressive, only modestly confirmed Meckel’s theory of recapitulation. In addition, the scientific community of the day were not fooled: they successfully recognised the nuances of the particular scientific context bearing on the measure of the significance of Meckel’s successful prediction. This case thus demonstrates the reliability (even in the early 19th century) of the intuitions of the scientific community when it comes to matters evidential. It also serves to show that ‘novel predictive success’ is not always significant evidentially, even when the prediction in question appears to be ‘bold’ or ‘risky’.

Chapter 4. The Tiktaalik ‘missing link’ novel predictive success and the evidence for evolution

Chapter 3 showed that if one wanted to defend future-proof science by putting weight on novel predictive success, one would have to carefully articulate the circumstances in which a novel predictive success really is of great evidential significance; if not in the Meckel case, then when? Chapter 4 considers a possible case from contemporary science, concerning the theory of evolution and the novel predictive success of the Tiktaalik ‘missing link’ fossil in 2004. This predictive success seems more impressive than the Meckel case, and scientists initially expressed their enthusiasm for the discovery, indicating that they considered it evidentially very significant. However, the closer one looks the less evidentially significant it appears to be. This further consolidates the thought that we should not put weight on novel predictive success in our search for future-proof science. This inspires a different approach to identifying future-proof science, based not on attempting to digest the first-order evidence, but instead on examining the scientific consensus.

Chapter 5. The judgement of the scientific community: lessons from continental drift

Chapter 4 showed just how challenging it can be for the individual (philosopher, or scientist) to judge the first-order evidence for a scientific idea. I argue that the best judgement belongs to the scientific community. Drawing on Oreskes (2019), some preliminary suggestions are made concerning how the individual might go about ascertaining, and critically assessing, the judgement of the community. It might be objected that sometimes the community judgement goes wrong. A famous case concerns the community attitude towards continental drift between 1915 and 1965: continental drift was widely considered to be absolutely impossible, even provably so, but by the end of the 1960s the scientific community had made a complete U-turn, and continental drift was considered not only possible, but highly probable. Does this show that the scientific community cannot be trusted to judge the weight of evidence in a given case? I argue that a definite distinction can be drawn between cases such as continental drift, and many contemporary cases where we really should trust the judgement of the scientific community when they tell us that a given scientific claim is factual, beyond reasonable doubt.

Chapter 6. Fundamental physics and the special vulnerability to underdetermination

Chapter 6 introduces Sommerfeld’s 1916 prediction of the hydrogen fine structure spectral lines. Sommerfeld’s success was extremely misleading evidentially speaking: it seemed to strongly suggest that his theory of the atom was at least approximately true. Whilst there was no strong scientific consensus regarding Sommerfeld’s theory, the case demonstrates the capability of physical theory to recover empirical results whilst making claims about the nature of reality that are radically false. This suggests that we might reasonably expect very significant scientific revolutions in this particular corner of the scientific endeavour over the course of the next 1000 years, in a way we shouldn’t in other scientific fields.

Learning from such cases, should we say that there is a special vulnerability to underdetermination in the context of ‘fundamental physics’ (broadly construed), as Hoefer (2020) claims? If there is, then it might be prudent to add a caveat, introducing additional epistemic caution purely for this particular corner of the scientific endeavour. At the extreme, an additional filter would block all claims coming out of ‘fundamental physics’ by fiat, refusing them passage through the future-proof filter.

              Various scholars have offered ways to draw a principled epistemic distinction. Perhaps most famously, Van Fraassen placed great epistemic weight on the observable/unobservable distinction. Hoefer describes the distinction as that between ‘fundamental’ and ‘non-fundamental’ physics. Taking each distinction in turn, I find significant problems. Ultimately I argue that we do better to avoid placing great epistemic weight on any such distinction, instead employing caution at the point of interpreting any claims coming out of ‘fundamental physics’ that enjoy sufficient support to make passage through the future-proof filter.

Chapter 7. Do we know how the dinosaurs died?

I have argued that we can identify future-proof science via scientific consensus, with criteria that must be met. But actually applying those criteria to a specific case can sometimes be extremely challenging; how is one to ascertain whether there is a 95% consensus amongst relevant experts? A highly relevant contemporary case concerns the claim that an asteroid impact caused the dinosaur extinction. Many scientists have been tempted to state the claim as a fact, and in 2010 a review article was published in Science hinting at a scientific consensus. There was a significant community reaction to this piece, however. In addition, there has been plenty of opposition to the claim in both the published literature and activity at (some) major conferences, all the way through from 1980 to 2020. This chapter navigates some of the challenges that can arise when we ask after the strength of feeling in the relevant scientific community vis-à-vis a specific claim. The case carries important lessons for how scientists go about declaring a consensus of opinion, a matter of crucial importance if – as this book argues – we are to identify future-proof science via a certain kind of scientific ‘consensus’.

Chapter 8. Scientific knowledge in a pandemic

During 2020-21, millions (even billions) of people living through the coronavirus pandemic urgently wanted, and needed, answers to questions concerning scientific knowledge. Were all of the deaths definitely linked via a viral cause? Did it definitely originate in China in December 2019? Were the vast majority of children really safe? It is one thing to look back to 2020-21, quite another to face up to the fact that there will be future pandemics, and humanity will one day be asking these questions once again. One thing lacking in 2020-21 was a clear account of how the individual could identify the future-proof scientific claims, distinguishing them from other types of scientific claim, such as (what might be called) ‘promising hypotheses’, or ‘useful speculations’.

              Looking to the criteria for future-proof science put forward in this book, a worry arises that nothing scientists were saying, in 2020, about the pandemic could responsibly be called ‘future-proof’, since in 2020 so little time had passed for relevant scientific claims to be internationally scrutinized. But scientists did in fact have some relevant future-proof knowledge, even only a handful of weeks after the onset of the pandemic. This chapter explains how this is possible, given that usually absolute confidence in scientific claims depends upon extensive international scrutiny, often taking many years.

Chapter 9. Core argument, objections, replies, and outlook

This concluding discussion comes in two parts.

First, the chapter draws on the lessons from all the previous chapters to lay out (i) the criteria for future-proof science, (ii) the core argument behind these criteria, and (iii) a workable strategy for actually identifying future-proof science. I build on the ‘externalist’ suggestion put forward by Oreskes (2019) that the best strategy is to use certain tools to critically assess the status of the scientific consensus, as a proxy for evaluating the entire wealth of first-order evidence from a large number of different perspectives. The shift from ‘internal’ evidence to ‘external’ evidence supports calls for adjustments to science education in our schools, with greater emphasis on teaching the ‘external’, second-order, or ‘sociological’ evidence for scientific claims.

Second, this chapter raises some possible, outstanding objections, and provides preliminary responses.