Work in Progress: The Paradox of Scientific History
Friday 24 June 2022
A problem that increasingly bothers me both in relation to civilization and to history (and, by extension, philosophy of history) is the reality of historical complexity in contrast to the simplified abstractions necessary to scientific understanding. This may prove to be one of the great barriers to the expansion of scientific knowledge, such as we experience with the limits of scientific observation (e.g., we can’t observe subjectivity and we can’t observe events before the retrodiction wall or after the prediction wall). At the same time, knowing an important limitation of science suggests opportunities to find ways around the limitations. Knowing that the limitation is there is far preferable to remaining within a framework and never knowing that one is bound by certain limitations.
I have often over the past few years pointed out that, while there is a significant body of literature in the philosophy of science on scientific method, there is only the barest treatment of scientific abstraction, which is equally important to the method of science. Robust abstractions are by their nature reductionist, but they are, at the same time, powerful tools that allow us to deduce much more than is not evident, which suggest experiments that make the initially inevident evident. From this we can easily see how research programs at the frontier of scientific knowledge can go wrong, and how there are conflicting results across research programs. Distinct sets of abstractions lead in slightly different directions, and the more powerful these tools of abstraction are, the further they lead us in a particular direction. But the particular direction of one set of abstractions will differ from the particular direction of another set of abstractions.
Of course, the reality of scientific research is much more complex. The enormous scientific research programs of the present day share many if not most of their presuppositions. The distinctive abstractions that characterize different research programs, and which lead these research programs in different directions, are relatively subtle compared to the robust abstractions of the recent scientific past, which allowed the enterprise of science to get started in a promising direction. Newtonian physics, for example, represents a number of robust abstractions that allow for the derivation of powerful results that closely approximate actual events. The Newtonian research program, however, has its limitations, and other physics paradigms, with other more subtle abstractions, have superseded it.
The various research programs that today give different results for the Hubble constant share most of the same presuppositions — they have the great majority of their physics in common — but point to different values for the constant. Last I read on this research, of the many clever methods used to measure the expansion rate of the universe, they tend to cluster around a couple of different values for the Hubble constant. There is no clear way to rationalize the results of the different methods at present, which suggests physics and cosmology not part of the standard model, which in turn suggests the possibility of new theories with new abstractions.
While the various methods for finding the value of the Hubble constant differ, the differences between the values, while stubborn and persistent, aren’t all that great; early methods of measurement were insufficiently precise to determine that different methods point in different directions. However, because of the anthropocentric norms built into science, because it is human beings who are taking the measurements, even a small discrepancy at the scale of the universe as we know it today, if extrapolated to the infinitely large or the infinitely small, would lead these methods in very different directions. Thus our physics and even our cosmology is most accurate the more closely it hews to a human scale; the further we depart from a human scale, the wilder our theories become, so that we really know little or nothing about the ultimate origins or the ultimate fate of the universe, since these occur at a scale that is so far beyond anthropocentric norms that present science cannot constraint them to a single scenario.
I’ve gone a little off topic here, but the point is that we use abstractions for constructing scientific knowledge, and different abstractions can lead us in different directions. Simplified models give us dramatic results, but these dramatic results, which seem to point to a definitive way of understanding the world (“settled science” in contemporary parlance) come at the cost of abstracting from a mass of detail, some of which may prove to be highly relevant, but which don’t register in the model we are using. Models of cosmology today use the abstract concepts of dark matter and dark energy to try to make sense of the evolution of the universe — abstractions that didn’t even figure in cosmology a century ago.
Similarly, when we attempt to use abstract concepts to understand history or civilization, we abstract from a great mass of detail that may prove to be important eventually, even if it does not seem particularly important at the present time. Hence the origins of the ongoing dispute within the philosophy of history whether history is a science like the natural sciences, or whether it is something rather different — not a science that converges upon principles and laws, but a science, such as it is, that seeks to lay out contingencies in all their contingency. Karl Popper called the attempt to formulate a predictive science of history “historicism.” This is an outlier usage, as most other philosophers of history use “historicism” for something else entirely, but that fact that Popper wrote an entire book devoted to the denial of what he called historicism is significant.
The attempts to produce predictive theories of history are uniformly disappointing. The philosophies of history that have taken this path are today something of a rogue’s gallery used to frighten children: Hegel, Spengler, and Toynbee. The thing about these figures is that they proposed suitably simple abstractions such as one would expect from the earliest stages of a science in development. Hegel cast the whole of history as a development of absolute spirit, which passes through distinct stages of development among distinct peoples in different parts of the world. Spengler and Toynbee are similarly schematic, and they similarly fail to capture either the realities we know in relationship to history, or to predict anything useful in regard to the direction in which history is headed.
There are ideas in Hegel, Spengler, and Toynbee that have value, if extracted from their over-simplified context and then employed in a more subtle and sophisticated context. Unfortunately, philosophy of history hasn’t been a cumulative process like the most successful sciences that build upon past efforts and go beyond them. The admittedly oversimplified abstractions of their theories of history could be “limiting cases” of a more sophisticated theory of history, as Einstein said of earlier physical theories, but the later theories have been formulated at approximately the same level of abstraction, so that they cannot realistically employ rudimentary abstractions at the same level of scientific knowledge as limiting parameters for a more exacting inquiry.
The problem with trying to produce a theory of history that converges on a realistic degree of complexity that reflects the complexity of actual history, is that the theory itself becomes overburdened with an excessive number of concepts and excessively complex concepts. Let me give a particular example. It is a commonplace (and as often disputed) that the histories of civilizations can be divided into an organic sequence. From a formal point of view, it doesn’t really matter if we break the life history of a civilization into three stages or five stages of some other number, except that some breakdowns of the life history of civilizations will be more intuitive than others. So I prefer five stages: origins, development, maturity, decline, and extinction. I think Spengler also employed five stages; ultimately, as I noted, the number of stages doesn’t really matter.
Suppose we divide the histories of all civilizations into five stages. Further suppose that there are two civilizations: these two civilizations could impinge upon each other (strictly in a temporal sense) in twenty-five distinct ways, from two civilizations impinging upon each other at their origins, to two civilizations impinging upon each other at their extinction. If three civilizations are present, there are one hundred twenty-five ways in which these civilizations might impinge upon each other in time. If we also distinguish the modality of their impinging on each other — say, distinguishing cooperation, competition, and conflict, or three modes of contact — then two civilizations might interact in seventy-five different ways, and three civilizations might interact in three hundred seventy-five different ways. If we reduce our stages from five to three, then two civilizations can impinge upon each other in nine ways, and three civilizations in twenty-seven ways, which somewhat simplifies matters, but probably not enough to constitute an intuitively satisfying abstraction for schematizing the possible interactions among civilizations.
We see here that simple abstractions can easily give rise to levels of complexity that approximate actual history, which is satisfying because this is what we want a scientific theory to do for us, but the extrapolation and elaboration of these simple taxonomies rapidly become unwieldy as they approximate the complexity of actual history.
Human history is supremely complex. It is more complex than natural history, whether the natural history of physics, cosmology, geology, or biology, because all of these factors are involved in human history, as well as further human factors that make the mixture more complicated. Abstract conceptions of human history that draw primarily on, say, biology, are useful up to a point, but we can see that such models need to be supplemented by the additional complexity of human beings who introduce large social groups, spoken and written language, advanced technologies, and other novel factors that a purely biological account would abstract from.
Even though the history of the universe as understood by cosmology is less complex than human history understood in traditionally humanistic terms (this in itself is another over-simplification, because the history of the universe includes human history, and the histories of any other intelligent species that might exist, or alternative forms of emergent complexity that have equally complex histories, albeit radically distinct from human histories), I argued above that, the further we go beyond anthropocentric norms, the less certain our predictions and retrodictions in cosmology are. With the greater complexity of human history, the prediction and retrodiction walls are correspondingly closer to the present, and block out the greater part of our history and our future.
In a future of cognitive enhancement, human beings might be able to hold much more complex abstractions in mind and find them to be a satisfying model of some phenomenon, overcoming some of the limitations we experience today as a function of human, all-too-human limitations. In the meantime, attempting to understand human history and civilization should be recognized for the paradox it presents: the more adequate our model, the further it departs from the tacit intuitive limits that we set on an informative theory.
One might well produce an adequate theory of civilization or of history, but the result is so complex that no one is willing to sit through an exposition of it, and no one finds it to be a helpful way to go about understanding the world, much less a helpful guide in living life or making political decisions. In this case, the theoretician of history must content themselves with some bastardization of their theory that they believe will get some traction with an audience, and the further this bastardization is pushed, the more it approximates the disappointments of Hegel, Spengler, and Toynbee. It would be nice to have a snappy and memorable name for this paradox, like the paradox of scientific history.