The social sciences, some would argue, suffer from a ‘soft’ problem.
As Laurence Smith et al. describe in a 2000 article published in the aptly-named, Social Studies of Science, “Dating back at least to the writings of Auguste Comte, it has been thought that the sciences can be arrayed in a hierarchy, with well-developed natural sciences (such as physics) at the pinnacle, the social sciences at the bottom, and the biological sciences occupying an intermediate position.”
This hierarchy indicates somehow the ‘hardness’ or ‘softness’ a discipline. The natural sciences are more purely ‘science;’ more genuinely a description of nature as it is. The social sciences, on the other hand, are ‘softer’ – less predictive, testable, rigorous, or, perhaps, simply more subjective.
It’s generally unclear just what defines the hard/soft hierarchy, but in comparing a number of different definitions, Smith continually found the same thing: physics is the hardest science, sociology is the softest. Chemistry and biology are both well in the ‘hard’ science camp, while the analytic social sciences of psychology and economics skirt the ‘soft’ boundary and approach ‘hard’ territory.
This model makes social science out to be the poor cousin of the more prestigious natural sciences.
Whether you agree with that assessment of the social sciences or not, the inferiority complex and sense of always needed to justify the existence of one’s field effects the way social science is done.
As Danish economist and urban planner Bent Flyvbjerg describes, “inspired by the relative success of the natural sciences in using mathematical and statistical modelling to explain and predict natural phenomena, many social scientists have fallen victim to the following pars pro toto fallacy: If the social sciences would use mathematical and statistical modelling like the natural sciences, then social sciences, too, would become truly scientific.”
This pushes the social sciences down a computational path – a route, Flyvbjerg argues, which leads these otherwise valuable disciplines to produce more and more amounting to less and less.
“The more ‘scientific’ academic economics attempts to become,” he writes, “the less impact academic economists have on practical affairs.”
Furthermore, the whole attempt is foolhardy. As Flyvbjerg argues in Making Social Science Matter, “social science never has been, and probably never will be, able to develop the type of explanatory and predictive theory that is the ideal and hallmark of natural science.”
In emulating the computational and analytical approaches of the ‘hard’ sciences, social science aims to be something it is not and looses itself in the process.
As an (aspiring) computational social scientist, this argument seems like something worth thinking about.
Perhaps Flyvbjerg is too quick to write off the value of statistical approaches in social science, but nonetheless I find he has a compelling point.
Rather than trying to capture the episteme of natural sciences, Flyvbjerg argues the social science would do better to embrace phronesis. As he explains:
“In Aristotle’s words phronesis is a ‘true state, reasoned, and capable of action with regard to things that are good or bad for man.’ Phronesis goes beyond both analytical, scientific knowledge (episteme) and technical knowledge or know-how (techne) and involves judgements and decisions made in the manner of a virtuoso social and political actor.”
Essentially, social scientists should not obsess with trying to measure and quantify everything, but should rather aim towards the humanist goal of seeking to understand what is good and what is bad.
Perhaps unlike Flyvbjerg, I don’t see an inherent conflict between these aims. I can imagine that amidst the realities of a bureaucratic academy and fervent publish or perish pressures, scholars might find themselves forced along a too narrow path – but I see this as a broader challenge facing academia, not a singular failing of social sciences.
There is, I think, great value in developing computational models for complex social systems; in seeking to quantify and measure numerous facets of human interaction. The failing in this episteme approach comes only when phronesis is ignored completely.
In his own work on urban development, Flyvbjerg has a great saying: power is knowledge.
“Power determines what counts as knowledge, what kind of interpretation attains authority as the dominant interpretation,” he writes in Rationality and Power. “Power procures the knowledge which supports its purposes, while it ignores or suppresses that knowledge which does not serve it.”
These words come amidst his in-depth account of the bureaucracy and power which continually corrupts an ambitious urban development project in Aalborg. Most notably, this corruption rarely comes in the form of overt suppression, but rather a subtle, persistent distortion of information. “Power often ignores or designs knowledge at its convenience.”
This reality is in sharp contrast to the democratic ideal which “prescribes that first we must know about a problem, then we can decide about it. For example, first the civil servants in the in the administration investigate a policy problem, then they inform their minister, who informs parliament, who decides on the problem. Power is brought to bear on the problem only after we have made ourselves knowledgeable about it.”
Accepting the distorting effect of power, it’s reasonable to be skeptical of computational “knowledge.” In this sense, an episteme approach would only serve to further the interests of power – adding scientific credibility to an already distorted presentation of knowledge.
This is a valid concern, but again I find it to be a question of extremes. All methodological choices have consequences, all findings require interpretation. Understanding that dynamic has more value than walking away.
Power is knowledge isn’t an admonition that knowledge ought to be abandoned all together – rather it is a reminder: knowledge isn’t produced in a vacuum. Power shapes knowledge. Try as you might to be neutral and unbiased, this dynamic is inescapable. The computational social scientist is intrinsically a part of the system they seek to study.
