Demonic and Golemic Science: Ethics in a World of Inductive Risk

“If any individual desire, and is anxious not merely to adhere to, and make use of present discoveries, but to penetrate still further, and not to overcome his adversaries in disputes, but nature by labor, not in short to give elegant and specious opinions, but to know to a certainty and demonstration, let him, as a true son of science (if such be his wish), join with us.” -Francis Bacon (Bacon 1902)

From the dawn of the Enlightenment, science was conceptualized as a step from subjectivity and towards objectivity (Goff 2019). This does not mean that scientists were thought of as being value-free, Bacon above makes a direct appeal to the wishes of his audience to recruit them (read: exclusively men) (Connor 2005) towards the scientific cause. Instead, the taking on of Enlightenment values and the abandoning of the “specious” old worldview would lead to grand discoveries of truth. Contemporarily, these values are described as epistemic values which, if held and followed, will take the holder closer to factual Nature than other forms of values, which is why some claim other forms of values should not have a place in the sciences. Epistemic values can be enumerated as a broad scope view of accuracy, simplicity, consistency, and fruitfulness (Kuhn 1977).

This perspective is certainly appealing, and deconstructing it completely would be problematic: it is hard to refute that science has taken humanity further than most other belief systems. However, if the scientific goal is to move towards the truth, the post-Baconian narrative must be examined more closely. Many philosophers of science have examined the topic, but this short paper will focus on the work of Heather Douglas (2000), whose critical eye asks us to take a closer look at Bacon’s idea that science can make us certain. Because he was wrong in that assertion, she says, non-epistemic values must take a serious role in the sciences.

Douglas leverages Carl Hempel’s concept of “inductive risk” (1965) to lay the groundwork for this issue: whenever we accept an empirically derived scientific belief, we are taking a gamble that we may be wrong. This is true both because of the potentially unsolvable Problem of Induction (Hume 1740; Russell 1946), and because many of our scientific theories are probabilistic (Talbott 2008) regardless of Hume’s problem. From Bacon to modern American pragmatism, science has been considered a tool to intervene in the world (Bacon 1902; Brinkmann 2013), but by the premise of inductive risk it seems that there is always a chance of this going horribly wrong. This is why, according to Douglas, other forms of values are required within scientific decision-making. She uses her extensive research on toxic dioxins to make this point (2000):

Imagine you are a scientist who is doing research on a potentially toxic chemical. You are running tests on rats to see how much chemical exposure can occur before the rats are harmed. Immediately, you are faced with a series of choices on how to structure your experiments. Do you search for a threshold where exposure becomes dangerous, or do you think of exposure as a sliding scale from least dangerous to most dangerous? How do you set your alpha levels in your statistical tests? P < 0.05 is considered the standard in many scientific fields, but are you willing to take on a 5% risk or greater that your study is wrong if the results you get could lead to human testing?

Douglas says that these sorts of choices exist in many forms of science, especially medical and social sciences. She demarcates the difference between sciences which do and do not require non-epistemic values in a somewhat arbitrary way, when “a scientist believes there is virtually no chance of being wrong.” Two metaphors for scientific models comes to mind: Laplace’s demon being able to perfectly calculate the entirety of both history and the future (Shermer 1995), and the golem of Collins and Pinch being a large, powerful, and soulless machine with the potential to do good or harm the innocent (McElreath 2020). Demonic science only requires the values which drive scientific progress, golemic science needs a guiding hand with ethical and moral thinking.

This arbitrary cut does not fully satisfy me; a more standardized probabilistic argument here could and perhaps should be made. How do we define the line between demonic and golemic science other than by assigning a Bayesian probability or p-value cutoff to assess such knowledge? Perhaps the physicists’ 5-sigma standard (Lamb 2012) could be used as a starting point for such considerations. Regardless, Douglas’s work offers a useful heuristic and a strong argument for why scientists should consider non-epistemic values. In our contemporary era, I cannot help but think of the social scientists who claim the side of Enlightenment values who have done irreparable damage while ignoring such thinking, such as Paul Krugman constructing arguments for economic globalization while developing models which assumed full employment despite employment being the primary concern for critics of the policies Krugman advocated for (Gredier 1997; Ruggiero 2005). With little consideration of the human factors, politicians leveraged his work to create a situation we are still suffering under today (Klein and Barlett 2008; Rose 2021). Francis Bacon’s vision for science was for it to create a better world, and in order to do that golemic scientists (economists, biochemists, and so on) must consider the ethical implications of their work.

References

Bacon, F. 1902. Novum Organum: Or, True Suggestions for the Interpretation of Nature. Translated by Joseph Devey. New York: P. F. Collier & Son.

Brinkmann, S. 2013. John Dewey: Science for a Changing World. New York: Routledge.

Connor, C. D. 2005. A People's History of Science: Miners, Midwives, and Low Mechanicks . New York: Nation Books.

Douglas, H. 2000. "Inductive Risk and Values in Science." Philosophy of Science 67: 559-579.

Goff, P. 2019. Galileo's Error: Foundations for a New Science of Consciousness. New York: Pantheon.

Greider, W. 1997. One World, Ready or Not: The Manic Logic of Global Capitalism. New York: Simon & Schuster.

Hempel, C. G. 1965. "Science and Human Values." In Aspects of Scientific Explanation and other Essays in the Philosophy of Science, by C. G. Hempel, 81-96. New York: The Free Press.

Hume, D. 1740. A Treatise of Human Nature. London.

Klein, D. B., and H. A. Barlett. 2008. "Left Out: A Critique of Paul Krugman Based on a Comprehensive Account of His New York Times Columns, 1997 through 2006." Econ Journal Watch 5(1): 190.

Kuhn, T. S. 1977. "Objectivity, value judgment, and theory choice." In The Essential Tension: Selected Studies in Scientific Tradition and Change, by T. S. Kuhn, 320-339. University of Chicago Press.

Lamb, Evelyn. 2012. 5 Sigma What's That? July 17. https://blogs.scientificamerican.com/observations/five-sigmawhats-that/.

McElreath, R. 2020. Statistical Rethinking: A Bayesian Course with Examples in R and Stan. Boca Raton: CRC Press.

Rose, E. L. 2021. "The decline of US manufacturing: Issues of measurement." Management and Organization Review 17(1): 24-28.

Ruggiero, A. 2005. "Paul Krugman and the New Economic Geography - assesment in the light of the dynamics of a "real world" local system of firms." ERSA Conference Papers. Amsterdam: European Regional Science Association. 5: 273-305.

Russell, B. 1946. A History of Western Philosophy. London: George Allen and Unwin Ltd.

Shermer, M. 1995. "Exorcising Laplace's Demon: Chaos and Antichaos, History and Metahistory." History and Theory 34(1): 59-83.

Talbott, W. 2008. "Bayesian Epistemology." Edited by Edward N. Zalta. The Stanford Encyclopedia of Philosophy Winter 2016 Edition. https://plato.stanford.edu/archives/win2016/entries/epistemology-bayesian.