The Cognitive Complexity of Ideologies and the Ambitious Aspirations of Ideologists

Abstract

Some ideologies, such as liberalism, conservatism, and socialism, are complex symbolic structures. Mastering them requires specialization, and because we are all amateurs in almost all symbolically rich domains, most people are not ideologically sophisticated. Instead, they reason about politics in a maturationally natural way, via friend-foe representations and inferences based on those representations (for example, friends of foes are foes). However, even complex ideologies are much simpler than the political, economic, and social systems that they are supposed to represent. Hence, all ideologies are inaccurate to varying degrees. More subtly, all are incomplete in various ways; in particular, they fail to anticipate some crucial events (for example, global warming), which leads to unanticipated value trade-offs and strategic conundrums. Ideologists sometimes adapt to incompleteness via recombinant innovation, producing hybrid ideologies (for example, ecosocialism). In turn, this tends to produce inconsistencies between new and old parts of an ideology. Thus, inaccuracy, incompleteness, and inconsistency are not pathologies of political belief systems. They are the inevitable result of ideologists grappling with a reality that is far more complex than their symbolic constructions can be. Therefore, evaluations of ideologies that identify errors, incompleteness, or inconsistency at a single point in time are often unenlightening. Following Imre Lakatos, evaluations should focus on how a sequence of ideologies—an ideological tradition—evolves.

Introduction

Many ideologies are complex symbolic structures. Because scholars’ work involves producing and comprehending symbols, we tend to underestimate the complexity of the symbolic structures that we call ideologies. Despite the persistent efforts of Philip Converse ¹ and his colleagues, political scientists took decades to accept the fact that most citizens are much less familiar with liberal or conservative ideologies than are most political scientists or to accept that ideological reasoning is difficult for most people. We forget how long it has taken us to learn what we know about politics and how long it has taken us to think ideologically.

This is true of specialists in other complex belief systems. For example, Pascal Boyer, a cognitive scientist who has done important research on religious thinking, remarks that few people in small societies think about religion in sophisticated ways. Indeed, “most frustratingly for anthropologists, they generally seemed to have no established set of religious beliefs.” ² Psychologists call this phenomenon—when those who have mastered some body of knowledge or belief have forgotten what it is like to be a novice—the “curse of knowledge.” ³ This makes social scientists, historians, and philosophers underestimate how long it takes to understand Karl Marx or John Rawls and to underestimate the mental achievement of advanced ideological reasoning.

Yet the complexity of ideologies pales beside the complexity of the political, economic, and social systems that ideologies represent. For a species that throughout most of its evolutionary history lived in bands of twenty to one hundred people, contemporary societies are orders of magnitude more complex. Indeed, they are fantastically complicated.

Hence, we should take as axiomatic the proposition that all ideologies are inaccurate in some ways (for example, some of their forecasts are incorrect). More subtly, even the most sophisticated ideologies will encounter new and unanticipated problems. Thus, they are seriously incomplete: they have holes or gaps. For example, before 1950 no major Western ideology—not socialism, conservatism, or liberalism—anticipated climate change.

This is not meant as a criticism. Inaccuracy and incompleteness are inevitable when humans tackle collective problems of staggering complexity. ⁴ And for people who specialize in politics—such as politicians, organizers, scholars, and journalists—there is no real alternative to ideologies. Decades of research in cognitive psychology have made it clear that humans are natural-born theorists. ⁵ As individuals, we need mental models and a large library of cognitive frameworks to cope with the information-rich environments we have created and live in. ⁶ Collectively, there are good theoretical and empirical reasons for believing that in order to organize political interests in large industrial democracies, shared action-oriented belief systems—that is, ideologies—are required. Recommending that politicians, organizers, and other political professionals give up ideologies is similar to recommending that scientists give up theories; both proposals are utopian. Ideologies are here to stay. ⁷

Hence, rather than abandoning them, when ideologists encounter unanticipated problems, they usually try to adapt them to the new circumstances. This tends to make ideologies still more complex, which fosters inconsistencies between different parts of a belief system. (This problem occurs, for example, in common law systems.)

Thus, the main theses of this essay are as follows. First, many ideologies are cognitively complex; mastering them requires specialized education. Hence, few people reason in ideologically sophisticated ways. Second, even complex ideologies are much simpler than the politicoeconomic systems that they are supposed to represent. Hence, all ideologies are incomplete as well as inaccurate. Third, adapting to unexpected circumstances makes ideologies yet more complex, fostering internal inconsistencies.

This terse argument can be unpacked as follows.

1. (1) Although thinking ideologically is a mental capacity that most adults can acquire, it is not a maturationally natural skill such as understanding and speaking one’s native language. ⁸

2. (2) Therefore, in some environments only a few people will learn to think ideologically; in most environments, only a few specialists use advanced ideological reasoning.

3. (3) As a widespread sociocultural phenomenon, even rudimentary ideological reasoning is fairly recent, going back only about 250 years. (A few specialists, for example, Plato, started much earlier.) In contrast, friend-foe thinking is ancient. ⁹

4. (4) Ideological thinking draws on a wide array of cognitive capacities, including problem-solving, rule-following, planning, and causal reasoning. No one dedicated cognitive module is responsible for ideological reasoning, though specific modules (for example, harm-based theories of morality) can be recruited as needed.

5. (5) Consequently, from a psychocultural perspective little about ideological reasoning is fundamentally mysterious. Although certain ideologies may be quite complex, they are typically built by normal psychocultural processes out of mostly preexisting elements.

6. (6) Ideological innovation therefore often involves combining familiar elements in a new way, referred to as “recombinant innovation.” ¹⁰

7. (7) Whether a particular combination of elements is internally consistent is an empirical question. Moreover, perceived coherence is not equivalent to logical consistency. An ideology’s adherents, both ordinary specialists and amateurs, may believe that it is consistent, even if hyper-specialized scrutiny (for example, Arrow’s Impossibility Theorem) proves that it is not.

8. (8) All ideologies are in varying degrees both inaccurate in that some empirical claims are incorrect and incomplete in that certain significant events and problems are completely unanticipated. An important class of unanticipated problems involves trade-offs among important values.

9. (9) Making an ideology more complete tends to produce inconsistencies between new ideas, which are added to address unanticipated problems, and old parts of the ideology.

10. (10) Because ideologies are action-oriented belief systems infused by values, motivated reasoning ¹¹ is common. Adherents—specialists as well as amateurs—often protect core ideas from falsification or from efforts that identify inconsistencies within an ideology. This protection is frequently enforced by loyalty tests based on friend-foe reasoning.

11. (11) Some ideologies are more difficult than others for motivated reasoning to protect. For example, those promising benefits in this world are more vulnerable to falsification than are those promising benefits only in the next.

12. (12) Ideologies that focus on benefits in this world can become intertwined with processes of high cognitive complexity—such as science and mathematics—that are not maturationally natural.

13. (13) Because both friend-foe reasoning and religious thinking are maturationally natural, combining either of them with political ideologies will remain possible for the indefinite future, even when scholars regard such combinations as inconsistent or even incoherent.

14. (14) Because friend-foe reasoning represents political opponents as enemies, it can destabilize the ideology of liberal democracy.

This argument involves changing our scientific reference point. Instead of an idealized representation of a rational citizen, possibly inherited from Enlightenment doctrine, I propose an empirical benchmark: the cognitive capacities of normal human beings. ¹² These mental capacities create the action-oriented political belief systems that we call ideologies, which are symbolic structures that, despite the cleverness of their creators, have holes and inconsistencies ¹³ as well as overt mistakes.

What this essay is not

Although this essay will analyze the incompleteness and the inconsistencies of ideologies, this is not yet another essay about human irrationality. There already are too many such essays, often based on an idealized and therefore misleading scientific reference point regarding human cognition. To state the point affirmatively, homo sapiens is undoubtedly the most intelligent living species and we are probably the most intelligent species ever to have inhabited this planet. Nevertheless, we are cognitively constrained in many important ways. In particular, at a conscious level we process information serially. ¹⁴ This sharply limits how fast we can learn new subjects; students can read only one book or article at a time. This property is not a kind of irrationality; it is a computational constraint of our minds.

That our conscious processing of information is serial has important implications for our understanding of ideology. As noted, this constraint places sharp limits on our speed of learning. Consequently, in contemporary societies, mastery of an informationally intensive domain—for example, accounting, aerospace engineering, astrophysics, and so on through the alphabet—is restricted to trained specialists. Mastering such domains is not maturationally natural.

Thus, the following powerful empirical regularity holds. Today every person, in every country, is an amateur in almost all symbolically rich domains. There are no exceptions to this pattern. None. This is not an indication of irrationality. It simply reflects a few basic psychological, cultural, and biological facts. (1) It takes between five and ten years to become a competent specialist in any informationally rich domain, that is, one that has many representations and methods for manipulating those representations. (2) Thanks to cumulative cultural evolution, ¹⁵ modern societies have many such domains. ¹⁶ (3) Given our relatively short lifespans and given facts (1) and (2), every person remains an amateur in almost all such domains throughout their adulthoods.

Consequently, superficial knowledge of ideologies and superficial ability to reason ideologically, as Converse and many others ¹⁷ find characterizes most American voters, tell us nothing new about the underlying cognitive capacities of human beings. Nor do we need to posit the existence of any mental properties in order to explain this superficiality. It is not a sign of irrationality or even of a “bias” in the sense used in the behavioral study of judgment and choice. It is just what should be expected of amateurs and, given the underlying representational and computational complexities of many ideologies, it is to be expected that most people are amateurs in this domain.

The rest of this essay is organized as follows. The first task is to examine nonideological thinking. This is important because many students of ideology seem to believe that all thinking about politics involves ideology. As stated above, I believe this is an empirically incorrect position that hinders our understanding of the cognitive complexity of ideological reasoning. The section on “If most people aren’t ideologically adept, how do they think about politics?” examines our default mode of nonideological thinking: friend-foe representations (claims 2, 10, and 13).

I then return to analyzing ideological thinking. The section on “Different kinds of incompleteness and inconsistency in ideologies” provides a brief conceptual analysis of different kinds of incompleteness and inconsistencies in ideologies and examines a few of the relations between these two properties. A heuristic that can reduce incompleteness is analyzed in the section “Incompleteness, recombinant innovation, and hybrid ideologies.” The heuristic of recombinant innovation can be used when different ideologies solve different political, social, or economic problems and combining them creates a hybrid (for example, ecoconservatism) that is less incomplete (claims 4–8). Claim 9, however, hypothesizes that this method of reducing incompleteness often produces new inconsistencies. The “Inconsistency in ideologies” section focuses on inconsistencies in ideologies. In the “Criticizing and evaluating ideologies” section, I analyze an ideology, populism, that is especially likely to being influenced by friend-foe reasoning and which is therefore dangerous to liberal democracy (claim 14). I then offer some concluding thoughts.

If most people aren’t ideologically adept, how do they think about politics?

Difficult as it is for scholars who study politics to grasp, there are nonideological ways of thinking about politics. One of these is maturationally natural; it involves friend-foe mental representations and operations (computations) on those representations. The following fleshes out the content of such representations:

Together, the egocentric and social aspects of identity suggest that most people mentally represent themselves as the salient central node in a personal social network … [and they] can automatically use this representation to structure their understanding of a wide variety of political settings…. Furthermore, once this type of representation has been evoked, one can carry out specific mental computations that facilitate the relevant setting. One can form coalitions (e.g., “in building my coalition I should look to my friends first”), associate with a political party (e.g., “my parents voted for this party, so the party is probably good for me”), and quickly estimate the value of unfamiliar policy proposals (e.g., “my enemies endorse this plan, so it’s probably bad”). ¹⁸

Since friend-foe representations (FFRs) are maturationally natural, people find it easy to create ideologies that incorporate an FFR component or a straightforward modification of an FFR component. Nationalism and X, where X may equal communism or Islamism, is such a class. ¹⁹

Before proceeding further, I should explain why I do not categorize FFRs as ideologies. I have already mentioned one cluster of reasons: how FFRs differ from prototypical ideologies. Recall that, spatially, FFRs are everywhere; temporally, they are historically old; developmentally, they are maturationally natural; psychoculturally, literacy is unnecessary. ²⁰ Prototypical ideologies have none of these properties.

To this set of differences, I add one more. It is perhaps the most important, as it may help to explain the others. The claim is that whereas FFRs are typically egocentric representations, prototypical ideologies are not. I mean egocentric in the morally neutral sense of the word used in research on spatial cognition. ²¹ In an egocentric spatial representation, the objects in my mental image are depicted in relation to myself. I close my eyes and I can “see” my desk in front of me and my bookcase behind me. I am an indispensable part of this mental representation.

A good example of nonegocentric spatial representations are conventional maps. A map of the American West shows the relation between Tucson and San Francisco without reference to my home in Palo Alto. In textual representations, latitude-longitude descriptions are nonegocentric. Spatial cognition researchers refer to nonegocentric representations as allocentric ones. ²²

FFRs are egocentric almost of necessity, for I have in mind my friends and my enemies. That much is clear. But why do prototypical ideologies tend to be allocentric? It is partly a matter of scale. Ideologies are belief systems designed to answer questions—for example, “How should a society be organized?”—that involve enormous numbers of people, hundreds of millions in some cases. As Benedict Anderson emphasizes regarding nationalism, systems of this scale cannot be seen; they must be imagined. ²³ For such objects the self is not a useful mental landmark. Something outside the self—language, place of birth, membership in an economic class—is likely to be better at organizing the mental landscape. The result is allocentric mental representations. ²⁴

The hypotheses that FFRs are egocentric and that most ideologies are allocentric, combined with the empirical regularity that humans find egocentric reasoning easier than allocentric, help to explain why FFRs preceded ideological thinking. This historically grounded analysis helps us move away from teleological theories of political reasoning. It is causally misleading to start with a reference point of how ideological specialists, whether politicians or scholars, think and to investigate how and why most people fall short of those standards. That teleological approach gets things backwards. It would be better to ask how allocentric ideological reasoning evolved, given the common starting point of egocentric friend-foe thinking.

Although allocentric belief systems tend to be more cognitively sophisticated than egocentric ones, they are not free of problems. We take up two of these next.

Different kinds of incompleteness and inconsistency in ideologies

Ideologies may be incomplete or inconsistent in different ways because they contain logically different kinds of statements. Virtually all ideologies contain at least three kinds of statements: (1) empirical claims about the world (for example, the hierarchical character of current political systems); (2) statements of value (for example, the importance of individual liberty or economic justice); and (3) injunctions, prohibitions, and recommendations (for example, one should obey one’s rulers or one should further the interests of the working class). Claims regarding an ideology’s completeness or its consistency can refer to any of the above kinds of statements, though the logic of these claims differs somewhat. For present purposes, I restrict the charge of “inaccuracy” to an ideology’s empirical claims.

Completeness is easiest to understand regarding the strategic logic of statements of type (3). Consider an algorithm that contains if-then statements: If the situation is x, do y. The algorithm is incomplete if it does not say what to do if the situation is not x. Similarly, an ideology’s strategy is incomplete if it is silent about what to do at an important historical juncture.

An ideology’s empirical content is incomplete if it says nothing about events (for example, climate change) that turn out to be important for its values and strategy. Thus, there are close links between incompleteness in an ideology’s strategy and in its empirical content.

An ideology’s strategy is inconsistent if in one place it says if x do y, but in another it says if x do w but y and w are not identical (though their actions may overlap). An ideology’s empirical content is inconsistent if in one part it predicts r and in another part it predicts not-r, or if one part predicts r and another part states that not-r is in fact the case. These two kinds of inconsistency are related; it may be advisable to do y if r occurs and to do w if s occurs, but if one cannot do both y and w, then adherents of the ideology do not know which action they should take.

An ideology’s value-content is inconsistent if it places high value on two values (for example, individual liberty and economic equality), but it turns out that achieving both to the level prescribed by the ideology is impossible: there are trade-offs. This inconsistency often generates a kind of incompleteness, as ideologies are frequently silent about how to resolve important value trade-offs.

Finally, an ideology can exhibit inconsistencies across the three different kinds of statements identified above. For example, its strategy may be at odds with its empirical content. The section below on “Inconsistency in ideologies” will analyze a well-known example of this complicated kind of inconsistency: tensions exhibited in the 1890s between the German socialist party’s practical ideology and several of Marx’s predictions that had been empirically disconfirmed.

Incompleteness, recombinant innovation, and hybrid ideologies

If ideologies are cultural tools for addressing difficult collective problems, then we can expect the use of basic problem-solving methods. A common heuristic, when facing an unanticipated problem, is to import several existing solutions to different sub-problems and combine them, that is, to engage in recombinant innovation.

For example, consider the invention of the armored tank in World War I. The problem—how an attacking force could reach the opponent’s front lines before being wiped out in the “killing zone” swept by artillery and machine guns—was not qualitatively new, but the lethality of modern weapons had intensified it enormously. A solution to the problem—the armored tank—was designed by importing ideas from four different domains. Protective steel armor was borrowed from warships; internal propulsion engines were borrowed from trucks; the tank’s main offensive weapon, its gun, was borrowed from artillery; most imaginatively, instead of wheels and tires, tank designers imported tread from farm tractors. Mix and match, et voilà, the armored tank! ²⁵

Coping with new and unanticipated problems is a fact of life for both military strategists and political ideologists. If solving a well-defined tactical problem such as getting offensive forces through a killing field induced decision-makers to import ideas from different domains, we should expect that solving an ill-defined problem—how societies should be organized and governed—would often prompt an open-ended search for solutions and components of solutions. Why impose ex ante limits on our imagination? Perhaps some old ideas (for example, religious ones) retain some relevance. Perhaps they can be combined, in surprising ways, with modern views about politics. Hence, we observe Christian socialism, Islamic nationalism, and religious Zionism. ²⁶

Incompleteness and hybrid ideologies revisited: Environmentalism

Environmental issues provide a useful example of how a new set of problems prompt the combining of general-purpose and special-purpose ideologies. None of the major European or American ideologies—classical liberalism, welfare-state liberalism, modern conservatism, or orthodox or revisionist socialism—anticipated environmental problems. That class of issues was not on their screens before 1960. In particular, no major ideology of the modern era anticipated global warming, perhaps the most dire problem facing humanity today.

This is a serious case of incompleteness, to say the least, but this is not meant as a criticism. As moral philosophers say, ought implies can. The failure to anticipate grave environmental problems is related to the growth of knowledge and, as Karl Popper argues, ²⁷ we cannot predict the growth of knowledge in detail; if we could, then we would have it today, not thirty years from now. When the burning of fossil fuels (primarily coal) took off in Britain in the mid-1700s, climate science was not even in its infancy.

Unlike conservatism or liberalism, however, environmentalism, especially its moderate versions, is a special-purpose ideology. ²⁸ Conservatism and liberalism are generally understood as belief systems that provide an overall perspective on how societies should be governed. It would be odd for an advocate of either to say that their ideology need not give guidance about a major political issue of the day. That obligation is not imposed on environmentalist ideologies. It is understood that they may have little to say about, say, national security policy. Such silence is acceptable for special-purpose ideologies.

Certain special-purpose and general-purpose ideologies can produce viable hybrids. This is the case with moderate versions of environmentalism and some versions of either conservatism or liberalism. Because I suspected that relatively few readers of this volume would be conservative, the more interesting case to consider pertains to the consistency of environmentalism and (certain versions of) conservatism. The argument is simple. The most rigorous part of free-market conservatism’s empirical core, hence the part where detecting inconsistency is easiest, is the microeconomic theory of competitive markets. ²⁹ It is consistent with that theory to tax negative externalities, for example, to impose a carbon tax. Indeed, the idea of taxing negative externalities goes back at least to Arthur Pigou, a well-known economist who wrote about this topic a century ago.

Thus, fixing this particular incompleteness does not produce any inconsistency in the hybrid. Moderate versions of environmentalism can be imported into market-based conservatism with surprisingly little strain. Also, because liberalism is more inclined to governmental regulation of markets, creating a hybrid green-liberal ideology is even easier. ³⁰

The existence of hybrid ideologies, especially those created by merging a special-purpose belief system and a preexisting general-purpose ideology, may help to answer a question about the dynamics of complex political belief systems. As I argued in the introduction above, ideologies that purport to provide more or less comprehensive guides to the organization and governance of societies are often highly complex symbolic entities. The evolution of complex systems can take a lot of time. How have ideologies as complex as ecoconservatism or ecoliberalism evolved so quickly? ³¹

In his pathbreaking work on the evolution of complex systems, Herbert Simon provides a hypothesis: complex systems evolve much faster if they are nearly decomposable. ³² From this perspective, a special-purpose ideology is a relatively stable symbolic module that can be plugged into a variety of general-purpose ideologies. Some combinations, of course, work better or are more stable than others.

Inconsistency in ideologies

Although the examples of green conservatism or green liberalism indicate that some hybrid ideologies are surprisingly coherent, it is easy to find examples of major inconsistencies. Given that “the core definition of ideology as a coherent and relatively stable set of beliefs or values has remained constant in political science over time,” ³³ such phenomena are a puzzle for many political scientists who study ideology.

Perhaps one could argue that coherence or consistency is a property of prototypical ideologies. But even for those belief systems much depends on how these words are defined. ³⁴ Probing investigations of, for example, democratic theory—such as Arrow’s Impossibility Theorem ³⁵—have revealed subtle inconsistencies in what are often considered to be coherent belief systems. Hence, I believe that it is methodologically unwise to make consistency a defining property of ‘ideology.’ Doing so may reflect an idealized view of our cognitive capacities; an underestimation of the complexity of the political, economic, and social systems that these belief systems putatively represent; and an underestimation of our imaginative and pragmatic problem-solving. We should not lightly assume that a desire for consistency always stifles the mixing and matching of ideologies. If one looks for hybrid ideologies with serious inconsistencies, one finds them quite quickly.

Indeed, one can find inconsistencies in most ideologies whether or not they are hybrid. I will discuss three important examples in the rest of this section. The first one, the inconsistency between property and freedom in American Revolutionary ideology, became well known to ordinary citizens during the debates over slavery in the first half of the nineteenth century. The second one, the inconsistencies in German socialism revealed by the debates between orthodox Marxists and revisionists in the late-nineteenth century, is somewhat less well-known. The third example, Arrow’s Impossibility Theorem, is known only to specialists.

Value inconsistencies and ideology in the American Revolution

Major inconsistencies can appear in an ideology’s value-content. Indeed, given the ambitious scope of general-purpose ideologies, it would be remarkable if major inconsistencies did not exist.

The conflict between freedom and property (in particular, the owning of persons under slavery) was one of the most (in)famous value-inconsistencies ever associated with an ideology based on liberalism and other Enlightenment ideals. The values of classical liberalism, especially the rights of individuals, were of major importance to those revolutionary ideologies. Within that set of values—specifically, in the rights attributed to individuals—there is a profound inconsistency between property rights and the right to life, liberty, and the pursuit of happiness. British opinion-leaders were quick to point this out. Samuel Johnson, for example, famously asked the scathing question “How is it that we hear the loudest yelps for liberty among the drivers of the negroes?” ³⁶ The strategic logic of any ideology that incorporated both property rights and the three rights articulated by the American Declaration of Independence were similarly inconsistent.

The U.S. Constitution did not resolve this inconsistency. Any proposal that would have significantly curtailed slavery would have been rejected by important states, notably Virginia. Hence the Founding Fathers and Framers knowingly kicked the can down the road. Some of them (famously, Thomas Jefferson) did so uneasily, but kick it they did.

If Garry Wills is correct, Abraham Lincoln attempted an ideological reorientation that would have greatly reduced the inconsistency, ³⁷ but that gambit had to be backed by the force of arms. And counterrevolution, the white terror in the post-Civil War South that restored white supremacy, greatly limited the impact of that ideological reorientation. Thus, the inconsistency in American belief systems endured.

Inconsistencies in German socialism

In the 1890s the largest socialist party in the world, the German Social Democratic Party (SPD), was roiled by an extended argument between orthodox Marxists, led by Karl Kautsky, and revisionists, led by Eduard Bernstein. The substance of the argument involved important claims about the world and related injunctions and recommendations. Bernstein asserted that important parts of the ideology’s empirical content was seriously inaccurate. Several key predictions of classical Marxism (for example, under capitalist systems the proletariat would become immiserated) had been falsified by well-known facts about German economic trends (German workers had become better off in the late 1800s). These errors, together with several other developments that classical Marxism had not predicted (for example, that socialists could win many parliamentary seats), led to strategic inconsistencies in the movement. Although orthodox Marxists continued to advocate revolution (or said they did; see below), revisionists recommended taking over the state peacefully through elections.

This was awkward, indeed. On the one hand, the theoretical predictions were major parts of the ideology’s empirical content and its claim to being (unlike many of its competitors) a scientific socialism. They could not be jettisoned lightly. On the other hand, the empirical observations (for example, the recent trajectory of workers’ wages) were too well known to be easily dismissed. Moreover, Bernstein had impeccable SPD credentials: from 1881 to 1890 he had been the editor-in-chief of an important emigre journal that upheld orthodox Marxism. ³⁸ Moreover, while in exile in England, he had been close to Friedrich Engels. As one scholar puts it, “when Bernstein challenged the accuracy of Marxian prophecy it was as if the pope declared there would be no Second Coming.” ³⁹ Significantly, this controversy involved little dispute about values; in particular, orthodox types and revisionists continued to agree on the importance of economic and political equality. Instead, there was strenuous disagreement about strategies: how to get from here to “there.”

There was no straightforward solution to the forecasting errors and ensuing inconsistencies underscored by Bernstein. All alternatives faced major difficulties. It is therefore not surprising that many of the major participants in this drama either tried to ignore the issue or denied certain obvious facts. The leaders of the SDP—August Bebel and Wilhelm Liebknecht—calmly pursued the revisionists’ in-practice strategy—orient the party toward winning elections—while mostly ignoring “theoretical” matters, that is, whether several of Marx’s key predictions had been falsified. Thus, they accepted living with an inconsistency between the SDP’s strategy, which had adapted to the realities of late-nineteenth-century German political economy, and the ideology’s empirical content, which was left unchanged. Equally unsurprisingly, Bebel and Liebknecht continued to maintain that the party was a genuine Marxist organization. Bernstein himself continued to identify as a loyal Marxist. Rosa Luxembourg, more clear-headed about the implications of Bernstein’s critique, said otherwise. ⁴⁰

Arrow’s Impossibility Theorem

Over 200 years ago, the Marquis de Condorcet discovered a curious fact about majority rule (and, hence, about democratic voting systems): it can produce inconsistent results. Consider the following example. Three legislators, L1, L2, and L3, face alternatives x, y, and z. L1 prefers x to y and y to z; L2 prefers y to z and z to x; L3 prefers z to x and x to y. Each legislator’s preferences are consistent, that is, the preference orders are transitive. For example, since L1 prefers x to y and y to z, she prefers x to z. ⁴¹ Majority rule determines the outcome.

Surprisingly, however, in this instance the individual-level property of consistency is not replicated at the institutional level. Majority rule does not produce a consistent (transitive) collective preference ordering. To see this, first pit x against y: the outcome is x. Next, pit the winner of that contest against z: the challenger, z, triumphs. Thus, the legislature prefers z to x and x to y. Common sense suggests that it would also prefer z to y. This, however, is incorrect. If y is pitted against z, y wins. Thus, the legislature’s preferences are inconsistent: it prefers y to z and z to x—but x to y.

Kenneth Arrow knew about this defect of majority rule as a graduate student. Believing that some collective choice rule will produce consistent collective preference orderings, he set out to find it. ⁴² As a good democrat, Arrow imposed a constraint: an acceptable rule could not be dictatorial. ⁴³ Furthermore, it had to be minimally responsive to what individuals want; if all voters prefer x to y, then the collective choice rule must rank x over y (the Pareto criterion). A third requirement was that the rule must work with whatever preferences the decision-makers have; it could not disallow specific profiles of individual preferences. The last criterion, independence of irrelevant alternatives, though less intuitive than the other three, also seemed mild. ⁴⁴ Indeed, apparently Arrow regarded all of them (that is, consistent collective choice plus the four he added) as necessary properties for any reasonable collective choice rule. ⁴⁵ To his surprise, he could not find a procedure that satisfied all five. Then, to everyone’s surprise, he proved that no such rule exists. Though each criterion by itself seems reasonable, collectively they are incompatible.

Arrow’s discovery stimulated other scholars to prove many related impossibility theorems about collective choice procedures. ⁴⁶ This body of work indicates that there is a big though often unnoticed gulf between stipulating a list of values or criteria that democracy or any other political institution should satisfy and analyzing the relations among those standards. Writing down a list of values that collective choice processes should achieve is easy; figuring out whether these evaluative criteria are consistent is often difficult. Upon careful inspection, the values may turn out to be inconsistent; if so, the entire set cannot be realized by any institution, existing or imagined. ⁴⁷ Something must give. As Isaiah Berlin warns us, in some circumstances decision-makers—even those who have embraced sophisticated ideologies—will face significant unanticipated value trade-offs. ⁴⁸ In short, whereas incompleteness is the probably inevitable result of a belief system grappling with a reality that is more complex than it is, internal inconsistency is the probably inevitable result of a belief system becoming even modestly complex.

Some readers may object that the foregoing demands too much of ideologies. If all serious ideologies exhibit holes and inconsistencies, then neither is a distinctive defect of any of them.

I completely agree. If my hypotheses about the incompleteness and inconsistency of ideologies are correct, then pointing out that a particular ideology (say, liberalism) is inconsistent may be completely uninteresting. Everything turns on the nature of the inconsistency. For example, the clash between trumpeting the value of individual liberty and supporting slavery is profoundly important; it must not be ignored or minimized as a mere anomaly. Similarly, it is an incompleteness of major proportions that anarchism does not address the issue of how to enforce justly established rules.

Thus, the argument is not whether there are good ideologies; with high enough standards, with sufficiently demanding values, no ideology will pass muster. The relevant issues are whether some ideologies are better than others and whether we can improve those that currently appear to be the most promising. Trying to do more than that runs into feasibility limits—and ought implies can. It is time, then, to turn to the topic of evaluating and criticizing ideologies.

Criticizing and evaluating ideologies

Thus far, this essay has focused on analyzing ideologies empirically. However, I believe that descriptive and normative analyses can enlighten each other. In particular, if we accept as a premise that ought implies can, then we should take seriously what ideologies can be. A major descriptive thesis of this essay is that humans cannot create complex ideologies that are internally consistent. Ideologies that are so simple that all inconsistencies can be identified and removed are too spare to help orient us to the complexities of modern life. They are also likely to be inaccurate in various ways, for example, by making incorrect forecasts. So there is little point, in our role as ideology critics, in playing “gotcha” in this regard. If all serious ideologies are inconsistent in one way or another, being told that a particular ideology is inconsistent is not enlightening per se.

I have offered a similar descriptive thesis regarding the incompleteness of ideologies. I have further argued that reducing incompleteness often winds up increasing ideological inconsistency; the more content an ideology has, the more elements there are that can be inconsistent with each other. ⁴⁹ If these two descriptive theses are accurate, then it is neither informative nor interesting for a critic of a particular ideology to point out that these properties hold, at a given point in time, for the belief system she or he has scrutinized.

This justifies maintaining, as I did above, that in static evaluations of ideologies we should prioritize “better” over “good.” Given sufficiently demanding criteria of evaluation, no ideology may be satisfactory, let alone good, but some are better than others. Relatedly, in dynamic evaluations we should try to figure out, building on Imre Lakatos’s evaluation of sequences of scientific theories, whether a sequence of related ideologies—an ideological tradition—is improving or degrading. ⁵⁰

Evaluating sequences of ideologies

In his famous modification of Popper’s work, Lakatos proposes that instead of assessing whether a single freestanding theory is falsifiable, we study sequences of theories. He argues that some sequences, which he called progressive, would become increasingly falsifiable (thus their empirical content would grow), whereas others would degenerate, as expanding belts of ad hoc auxiliary hypotheses would shield a research program’s core assumptions against empirical challenges. ⁵¹

Because ideologies contain not only factual claims but also value statements and political strategies (action injunctions and prohibitions), they cannot be falsified, taken as a whole, the way scientific theories can. Hence, Lakatos’s idea of assessing sequences of related belief systems must be used with some care. The empirical parts of ideologies can degenerate in Lakatos’s sense, but one cannot say that, for example, Marx’s concern for politicoeconomic equality has been falsified at any point in history. That would be a logical error.

Nevertheless, Lakatos’s idea has direct implications for the strategic parts of ideologies. Strategies are usually based partly on empirical claims. In particular, (a) even “thin” ones typically include a causal diagnosis of the current political situation (“it’s all the immigrants’ fault” is a causal claim as well as a condemnation) and (b) the strategy itself is based on a forecast that if one follows the ideology’s action recommendations, then the ideological goals will be reached. Thus, although strictly speaking one cannot predicate truth or falsity of a strategy, one can say that it is effective or ineffective, useful or useless, and those assessments are often tied closely to the accuracy of the strategy’s empirical premises.

A good example of sequences of ideologies involves the split between the two kinds of revisionism to orthodox Marxism: Bernstein’s reformist position and Vladimir Lenin’s revolutionary one. ⁵² Bernstein’s strategy was based on two key premises: cross-class coalitions were feasible and ensuing parliamentary victories in so-called bourgeois democracies could deliver significant policy gains for workers. A similar strategy devised by Swedish social democrats worked largely because both claims proved accurate (in Sweden). ⁵³ Thus, this hybrid ideological tradition united socialism and democratic theory in a fruitful way.

In contrast, Lenin’s modification of Marxism has turned out disastrously, partly because several key premises of his revisionist ideology proved inaccurate, including the beliefs that a one-party dictatorship would be both competent (it would outproduce the West) and distributionally equitable. The Marxist-Leninist hybrid, that is, the melding of socialism and one-party dictatorship, has been a degenerating sequence in ways similar to Lakatos’s original sense; its “hard core” has been increasingly protected by a belt of auxiliary claims that insulate the core from normative or empirical criticism. To take a well-known example, when a society would reach the stage of pure communism could be postponed indefinitely by regime leaders in Soviet-style systems.

The central issue dividing Bernstein and Lenin was the kind of hybrid socialism that should be created in order to correct the empirical and strategic deficiencies of orthodox Marxism. Both saw the passivity of classical Marxism as a key problem. ⁵⁴ Both recognized the importance of human agency. They differed over the strategic and institutional solutions to the common problem. The key difference was their stance toward democracy versus one-party dictatorship. This difference had long-lasting consequences, as descendants of each variant continued to endorse the design choice of that variant’s founder. ⁵⁵ Certain choices are dangerous.

Dangerous ideologies

Dangerous to what? Danger should be defined in relation to some state of affairs or goal-state. Mine is straightforward. I believe that liberal democracy is a better system than any other type of regime we know to be feasible. Hence the question is: What ideologies are especially dangerous to liberal democracy?

An obvious answer is: Those ideologies that explicitly justify authoritarian rule, such as fascism, are dangerous. However, liberal democracies today seem stable in the face of such overt threats. Hence, students of comparative politics are exploring a subtler answer: the proximate danger comes primarily from ideologies that can destabilize liberal democracy while claiming to uphold the purest type of democracy—that is, those that advocate rule by “the” people and by a leader who is somehow in direct contact with them and therefore knows what they want. ⁵⁶ This, of course, is populism. ⁵⁷

This is a credible threat assessment. Hungary is the prototypical example of a system that has been gradually transformed into an illiberal democracy. Prime Minister Viktor Orban has not used force, but he has chipped away at liberal democracy from within, justifying his actions on the basis of ethnonationalist populism. President Tayyip Erdogan’s conduct in Turkey has been similar. This clever strategy of “stealth authoritarianism” ⁵⁸ has become more common in recent years. ⁵⁹

Several who support this threat assessment have pointed out a fundamental incompatibility between liberal democracy and populism. In the former, one has opponents; in the latter, enemies. These categories differ cognitively, emotionally, and strategically. One negotiates with opponents; one tries to defeat enemies. This is an enormous difference. It is easy to see why mental representations that focus on enemies can destabilize liberal democracy.

To this I add two hypotheses. First, FFRs are central to populism. This is unfortunate not only because FFRs prompt thinking about enemies, but also because FFRs are maturationally natural (claim 13 above). ⁶⁰ They are cognitively simple and can easily be learned by young children: someone who wants to do bad things to one’s family is a foe. They require no formal instruction; they can be learned at home. Hence, causal inferences based on FFRs, such as “bad outcomes are caused by bad people,” are appealing. Consequently, they spread easily and, having spread, are cognitively sticky. ⁶¹

In short, friend-foe reasoning is for most people the default way of thinking about politics. ⁶² This helps populism flourish. As scholars we may disdain the “thinness” of populist ideologies, but for meme-diffusion ⁶³ thinness is an advantage. ⁶⁴

Second, populism is well-suited to the egocentric property of FFRs. There are “the people” (my people, naturally) and then there are, as Stalinists put it, the enemies of the people.

Because this match between populism and underlying FFRs is a permanent one and because FFRs are our default way of thinking about politics, populism is not merely strategically advantageous to contemporary demagogues; its properties make it a permanent threat to the ideas underlying liberal democracy. In good times, populism may remain latent or restricted to the electoral fringe, but in bad times, it can be activated or mainstreamed by unscrupulous politicians.

As many scholars note, populism easily combines with more detailed ideologies, left or right, that give it programmatic content. In terms of this essay’s argument, populism’s potential for recombinant innovation is high. This helps it to recur. ⁶⁵

Conclusions

If the evaluations made throughout this essay sound suspiciously neo-Enlightenment, you would be right. I cheerfully admit that Enlightenment belief systems form an ideological tradition in the nonpejorative sense of this essay. It is, however, an unusual ideological tradition. It embraces science and, more generally, the view that whenever possible, important claims should be justified by reason or evidence. I believe that this stance gives the Enlightenment tradition stronger self-correcting capacities than those of most ideological traditions. And if it is true that all significant ideologies are both incomplete and internally inconsistent, as well as inaccurate, then self-correcting properties are extremely desirable.

One can extend this argument from ideas to corresponding institutions. Following John Stuart Mill, John Dewey, and Popper, I conjecture that the above claim about self-correcting properties holds in practice, that is, regarding real institutions, as well as in theory. That is, I believe that the main institutions highlighted by the Enlightenment tradition—science, democracy, markets, and the rule of law—on average have stronger self-correcting tendencies than do any known alternatives. ⁶⁶

Are these good institutions? Not necessarily. Much depends on how demanding the criteria are we pack into the standard of “good.” ⁶⁷ Arrovian impossibility theorems, for example, set limits on what can be attained by any institution—as do empirical constraints. In particular, a pre-Enlightenment institution, the state, is here to stay (pace anarchism) for the foreseeable future. If that prediction is correct, then pro-democracy ideologists should work on how to hold the state’s decision-makers more reliably accountable to citizens.

Beyond the foreseeable future, who knows? Philip Tetlock tells us that forecasts of political events more than five years out are virtually worthless. ⁶⁸ Hence, a premium should be placed on ideologies that create resilient institutions—those that use scientific reasoning and evidence (deduction and induction) to adapt to unanticipated events—and that seem likely to give rise to subsequent ideologies that are pro-democratic.

Consistency requires, however, that these methods be deployed on the above recommendation itself. And here the news is sobering. Considerable evidence indicates that scientific thinking is not maturationally natural. ⁶⁹ Learning it requires years of systematic study in formal institutions. Scientific thinking is a beautiful hothouse plant: a special environment is required.

In contrast, friend-foe thinking is a weed: it grows everywhere. And there will always be a probabilistically steady stream of demagogues, eager to use the weed for their own purposes.

Hence, the Enlightenment “project” will never be completed. It cannot be. Given that science is both corrigible yet not maturationally natural, democratic norms might be less stable than we had thought, friend-foe thinking is maturationally natural, and there will always be demagogues. Pro-democracy ideologists have our work cut out for us.