Climate science models and the scientific method: experimentation verses deduction?

Is this science and is it valid: What would Thomas Hobbes think?

In Hobbes view, Boyle’s experimental solution to the problem of order was not possible; it was not effective; and it was dangerous. (Shapin and Schaffer 1985: 80-81).

In discussing the methods of climate science, the use of models and the process of simulation are often mentioned as either a key strength or key weakness of the science.  While climate models receive a great deal of attention (and funding), they are generally a very marginal part of the science – much more time is spent examining actual data, identifying relevant data sources, developing statistical methods and making sure that observations and calculations are contextualised in terms of other data sets.  The fact that some statistical methods assessing correlations in actual data is also called modelling is confusion.

By modelling, I will refer here only to simulation using models, a type of experimentation.  I will contrast modelling with the statistical analysis of actual data to determine specific correlations that does not depend on extrapolation or simulation.  While examples of either approach can be contested, the issue I am concerned with here is the validity of simulation and experimentation using models in climate science.  This issue is fundamental if we are to understand the criteria such modelling must meet if it is to follow the scientific method. 

The scientific method seems to be one of the more straightforward issues in climate science - we all know what the scientific method is, right? We agree with the Oxford English Dictionary that it is “a method of procedure that has characterized natural science since the 17th century, consisting in systematic observation, measurement, and experiment, and the formulation, testing, and modification of hypotheses” I assume.  But let's go back to the 17th Century for a moment to understand the emergence of the concept.  The status of experimentation was the source of an important dispute between Robert Boyle and Thomas Hobbes during the 1660s and 1670s.  The disagreement between Boyle and Hobbes is analysed in great detail by Steven Shapin and Simon Schaffer in their book Leviathan and the Air-Pump (Shapin and Schaffer 1985).  The book examines, among many its themes, the boundaries and methods of science in relation to questions of political philosophy.  This is a good place to start because there were two competing approaches to “the” scientific method, best expressed in terms of the relationship between scientist Robert Boyle and the political philosopher Thomas Hobbes, or rather the science and political thought of Boyle and the political thought and science of Hobbes.  Hobbes and Boyle are key figures in the Enlightenment tradition, their approach to method being that of the mechanistic philosophy while their epistemology developed along similar lines within the rationalist tradition, but where they differ is in the status of experimentation itself.

Boyle’s approach to science is important because he argued for a new source of legitimation.  Shapin and Schaffer take this to be as much a political act, as it is a scientific operation.  This is because rather than appealing to a metaphysical condition, Boyle argued for an empirical condition by referring to specific experiments and emphasising the importance of witnesses in relation to the outcome of each experiment.  Boyle argued for the authority of the opinion of the witness by referring to an English legal act (Clarendon’s 116 Treason Act) stating that, in a trial, two or more witnesses were necessary in order to convict a suspect (see Shapin and Schaffer 1985: 327).  Boyle’s empirical method similarly relied upon the credibility of reliable witnesses, whose senses were corrected and disciplined by scientific instruments, analogous, perhaps to the discipline imposed on the senses by reason.  Boyle was interested in that which was perceptible, and his air-pump was an effective tool in that it controlled conditions for drawing out the relationship between various perceptions.  Hobbes opposed this method of observation and opinion, claiming instead that the only method able to gain universal consent was that of mathematical demonstration.  Hobbes seems to suggest that Boyle’s method was dangerous because it separated knowledge from power.

For Hobbes power is knowledge and in this unity there is only one type authority i.e. the judgements of civil society as represented by the figures designated by the social contract.  No class or individual, not the king not God and certainly not the masses, are above the law or constitute the basis of an alternative authority to challenge the judgements of law.  This is because the social contract, which forms the basis of civil society, is entered into by common consent through the calculations of individuals aiming to protect themselves from the fate of a nasty, brutish and short existence within the state of nature.  Hobbes argued that the method that he himself had developed was the only way of securing order.  Such social order that would avert civil war was for Hobbes embedded in the order that could be calculated from the natural world through philosophical practice.  According to Hobbes, then, The Royal Society, and Boyle in particular, broke the unification of knowledge and power, creating, instead a pool of opinion, opinion that was independent of the state.  Such opinion could never be universal and mathematical, but must be formed on the basis of inconclusive experiments and the (demonstrably!) deceptive senses.  With the implementation of Boyle’s methods, the basis of order had been undone.  With it, the academics and scholars challenged not just the king’s authority but the traditional notion of authority itself in the name of nature or in the name of the matters of facts produced by such things as the air-pump.

The air-pump, then, was in Boyle’s view able to establish a matter of fact independently of the machinations of political order, to serve as the foundation of knowledge and secure assent insofar as it is not regarded as man-made.  In this way, matters of fact can be stated as a given, mediated, of course, by a “neutral” technological device, but nevertheless with the authority of nature and not merely a matter of human opinion.  For Hobbes, this would lead to disagreement over opinions and, ultimately, political uncertainty.  For Boyle it would lead to scientific process, divorced from any such political considerations.

Shapin and Schaffer attempt to draw out an immanent critique of Boyle’s methodological assumptions arguing that scientists do not merely become individuals through which the objects and the “matters of fact” themselves speak.  They indicate that Boyle develops conditions for the establishment of facts and that these depend upon various conditions of credible witnessing.  Such witnessing was to be public and collective, but the conditions for being part of the reliable witness community or being an appropriate public site were developed in a way that is neither intuitive nor unproblematic.  Hobbes himself took Boyle’s conditions to almost exemplify the practices of a secret society.

Testimony was to be obtained from men hoping merely to advance natural philosophy rather than their own reputation or interests.  Theories, hypotheses and speculations were to be spoken about with caution.  In disputes over matters of fact, experiments would decide the case, not rhetoric.  Such evidence would enable those in error to renounce their former position.

But Hobbes’ method is entirely problematic.  It is clear that his a priori rationalist philosophy is not a viable alternative to experimentalism, because it is based on an invalid argument.  That is to say, it is not possible both to make up definitions and postulates in an arbitrary manner and claim that deductions from these definitions and postulates give certainty about the real world.  The example of Hobbes’ hostility to the air-pump experiments doesn’t really clarify how Hobbes intended to otherwise deduce truths. 

In terms of climate science simulation models, the analogy should have dawned on the reader some paragraphs ago, but I’ll make the point explicitly – for Boyle read climate science modeller, the air pump, a computational model, for experimentation read simulation.  For Hobbes, the analogy is the statistician, sceptical of simulation and the “secret society” of modellers.

In my view, there is no method of science that has a logic and set of principles that can crank out truths like a machine, the way Boyle seems to describe his own method.  However, it seems equally true that deducing truth about the world from firm foundations, Hobbes’ preferred method, is unfeasible.  Very little could be done with a strict adherence to either type of methodological principle.  I see simulation as a useful heuristic device, supportive of, or triangulating other methods, illustrating key issues and testing specific relationships, and as artificially (counterfactually) as the air pump.  But methods are not everything – they must involve the craftsmanship of the experienced and capable researcher.  As Bruno Latour concludes:

Shapin and Schaffer have access to thousands of archival pages on Boyle’s ideas and Hobbes’s, but nothing about the tacit practice of the air pump or the dexterity it required. (Latour 1993: 82)