April 11, 2020 Reading Time: 5 minutes

Yes, epistemology is a big word but it is just philosophy-speak for the study (logy) of knowledge (episteme), of, as Homer Simpson might say, knowing how we know.

Trust me, reading a full dissertation on the subject is not for everyone, but it is essential for everyone to understand the basics:

  1. Human brains naturally impose meaning on the world. 
  2. In addition to their rather limited senses of sight, hearing, smell, taste, and touch, humans must contend with cognitive constraints, ingrained ways of interpreting stimuli.
  3. All empirical data is a cultural artifact, created by somebody or some entity for some purpose. (See James C. Scott’s Seeing Like a State for a brilliant introduction to the ways that governments try to make everything “legible” by imposing quantification upon it.)
  4. Even if pure empirical data existed, it alone could not create meaning outside of some ideological, hypothetical, or theoretical construct.

A group of philosophers, called postmodernists, concluded from those four points that Truth cannot exist independently from Power. Claims to know some proposition X were made and maintained because some powerful person or persons benefited from others believing X.

The postmodernists had a point but they took it too far because if a Real World test for X exists, it will in due time be found wanting, brilliant, or good enough for now:

Humans thrive under Socialism = wanting

Every nation always has a comparative advantage at producing something = brilliant

COVID-19 is more dangerous than the flu = good enough for now

The classic “scientific method” is supposed to help humans to reject propositions that don’t match the real world while keeping those that do. Unfortunately, scientists cannot always run controlled experiments with double blind, randomly-assigned test subjects and social scientists certainly cannot do so, relying instead on so-called “natural experiments,” which are always a bit messy, and instrumental variables (IV), which are also flawed but a big improvement over previous statistical testing methods.

But as the Coronavirus Crisis has taught anyone paying attention, even “hard” science does not have it “easy” when it comes to discerning Truth from Error. Data can be “massaged” by removing “outliers,” putting in “lags,” smoothing in different ways, and on and on, hence the old joke about white lies, damn lies, and statistics. The distortions do not have to be deliberate to be a major problem.

Then there are models, equations used to make sense of empirical data. They are not infinitely malleable but they are certainly more flexible than my legs after a month out of the gym! Alas, they often tell us more about the model maker than the real world.

Somebody opposed to lockdowns could, for example, construct a model predicting the horrific death of any number of Americans desired due to recent government policies shutting down major swathes of the U.S. economy. 

Let’s start by guesstimating, say, one million Americans will die due to suicide, drug use, malnutrition, and so forth. Then assume that they were all fertile and each of those women killed by the lockdowns would have had 20 children and each man would have sired at least that many. 

So, by the terms of our model, the lives of at least 21 million Americans will be stamped out by the government’s reaction to COVID-19. But of course those 20 million unborn would have had 20 children each, whose existence has also been squashed by the lockdown.

Yes, that model, which makes the Holocaust look inconsequential, is ridiculous but what if the person espousing it is in a position of power? The powerful person has numbers that “add up” so they are Truth, and should be used to make policy, right? Policy that we must all obey, or else. Right?

No! A model is just one of many different possible views of the world. But recognizing that does not mean that we should fall into the postmodern trap that says that all ideas are equally useful (or useless) because ways of distinguishing stronger models from weaker ones exist.

Structural models, for example, are generally preferable to reduced-form ones. Reduced form models go right from cause to effect, e.g. SARS-CoV-2 → death. 

A structural model posits a causal chain, e.g. SARS-CoV-2 → transmission → COVID-19 → hospitalization → death. The model allows scientists to see the ways that the virus might not lead to death, through “leakages,” like people not getting infected, infected people remaining asymptomatic or subclinical, receiving treatment at home, or effective intensive-care in the hospital.

But even similar structural models can lead to very different projections of death based on how big the “leakages” are believed to be. Different scientists will form different assumptions from historical analogs, early case studies of the disease, or, as a last gambit, parameterization or the guesstimation of lower and upper bounds.

If predictive modelling is starting to sound more like “art” than “science” to you, great job at paying attention. We are almost “there.”

Because science is not just hard “facts” but interpretations of the world, logical (one hopes) and empirical (again, ideally) interpretations but interpretations nonetheless, governments should never ever, ever, ever, ever, ever base policies, especially ones with earth-shattering implications, on one model alone. Even science needs checks and balances.

For starters, society needs to have lots of different models (about everything, not just epidemiology), from lots of different people paid from a variety of sources, from competing think tanks, universities, and different levels of government. Then people with a stake in discerning the more realistic from the more fanciful models need to test them against the real world, not public opinion. And if there are no clear winners in the competition to model some important aspect of the real world, policymakers should heed the medical Hippocratic oath and “do no harm.”

But what if time is of the essence, as with the novel coronavirus? Let’s be real here, America lost that battle in January precisely because it relied on one model of the world, the government’s. The snap policy decisions made since then have been the most tragic case of CYA since Watergate. If the government was not always presenting itself as omniscient and omnipotent, Americans would once again expect little of it. 

Instead, it has rendered itself a “compulsory monopoly” in the words of AIER president Edward Stringham. One fears that if an actual zombie apocalypse began, the federal government would put troops in the streets to make sure everybody was (re)killing the undead using only FDA-approved bullets and baseball bats.

A one stop shop with a God complex is a dangerous combination because poor decisions generally stem from top down bureaucratic structures, especially ones, like ours, with poorly designed incentives. If America had taken a more pluralistic approach to policy, with lots of competing models and better alignment of incentives with goals (minimize deaths now is it?), we would not be in the current mess, with revolution potentially just a blackout or food shortage away and sharks probably in the process of being jumped. 

But surely, you might ask, don’t we need a “strong leader” to “get things done” and “make the tough decisions”? Please refer to the long string of “no” above. A single approach might be fine for Andorra but in a large country like the United States a monolithic approach is almost certainly going to be the wrong one for many. 

When there is no clear right answer, heterogeneity is definitely our friend because it can show us quickly what works and what doesn’t and under what circumstances desired outcomes arise. We have a little of this in the United States in the states that have not locked down but even those have implemented policies, like closing schools, that might have been counterproductive.

In sum, post-crisis we do not just need new leaders, elected and bureaucratic, we need a whole new approach to policy problem-solving and decision-making, a more pluralistic one replete with checks and balances. You know, the Constitutional and traditional limits on the arbitrary exercise of political power that made America great.

Robert E. Wright

Robert E. Wright

Robert E. Wright is the (co)author or (co)editor of over two dozen major books, book series, and edited collections, including AIER’s The Best of Thomas Paine (2021) and Financial Exclusion (2019). He has also (co)authored numerous articles for important journals, including the American Economic ReviewBusiness History ReviewIndependent ReviewJournal of Private EnterpriseReview of Finance, and Southern Economic Review. Robert has taught business, economics, and policy courses at Augustana University, NYU’s Stern School of Business, Temple University, the University of Virginia, and elsewhere since taking his Ph.D. in History from SUNY Buffalo in 1997. Robert E. Wright was formerly a Senior Research Faculty at the American Institute for Economic Research.

Find Robert

  1. SSRN: https://papers.ssrn.com/sol3/cf_dev/AbsByAuth.cfm?per_id=362640
  2. ORCID: https://orcid.org/0000-0003-3792-3506
  3. Academia: https://robertwright.academia.edu/
  4. Google: https://scholar.google.com/citations?user=D9Qsx6QAAAAJ&hl=en&oi=sra
  5. Twitter, Gettr, and Parler: @robertewright

Get notified of new articles from Robert E. Wright and AIER.