What Is a Law of Politics?

In the introduction to his original article, Cuzán (Reference Cuzán2015, 415) clarified what he meant by a law of politics: “By a ‘law’ of politics, I mean an invariant or almost invariant empirical regularity that is descriptive of intrinsic properties of politics and the state.” I adopt this definition and agree with Cuzán (Reference Cuzán2015, 415) that political scientists should not refrain from claiming that there are laws of politics in this sense.

An important feature of this definition is that laws are a subset of empirical regularities. Some philosophers of science (e.g., Cohen and Callender Reference Cohen and Callender2009) proposed to call the principles that best systematize the knowledge in a given discipline “laws.” This is a piecemeal approach to laws that states that every scientific discipline (not only physics) can have laws. Cohen and Callender (Reference Cohen and Callender2009, 24) suggested that there are laws (in this sense) in not only, for example, high-energy particle physics and fluid dynamics but also botany, ecology, and economics. This concept of law does not seem useful in political science: it is unclear whether political science can be systematized in any meaningful way. Doing so would require that all important knowledge in the domain is derived from a few theoretical principles. However, what is crucial is that laws—for both me and Cuzán—are empirical generalizations, not axioms in a deductive system, as Cohen and Callender (Reference Cohen and Callender2009) proposed. In other words, we consider empirical laws, not theoretical laws (e.g., Newton’s laws of motion).

The second important feature of the definition is that it does not require laws to be deterministic: laws are invariant or almost invariant empirical generalizations. It has been emphasized in both the general philosophy of science and the philosophy of the social sciences that laws can have exceptions. For instance, Mitchell (2001, 248–49) explicitly stated that her pragmatic account of scientific laws allows them to have exceptions. She provided Mendel’s laws in genetics as an example. In the philosophy of the social sciences, Goertz (Reference Goertz and Kincaid2012, 103) defended the view that the canonical form of laws in the social sciences is “(Almost) all/none A are B.” The idea is that in each discipline, some empirical regularities of this form are singled out as special and thought to be deserving of the label “law” (Goertz Reference Goertz and Kincaid2012, 98).

The third important feature of Cuzán’s definition is “intrinsic property.” Cuzán did not clarify this notion. Near the conclusion of his original article, he stated that laws describe “properties and patterns inherent in the very nature of politics in the state” (Cuzán Reference Cuzán2015, 418).

Empirical Evidence for the Five Laws

In Cuzán’s original article, four of the laws are presented as spanning democracies and dictatorships; the fifth law is believed to set the boundaries between the two (Cuzán Reference Cuzán2015, 415). In his follow-up article, the laws are said to govern elections in democracies. The autocratic regimes disappear from the analysis. As previously mentioned, the democratic part of the 2015 evidence base contained data from 426 elections in 23 democracies. The original dataset also included more than 100 controlled “elections” in 15 autocratic regimes. Cuzán was confident about his data: “The regimes hail from large and small countries and different times and cultures. Thus, there is sufficient variation in the data to enable one to generalize with confidence” (Cuzán Reference Cuzán2015, 415).

This confidence is not unfounded, especially when we consider the enlarged 2019 evidence base. As for geographic distribution, Cuzán analyzed election results from Africa (i.e., Botswana and Namibia); America (i.e., Canada, the Caribbean, Latin America, and the United States); Asia (i.e., India and Japan); Europe (i.e., the developed democracies of Western Europe as well as post-communist Europe); and the “antipodes” (i.e., Australia and New Zealand). Concerning time, some data date from the nineteenth century (i.e., Canada and the United States), other data to the first half of the twentieth century (i.e., Australia, New Zealand, Sweden, and the United Kingdom). In general, two decades of free elections were required for inclusion in the evidence base. As previously mentioned, the countries included both presidential and parliamentary democracies.

Spatiotemporal Stability

In the general philosophy of science, the idea of “law” often is connected to the notion of spatiotemporal stability. Empirical regularities can have more or less temporal stability and more or less spatial stability. A well-known example from the philosophical literature on scientific laws illustrates this (see, e.g., Mitchell Reference Mitchell2000, 246). Consider the following claims:

1. (G) All gold spheres on Earth have a diameter of less than 10 meters.

2. (U) All uranium spheres on Earth have a diameter of less than 10 meters.

Both claims are true now; however, there is an important difference. The critical mass for enriched uranium is only a few kilograms. Therefore, we cannot create a uranium sphere with a diameter of 10 meters because there is a theoretical limit to how big uranium spheres can be, given their instability. This means that we have good theoretical reasons to believe that (U) will be true in the future and has been true in the past for a long time. In other words, we have reason to believe that (U) has high temporal stability.

This contrasts with (G). This claim is true because it happens to be the case that no one produced such a sphere until now. There is sufficient gold in the world and there is no reason to believe that the sphere would explode. Therefore, claim (G) should be attributed a lower temporal stability; that is, it is true now but it very well may be false in one month, in one year, or in a thousand years.

The same line of reasoning can be developed with respect to the spatial regions in which a claim is valid. Some claims are considered to be valid in the entire universe and some in more restricted regions (e.g., on the entire Earth or on parts of it). This means that in addition to degrees of temporal stability—as illustrated in (G) and (U)—we can distinguish degrees of spatial stability.

In the general philosophy of science, the idea of “intrinsic” or “inherent” properties is analyzed in terms of spatiotemporal stability. A regularity that describes intrinsic properties of a category of objects has considerable spatial stability (i.e., its validity is not limited to a small region) and also considerable temporal stability (i.e., it was true in the past and will remain true as long as certain conditions hold). These conditions are determined by the underlying mechanism that produces the regularity—which brings us to the second core concept.

Social Mechanisms

In the general philosophy of science, the standard definition of “mechanism” is “a mechanism for a phenomenon consists of entities (or parts) whose activities and interactions are organized so as to be responsible for the phenomenon” (Glennan and Illari Reference Glennan, Illari, Glennan and Illari2018, 2). This is a minimalistic characterization because not much is required for something to be a mechanism; many types can exist. The most important constraint is that mechanisms are compounds—that is, they can be decomposed into entities. Types of mechanisms can be identified by imposing constraints on the types of entities and activities. Social mechanisms are one such type. Steel (Reference Steel, Jarvie and Zamora-Bonilla2011, 298) characterized them as “Social mechanisms in particular are usually thought of as complexes of interactions among agents that underlie and account for macro-social regularities.” In this definition, agents are not only individual persons. For instance, corporations, government bureaus, and charitable organizations can be perceived as agents because they are coordinated groups of individuals with common objectives (Steel Reference Steel, Jarvie and Zamora-Bonilla2011, 298). Social mechanisms usually involve categorization of agents, and the behavior of agents is determined at least partly by the social roles that they occupy (Steel Reference Steel, Jarvie and Zamora-Bonilla2011, 299).

A Regulative Principle

My view is that all disciplines in the social sciences should adopt the following regulative principle with respect to the concept of “law”:

1. 1. As a general rule, call empirical regularities of a given discipline that are spatiotemporally stable in a relative sense (i.e., compared to the bulk of regularities in that discipline) “laws” as a mark of their special status within the discipline.

2. 2. To decide which theses deserve this special status, assess their spatiotemporal stability by inquiring into the social mechanisms that are responsible for the relationship described in the claim.

This regulative principle creates an intimate link between (social) laws and (social) mechanisms. First, laws ontologically depend on mechanisms: relatively stable relationships at the macro level are produced by relatively stable underlying social mechanisms. Second, there is epistemological dependence—that is, we cannot judge whether a claim deserves to be called a law if we do not have sufficient knowledge about the underlying social mechanisms.

Before I apply my account to Cuzán’s proposals, I provide an example of how it could work in the case of Duverger’s law. This example serves as a contrast: whereas I think that we must suspend judgment with respect to the five alleged laws that Cuzán proposes, I believe there are good reasons for calling Duverger’s law a “law of politics.”

Inconclusive Arguments

Cuzán paid attention to spatial variation in data: wide geographical distribution was important for him, and his data for some of the countries date from the nineteenth century. However, spatial and temporal variation is not the same as spatial and temporal stability. Two crucial questions to be answered are as follows:

• • What reasons do we have to assume that if a new democratic regime is founded somewhere, its elections will be governed by the same regularities as the elections analyzed by Cuzán?

• • What reasons do we have to assume that the trends that have been observed in the distant past (for Canada and the United States) and the more recent past (for all countries) also will manifest in the future?

These questions are not answered in Cuzán’s articles. Therefore, there is no argument for the spatial or temporal stability of the alleged laws. Cuzán did not investigate or describe the underlying mechanisms in detail; there are only a few hints. In the discussion of Law #2 (i.e., incumbent advantage), he wrote that “the selectorate may be somewhat biased in favor of the ‘devil’ they know” (Cuzán Reference Cuzán2015, 416). The conclusion of the follow-up article suggested that the laws may be “a function of mechanical or statistical features of democratic institutions” (Cuzán Reference Cuzán2019, 459). A good argument for or against considering the five theses as laws can be developed only by substantially elaborating on these hints.

Implications

Because the arguments are inconclusive, we must suspend judgment. The issue can be settled only by an extensive investigation of potential underlying mechanisms. Cuzán referred to a few papers that contain brief descriptions of mechanisms that could be elaborated and investigated further. For instance, Stokes and Iversen (Reference Stokes and Iversen1962, 159) suggested that interest groups have a biased memory (i.e., favors granted are less well remembered than favors not granted). Another mechanism they suggested is that parties that are not in power can make extravagant promises, whereas parties that are in power are bound by what they can deliver (Stokes and Iversen Reference Stokes and Iversen1962, 159). These may be mechanisms that are relevant for the third regularity. Lebo and Norpoth (Reference Lebo and Norpoth2007, 72) suggested a different mechanism: a party that is in power is bound to make mistakes and disappoint its supporters. These brief descriptions are avenues that should be investigated to make a final judgment on the status of Cuzán’s laws.

Popular Cases

My view on Duverger is that his three propositions deserve to be called laws. My argument for this, summarized previously herein, is presented in Weber and De Bal (Reference Weber and Bal2018). As a consequence, I disagree with Benoit (Reference Benoit2006, 76), who wrote that Duverger’s law “retains the label of ‘law’ mainly through force of habit.” I believe there are good epistemological reasons for calling Duverger’s propositions laws.

I also believe that there are good epistemological reasons for agreeing with Davenport (Reference Davenport2007) that there is a law of coercive responsiveness, which is the main example in Weber et al. (Reference Weber, Makhnev, Leuridan, Barman and Coninck2022). We argued that if our regulative principle is applied, a positive argument results.

For other interesting cases, I suspend judgment because I have not applied the regulative principle to them. The first case is the democratic peace proposition. Levy (Reference Levy1994, 352) famously wrote:

In future research, I plan to scrutinize this “confidence” through my account. Another interesting case is the so-called iron law of oligarchy, which states that organizations—after some time—inevitably are ruled by a small elite group. This thesis was originally presented by Michels in 1911 and in German was translated as Michels (Reference Michels1915). My third case is the principle of balance of power, stated as follows:

An interesting starting point is Levy and Thompson (Reference Levy and Thompson2010), in which they argued that this principle is not an “iron law.” To summarize, these are at least three examples that I want to investigate in future research.

Laws, Mechanisms, and Explanation

Because of the ontological and epistemological dependence of laws on mechanisms, my account can be called a “mechanistic conception of laws.” Many social scientists (e.g., defenders of mixed-methods research) emphasize that knowledge of causal mechanisms has an explanatory role: it is important because it provides insight into how causal relationships at a higher level are produced. I agree with this and add an element to the global picture: underlying mechanisms can be more or less stable, which warrants singling out some claims as laws. This also is what my account adds to the idea of evidential pluralism—in medicine and in the social sciences—as it is elaborated and defended in Shan and Williamson (Reference Shan and Williamson2021).