Selling off the soul of science

As Dan Hind points out in The Threat to Reason, "Reason and science can be empolyed for swindling ends but they can also serve in the cause of human liberation. The decision to treat human beings as objects of rational administration does not derive from the operations of rationality. It is an act of will." Below, Eadaoin O'Sullivan tries to rescue science from its ideological hijacking, and suggests that in fighting against technocracy, we should be wary of being drawn into a fight about who's got the smallest p-values.

Some weeks ago Colm McCarthy, during an interview on Morning Ireland, was asked about the unreliability of the ESRI’s economic forecasts. His response was brusque, and along the lines of ‘Economics is an inexact science. Forecasts are not gospel and are only estimations.’ There are two possible responses to this burst of professional honesty: (1 - obvious) You lot weren’t saying that five years ago; (2 - important) Exactly. So why do we persist in making the hypotheses generated by economic research into the immovable fulcrum around which all debate about politics and society must revolve, given that those hypotheses are only ever tentative and wide open to falsification and consequently no more valid a basis for political decisions than other tentative hypotheses about the importance of justice, fairness and equality?

Fundamental to the idea of technocratic administration is an obsession with quantification. Targets must be set, checkboxes checked, stats run and judgement passed. Underpinning this obsession is a nebulous grasp of the power of the scientific method to produce information that can usefully be turned into knowledge and applied. Used in material sciences, this method is incredibly powerful and effective. Used in human sciences it is less so.

Very simply, the scientific method involves the following steps:

1. Make a set of observations regarding the phenomenon being studied.
2. Form a hypothesis that might explain the observations. (Inductive Step)
3. Identify the implications and outcomes that must follow, if the theory is to be true.
4. Perform other experiments or observations to see if any of the predicted outcomes fail.
5. If any predicted outcomes do fail, the theory is proven false since it is true that if A implies B, then not B implies not A. (Deductive Step) It is then necessary to change the hypothesis and go back to step 3.

That’s taken from Wikipedia and step 5 should be clarified – while a theory may be proven false, that does not imply that any theory is ever proven. Hypotheses are only ever supported or unsupported, they are never proven to be true. This is central, as it means that enquiry never stops, and yesterday’s findings may be contradicted by today’s with no invalidation of the method.

Because the scientific method is fundamentally about process rather than certainty, it tends not to do very well in mediated public discourse. In this bowdlerised version, scientists are always discovering, showing, proving, and predicting. They are never tentatively suggesting a possible link that needs much more work before it can be supported with a degree of certainty that’s only as certain as the constrained parameters of the study at hand can provide and bearing in mind its limitations and its place within its field and the fact that any inferences drawn from its results are just inferences, and may or may not be supported by the evidence at hand due to the subjective nature of inference meaning those inferences themselves must be tested, modified, retested…and on and on.

Material science suffers less from problems with inference for the simple reason that its predictions are more easily tested. If I infer from the results of my research that one extra strut on a three strut bridge will make that bridge more stable I can test that inference quite easily by adding the extra strut and testing the stability of the bridge, keeping all else constant. If, however, I infer from the results of my research that breastfeeding for six months leads to higher scores on IQ tests when a child is six, I can’t test that inference by taking two groups of women, preventing one group from breastfeeding and forcing the other to do so, and keeping both groups in identical environments for six years. Even if I could, I couldn’t discount the potentially confounding role of some genetic aspect. Consequently, I can’t say with any degree of certainty that the variable (breastfeeding) I have inferred from descriptive or observational data as primary cause in higher IQ scores at age six is indeed the primary causative factor. I can only propose it and go on to investigate other contributory factors, all the while watching time march ever onwards, throwing up previously unimagined contributory factors. None of which is to say that such research is not valuable, it certainly and absolutely is, but it’s only interpretable if its bristling array of caveats are given equal cognitive airtime as its headline results. And they never are.

Imagining that I did overcome methodological hurdles and find a strong and reliable correlation between breastfeeding and higher IQ scores at age six that’s only the beginning of my problems, however. Having found that such a correlation exists, the results of my research are now thrown into the bear pit of public and policy debate, where decisions about whether or not to breastfeed for six months must be informed by the strength of the correlation as offset against the disadvantages or difficulties of spending six months breastfeeding, as well as choices about whether to rely on breastfeeding to bump the kid’s IQ score or to engage in some other kind of compensatory activity (spending a couple of hours a week doing practice tests, for example). Methodological soundness within one set of limited parameters is all well and good, but once the results are set loose to bump up against a much wider set of parameters they become objects for discussion, not discoveries.

There are very serious arguments to be had about the applicability of scientific methods to human sciences. If the results of this method of enquiry are seen as tentative, problematic, and uncertain; contributions to ongoing arguments about human and social life that are as valid as but not more valid than the knowledge gathered through methods that rely more on heuristics than hard data; then all is well. If, however, they are seen as more valid than qualitative, heuristic approaches, there is a problem. And the problem has nothing to do with the method. Nor does it devolve on some kind of post-modern rejection of objective facts. Frequently, criticism of the use of quantitative methodologies in human sciences is met with sputtering about extreme constructivism, and accusations that the critic is living in a fairy la la land where everything is relative and nothing really exists. Such sputtering betrays only a failure to grasp that objective facts in science are only facts that are well supported by the evidence; whether or not they have an ontology in a philosophical sense is another argument entirely. That objective facts are bowdlerised in this way betrays not just a failure to grasp the limits of the scientific method, it betrays also a worldview that is a confused mish-mash of misunderstandings about science, yoked to imperialising ambition for it, an ambition itself derived from the confused mish-mashes of misunderstandings about science that have been a feature of public discourse since the 18th century.

…scientistic prophets have regularly made [the] mistake of expecting the wrong kind of thing from science. They have been unconscious flatterers who got it the wrong kind of reputation. What they promoted as scientific thinking was actually a series of uncriticised ideologies, which gradually diverged from mainstream Enlightenment thinking in various alarming directions.

(Mary Midgley, The Myths We Live By, p. 17.)

Many people have an intuitive grasp of the fact that science is inflected by ideology and assymetries of power. What’s less well understood is the fact that the science they distrust is not the method itself, but its ideological transliteration. As Midgley points out, expecting the wrong kind of thing from science has got it a bad reputation. This in turn has led to a (frequently dangerous) backlash against science – most obvious in the popularity of homeopathy, crystal healing and magical magnets – with that backlash, as Dan Hind explores at length in The Threat to Reason, being opportunistically seized upon by the very people responsible for the bastardisation of science and trumpeted as evidence of the irrationality and idiocy of the public. With the public so characterised, calls for public administration by rational experts are plausible. The debate, at this point in proceedings, has been very efficiently moved along – away from talk about limits, methods, and details, to talk about how best to use knowledge now characterised as unproblematic and certain to arrange social life. The veil has been drawn and methods obscured, and the terms of debate are now solely about who has the biggest and bestest numbers.

Throughout Technocracy Now! we have pointed to the structures of power that are in play in talk of expert or elite administration of public life. This quick and dirty rundown of the limits of science is a small addendum to that, in the hope of offering another weapon with which to fight the forces of technocracy. Before ever Desmond and co get to the point where they’re talking about technocratic administration we should be taking a leaf out of Micheál Martin’s playbook in his one-on-one with Vincent Browne and challenging the very premises of their enterprise. Opponents of the status quo were given a gift when the financial meltdown of 2008 cast the flaws in neoliberal economic orthodoxy into sharp relief. It’s a gift that keeps on giving, and an invaluable rhetorical trump card, but there should be some leeriness about allowing dissent rely too heavily on statistical catfights. We should also focus on moving the Overton window around, pointing out again and again that political choices can only ever be negotiated, not determined.