International Political Theory and Its Critics

Critique of the privileged position occupied by theoretical research in IR studies acquired traction at the turn of the 2000s-2010s. It received the backing of influential works and a growing number of followers. The most radical criticism arose from keen interest in forecasting. Its proponents object not only to macro traditions known as “-isms,” but also other types of explanatory work (including ‘theories of the middle range’). Conflict studies became an incubator of sorts, hatching that anti-theoretical movement. They have witnessed intense accumulation of vast statistical data and introduction of quantitative methods since the 1960s.

Building upon the extensive empirical testing, researchers concluded that theoretical explanations have limited worth in predicting future events. Thus, a growing number of scholars suggest an alternative approach to the development of the discipline relying on three propositions. First, correct prediction is the indispensable validation of scientific truth. Second, skepticism expressed by certain scholars about the ability to predict reveals denunciation of the falsifiability of theories and thus, following Karl Popper’s principle, their unscientific nature. Third, explanatory theories do not comprise a necessary condition for the forecasting of international events. Instead, machine learning provides tools for better predictions without explanations.

Russian academic community express growing interest in machine learning in IR studies. However, the discussion about the methods is disconnected from the debate about the role of theoretical knowledge, the relationship between the improvement in explanations, and the development of analytical tools. The task of advancing forecasting does not appear high on the agenda of Russian scholars.

“There is nothing more practical than a good theory.” [1] This attitude defined academic inquiry into international relations since the mid-twentieth century. Faculty surveys as well as bibliometric indicators suggest that generalized explanations provide the greatest recognition for a scholar within the community. Theorists top the lists of researchers producing the greatest impact on the scholarly debate, and their publications are quoted most frequently than empirical works.

Still, a fierce debate shook the leading academic journal in the 2000s, as the privileged position of theoretical research came under doubt. Critics claim that research should aspire not so much explanations for events that have already happened, but predictions of what may happen in the future. Their rhetoric reflects a pivot from a deductive-nomological model of knowledge to a renewed version of inductivism, bolstered by increased capacities for data-analysis.

Russian academic community achieved major progress in acquisition of theoretical research in recent decades. Moreover, it demonstrated increasing (even if somewhat belatedly) interest in sophisticated methods of empirical research. Nevertheless, the intensified debates on the purpose of IR studies so far remained foreign for Russian scholars. This publication aims to instigate domestic debate by introducing existing approaches to the issue.

Towards a new “great debate”

Alexander Chvorostov:
What Is Democracy and How It Can Be Measured?

The established chronology of the International Relations theory revolves largely around the so-called ‘Great Debates.’ Their role in the evolution of the discipline should be taken with a grain of salt. Commentators differ in describing their timing, participants, and the very subject, leading to the curious phenomenon of “debates over debates”. Nevertheless, the notion of “Great Debates” reflects major points of rethinking of IR studies, and revision of what constitutes a good theory.

The first debate centered on the relationship between normative and positive theory; the second focused on the clash between scientistic and metaphysical approaches to the study of international politics. The third emerged from reflexivist critique of the previously dominant rationalism and positivism. Thus, during every ‘Great debate,’ standards of what is a fruitful theory came under substantial revision.

Therefore, they represented a sort of “scientific revolutions”, suggested by Thomas Kuhn. [2] In the 2010s, it became fashionable to declare the end of the “Great Debates” abandoning any challenges to metatheoretical foundations of the discipline. The idea of their scientific futility led to the growing acceptance of theoretical relativism associated with the transition of IR studies to the stage of “normal science” (as Kuhn understood it).

Meanwhile, this period witnessed the unfolding of the newest “great controversy,” equal in importance (or perhaps superior in what was really at stake) to the aforementioned debates. It put under question not just various types of theorizing, but the scientific significance of the theory as such. It proceeded from claims that the IR theory (and maybe even theory in general) is anti-scientific and simply harmful to the advancement of knowledge.

Opposing the “-isms”

Criticism of the privileged position occupied by the theoretical research in IR received traction at the turn of the 2010s. Later, it acquired the backing of influential works and a growing number of followers. The attacks on theory came in two waves and from two angles, of which the first was moderate and the second became much more radical.

The moderate critics mostly opposed the so-called “-isms”. They denied the need to advance macro-narratives with their endeavor to provide a universal theory explaining international relations as a whole. Representatives of this approach argued that fixation on the development of a universal framework defining the general logic of international relations limited the explanatory power of the discipline. They contrasted that focus with a more pragmatic approach pursuing solution of specific puzzles.

Contrary to the traditional schools (realism, constructivism, liberalism, Marxism), moderate critics called for advancing ‘theories of middle range.’ The most famous exponents of this approach were Rudra Sil and Peter Katzenstein, who advanced the notion of ‘analytic eclecticism.’ They suggested to focus on specific issues and then select and combine theories in a way to find best possible explanations, even if theoretical approaches in different issue areas do not dovetail.

Anton Kolonin:
What’s Your Power, Strong AI?

The strongest response to this moderate revisionism came from John Mearsheimer and Stephen Walt, who emphasized the costs of the denial to fit individual studies into a broader theoretical framework. Primarily, they pointed to the compromised possibility of knowledge accumulation (i.e., in essence, they reproduced Hume’s critique of naive empiricism). Stressing the importance of theory, they emphasized the essential role of explanation in acquiring ability to forecast.

Rehabilitation of forecasting

It is the keen interest in forecasting that gives rise to radical criticism of theoretical research. It covers not only macro-approaches known as “-isms,” but also theory as such (including mid-level theorizing). Conflict studies became an incubator of sorts, hatching that anti-theoretical movement. They have witnessed intense accumulation of vast statistical data and introduction of quantitative methods since the 1960s.

Building upon the extensive empirical testing, researchers concluded that theoretical explanations have limited worth in predicting future events. For example, Skyler Cranmer and Bruce Desmarais ascertained that, despite the extensive record in studying the causes of interstate conflicts, the predictive accuracy of models built on existing theoretical explanations does not exceed 7%. Thus, using the theories available, it is possible to predict the onset or absence of conflict initiation in less than one in ten cases.

Moreover, theorists repeatedly questioned the very possibility of predicting international relations. One of the founders of the discipline Hans Morgenthau expressed doubts in this regard in his Scientific Man Versus Power Politics: he justified them by reference to incomplete information and human free will. [3] One of the most authoritative statements of similar thoughts appeared in the joint article by Steven Bernstein, Richard Ned Lebow, Janice Gross Stein and Steven Weber under the provocative title God Gave Physics the Easy Problems.

Nevertheless, a growing number of scholars are not satisfied with the deep-rooted fatalism of theoreticians. They suggest an alternative approach to the development of the discipline relying on three propositions. First, correct prediction is the indispensable validation of scientific truth. Second, skepticism expressed by certain scholars about the ability to predict reveals denunciation of the falsifiability of theories and thus, following Karl Popper’s principle, their unscientific nature. Third, explanatory theories do not comprise a necessary condition for the forecasting of international events. Instead, machine learning provides tools for better predictions without explanations.

Back to data?

Andrey Kortunov:
Vodka or Cognac: Two Tastes of Global Politics

The leadership in the critique of theory and advocacy of predictive research in the mid-2010s belonged to Michael Ward. He and his followers claimed that so far researchers have focused too much on explanation, rather on attempts to predict future events. According to Ward, the introduction of machine learning tools, combined with the accumulation of data sets, enables predictive estimates of high accuracy even without a supportive theory.

Currently, software algorithms that do not rely on predetermined patterns, but extract correlations from a large amount of data instead, became widespread. They can draw conclusions about new information on the basis of multiple analogies with previously processed data. In fact, they follow the path of inductive reasoning from empirical experience, providing forecasting assessments without deliberate explanations. Such algorithms do not provide justification for the assessments they make.

The common example of machine learning refer to the algorithm which after examining multiple images of cats start to identify similar animals in the subsequent pictures. Another prominent application of machine learning provides voice assistants, such as Alisa, created by Yandex, which steadily improves at answering requests of the clients. Scholars claim that similar algorithms are capable of predicting conflicts, protests or negotiations. These proponents of machine learning insist on their incorporation in International Relations studies. Starting from the 2010s, the number of articles with predictive assessments based on machine learning in places like Journal of Conflict Resolution proliferate by the year.

Nevertheless, the lack of explanation limits the practical value of predictions, despite the arguments put forward by proponents of machine learning. The refusal to parse the patterns that shape the international events weakens the ability of practitioners to respond to emerging risks or to seize opportunities, even in case of correct assessments about future events. To consciously impact events, it is necessary to understand their logic – this is where theory steps in.

Provisionary results

In assessing the prospects of the unfolding discussion, it is worth recalling that the previous rounds of “Great Debates” rarely ended in a clear victory for one of the sides. Usually, they developed in accordance with the Hegelian dialectics. As time passed, the radicalism of the disputants faded, and the subsequent generation of academics sought to synthesize elements of initially irreconcilable approaches.

The preconditions for similar dynamics also emerge in the field of conflict studies. For example, the Journal of Conflict Resolution published a series of correspondence articles in 2020–2021 including the one co-authored by Ward. In this piece, he and his colleagues accepted that theoretical explanations are of interest insofar as they provide better predictive results than analyses based on pure induction. That is, they moderated their skepticism toward theorizing without abandoning forecasting as a criterion for validating scientific knowledge.

In conclusion, it should be noted that there is a growing interest within Russian academic community towards machine learning in IR studies. At the same time, the discussion about the methods remains disconnected from the debate about the role of theoretical knowledge, as well as the relationship between the advancement of explanations, and the development of analytical tools. So far, the task of promoting forecasting does not appear high on the agenda of Russian scholars. However, this situation slowly starts to shift.

This material is a revised version of the report presented at the Russian International Studies Association's Convention on October 14, 2022.

1. Lewin K. Field theory in social science: Selected theoretical papers by Kurt Lewin. London: Tavistock, 1952. p. 169.

2. Kuhn T. The Structure of Scientific Revolutions.

3. Morgenthau H. J. Scientific man vs. power politics. London: Latimer House, 1947. p. 120.