Subtractive Epistemology: How to Add Value by Taking Things Away

The aim of science is not to open the door to infinite wisdom, but to set some limit on infinite error.—Bertolt Brecht

Pathological hoarding (syllogomania) is a well-known psychological disorder that can prove fatal, as in this case described by Jacob Shelton:

A 51-year-old compulsive hoarder in Spain died in early 2016 when one of his piles of garbage collapsed around him. His body wasn’t discovered until a friend … became worried after not hearing from his hoarding buddy for a while and called the police. The mass of trash … was so substantial that firefighters had to … help remove the body.

Hoarders’ collections can take up so much space in a home that, in order to move from room to room, the hoarder has to negotiate narrow paths through heaps of junk. A study by psychologists at the Worcester Polytechnic Institute Project Centre in Melbourne found in that, in Australia, almost a quarter of all deaths by fire in homes are caused by pathological hoarding.

The Benefits of Subtraction

The only way to alleviate the symptoms of pathological hoarding and improve sufferers’ quality of life is to get rid of almost all the unnecessary items. This is a vivid example of how subtraction can sometimes be, in effect, addition. And hoarding is not the only sphere in which this rule applies. For example, an accumulation of bureaucratic procedures can paralyse administrative systems; reducing their number can improve the systems’ effectiveness.

Subtraction can also be used to produce a work of art. As the philosopher Proclus of Athens once said, “Statues are made by subtraction.” When Michelangelo was asked how he had created the majestic statue of David, he is said to have replied, “It’s very simple: it was enough to remove everything that was not David.” He liked to say that he saw figures hidden in the marble, and that his only role was to bring them forth. And his four unfinished sculptures depicting slaves still partially trapped in their marble blocks show that this was indeed his approach.

Subtraction can also lead to addition in the area of human health. For example, giving up tobacco and reducing excess consumption of alcohol and sugar arguably yields more health benefits than, say, adding vitamin supplements to one’s daily regimen. Smoking is so deadly that, as Nassim Taleb details in his book Antifragile, if everyone in the world were to quit, the net health benefit would exceed the entire health benefit produced by the interventions of all the healthcare systems in the world.

Junk Science

Just as, over time, many of our homes become cluttered with objects that we no longer use or need, so too some scientific disciplines have become cluttered with theoretical constructs that are no longer used or needed. For example, when a group of health care professionals were looking for a way to encourage healthcare workers to wash their hands more frequently, they identified 33 different psychological theories claiming to explain why handwashing is neglected and suggesting how it could be increased: together, these theories referenced no fewer than 128 theoretical constructs. It is unlikely that all these theories and constructs persist because they are needed in order to solve such a simple problem. More probably, they have simply accumulated over time, because scientists tend to focus on adding new theories, rather than on discarding outdated or duplicative theories. For example, when a group of researchers reviewed the literature in psychology, anthropology, economics and sociology on how to encourage healthy behaviour, they counted 82 theories on the topic, only a handful of which had been subjected to rigorous empirical verification.

This kind of unhelpful accretion occurs partly because of practices that might fairly be called junk science. For example, some studies are commissioned by companies seeking to provide support for a product they sell—and their results are thus more likely to be biased or even deliberately misleading. And many other studies are produced simply because researchers must publish or perish, even though their findings are often of little or no value (as evidenced by the enormous number of studies that are cited by no one—not even their authors—in the first five years after they are published). The proportion of studies that fall into one of these two categories is so high that these disciplines may remind one of the home of a pathological hoarder: the proportion is estimated to be 82% in the humanities, 32% in the social sciences, 27% in the natural sciences and 12% in the medical sciences. (Although the proportion is lower in the medical sciences than in other disciplines, it is still a troubling statistic, especially when we consider the enormous cost of medical research.)

The Collecting Mindset

Another area in which clutter tends to collect is our shared knowledge base. In his 1981 book Critical Path, Buckminster Fuller calculates that, beginning about 2,000 years ago, our species’ knowledge started growing exponentially at an ever increasing rate—doubling in the first 1,500 years, doubling again in the next 500 years, then again between 1900 and 1950, again between 1950 and 1970 and again between 1970 and 1980. At this rate, by now the amount of knowledge is presumably doubling about every 12 hours.

Most of the changes to our knowledge base are additive rather than subtractive: we tend to accumulate new facts and ideas much faster than we discard old, useless facts and ideas. In the field of history this is probably inevitable: new things are always happening—at an ever-increasing rate, since the number of people on the planet is increasing, which means that there is an increasing amount of human activity to record. But in the field of science, knowledge is also increasing at an exponential rate. For example, whereas in the 1970s about 500,000 scientific articles per year were published worldwide, today that number exceeds four million. As a result, the sheer volume of information to which we now have access—most of which is of no practical value to us—can make it difficult and time-consuming to find the knowledge we need, despite the internet’s remarkably efficient search algorithms.

It has been estimated that only about 10% of new ideas are about eliminating previous ideas or activities. This suggests that the human mind prefers additive thinking. And indeed, that preference was confirmed by researchers at the University of Virginia, who in 2021 the results of eight experiments examining how people think when they are looking for ways to improve a procedure to more effectively influence other people’s views or behaviour. Participants were variously asked to look for ways to ease overburdened schedules, cut red tape, improve Lego constructions, improve golf courses and minimize human impact on the planet. Regardless of the task, participants tended not to consider how to reduce, rather than increase, the number of actions involved—even when such reductions were suggested to them. The experimenters hypothesise that this tendency may be due to cognitive overload: it is simply easier for us to add something to an existing procedure than it is to subtract something.

One of the co-authors of that research, Leidy Klotz, has elaborated on its findings in his 2021 book, Subtract: The Untapped Science of Less. Klotz shows how the gathering of food, and also other items, activates the same neurophysiological reward system in the brain that is responsible for the feeling of pleasure while eating and is stimulated after taking cocaine or browsing ‘likes’ on Facebook. In the case of pathological hoarders, it is probably triggered every time they add an item to their collections.

Our tendency to avoid subtraction is also described by prospect theory, which was developed by Daniel Kahneman and Amos Tversky based on research showing that we experience the pain of losing something much more strongly than we experience the pleasure of gaining the same thing. (For example, the pain of losing $100 is not compensated for by a subsequent gain of $100.) And getting rid of things—even things that we are not using or enjoying—can feel like a loss.

Via Negativa and Occam’s Razor—The Cleanup Tools of Scientific Subtraction

The instincts to take pleasure in collecting—and to be averse to loss—that helped our ancestors survive in the Pleistocene savannah have in some ways become a disadvantage in the Anthropocene. One way to help us overcome these instincts may be to adopt Nassim Taleb’s subtractive epistemology, according to which “the greatest and most robust contribution to knowledge consists in removing what we think is wrong.” (Subtractive epistemology is not to be confused with reductionism, which is an approach that studies reality by breaking a complex whole into smaller parts and attempting to describe those parts in order to help build a more detailed picture of the whole.)

The tenets that underlie subtractive epistemology were articulated in the Hellenistic era by Pyrron of Elida, the founder of philosophical scepticism, who emphasised the uncertainty of our judgements about the nature of reality. The Pyrronists refrained from making judgements in order to achieve peace of mind, and questioned the judgements made by other schools of philosophy. They pointed out many of the errors in the philosophical views of the time, and raised evidentiary standards in general.

Other ancient philosophers, such as Proclus of Athens and Gregory of Nyssa, together with Thomas Aquinas, helped develop subtractive approaches to problem solving. Apophatic theologians, in trying to discover the nature of god, developed an approach known as via negativa, which advocates defining a phenomenon (such as the nature of god) by what it is not. The via negativa continues to be widely applied, and not just in theology.

The medieval Franciscan monk William of Occam was an heir to these traditions. In postulating the principle we know today as Occam’s razor, he hoped to bring order to the multitude of speculative systems that medieval scholars had developed. He proposed that, when explaining phenomena, one should prefer explanations that are based on the fewest possible assumptions and concepts. The best-known expression of this principle is entia non sunt multiplicanda praeter necessitate (“entities must not be multiplied more than necessary”).

Add or Perish: Obstacles to Clearing Science of Clutter

Subtractive epistemology holds that it is just as important to discover which practices are ineffective as it is to discover which are effective. Identifying and eliminating ineffective practices is like chipping a piece off the artist’s block of marble—thanks to which the outline of the final sculpture becomes clearer. By identifying irrelevant relationships, ineffective methods and unproductive constructs, we come closer to the shape that the sculptor’s chisel seeks. But today’s scientific community has neglected this practice; instead, in general, scientists simply don’t publish negative research results. The result is that scientists have no systematic way to check whether someone has already asked the questions they are interested in and obtained negative results. (Only recently have there been attempts to change this state of affairs.)

Another reason why unnecessary knowledge accumulates in science is that authors routinely seek to cite the most recently published studies—partly because, if they don’t, peer reviewers often criticise their citations as not up to date—and they often neglect to cite older studies that remain valid. This is counterproductive. Something as solidly established as, say, theYerkes–Dodson law, described in a 1908 paper, does not lose its explanatory power purely through the passage of time. On the contrary, strong, well-described relationships tend to hold up over time; newer ideas, which have not yet been subjected to the test of time, are more likely to be discredited. The belief that what is new must be better than what is old is a cognitive bias known as neomania, and unfortunately, scientists are no more proof against it than lay thinkers. Scientific reports today can resemble the headlines seen on news websites—they are often just as brash in tone, and just as ephemeral.

Subtractive epistemology as applied to science primarily requires taking three things into account: negative research results, failed attempts to replicate previous studies, and reviews that are critical of research results. I am sceptical that this project will be widely undertaken by scientists. For one thing, many of them may fear that an emphasis on what isn’t working would discourage pluralism and creativity and damage credibility—as evidenced by the heated debates that tend to arise in response to reports of failed replication attempts.

Scientists should overcome these concerns: science demands orderliness, so it is important to work diligently towards institutionalising these subtractive approaches. Even though this will probably more quickly consign many newer hypotheses to the dustbin of history, the process will, like the sculptor’s chisel, bring us closer to the beauty hidden inside the marble block. Subtractive epistemology should complement rather than replace traditional research processes (without which there would be no progress). It could rescue us from the clutter and disorder that results when the creation of new constructs is not accompanied by the discarding of unnecessary ones. Unless scientists begin systematically using subtractive tools, science may simply suffocate and die—like an unfortunate hoarder, crushed by the inability to get rid of junk.