Cities today are both the engines of innovation and economic growth, and the settings for concentrated social problems. As cities around the world expand in size and impact, advances in the scientific understanding of cities, urbanism, and urbanization take on increasing urgency. Do fundamental urban processes exist that generate a basic kind of city, in whatever context or culture or time period? Or are the expressions of urbanism too diverse to include in a single model? Is contemporary urbanization simply an elaboration upon past urbanization processes, or is it a fundamentally different kind of process?

Urbanization through the ages has manifested an enormous variation in the spatial and social forms of cities, their size, functions, activities, and growth patterns. Given this great variability in so many domains, it is not hard to argue that any notion of a single urban form or process or pattern throughout history must involve over-simplification. The differences among cities, across space and time, would seem too great to fit into a single type or model. Nevertheless, scholars in a variety of disciplines have argued that cities—regardless of their size, geography, temporal setting, or cultural milieu—share many underlying social characteristics, and play similar functional roles in different human societies (Mumford, 1961; Jacobs, 1969; Hall, 1998). There is a growing recognition that human settlements from Ur to Mumbai share enough in common that the term “cities” can be used to meaningfully refer to entities separated by thousands of years (Algaze, 2008; Smith, 2010b; York et al., 2011; Barthel and Isendahl, 2013). Yet this work has failed to specify just what characteristics of cities through the ages allow them to be considered or analyzed as a single phenomenon, in spite of their obvious differences.

Cities (and human settlements more generally) are essentially about the advantages afforded by aggregation which in turn are a manifestation of human sociality (Boyd and Richerson, 2005). Population size is both a major determinant and consequence of social evolution (Henrich, 2015). The basic demographic dynamics of expansion, maintenance, and decline are essential to cities, ancient, pre-modern, and modern. Over their existence all cities documented by archaeology and history went through phases of expansion (in spatial extension and population size) before eventually declining. The existence, extent, and relevance of economic growth (which can mean either increases in material output or increases in output per capita) in ancient and pre-modern societies is a topic of considerable debate (Erdkamp, 2016; Jongman, 2016; Stark et al., 2016),(Greene, 2000; Scheidel, 2004; Pryor, 2005). The reality of urban growth, on the other hand (either an increase in the proportion of a society's population residing in urban settlements or the increase in the population size of individual urban settlements), in ancient and pre-modern societies is not. Understanding what is common and what is not with regards to urban (population) growth across time is key to understanding what is common to urban life across eras and civilizations.

The question posed in our title—one thing or many?—has two answers. First, when we focus on the institutional framework of cities, including the mechanisms that generate urban growth, there are two fundamentally different forms of cities. We call these economic cities and political cities. In urban economics, this contrast is often discussed in terms of differences between “normal” cities and primate cities (de Long and Shleifer, 1993; Ades and Glaeser, 1995), but in fact the distinction runs deeper than this. Most cities before the modern era were political cities, meaning that their dominant institutions were in the realms of power and administration by a ruler or ruling elite. A few pre-modern cities and most cities today are economic cities, meaning that economic considerations dominate the locational decisions of individuals and production units, and that economic activities largely shape their social structure and economic forces dominate their processes of growth.

Notwithstanding the very real and important differences between economic cities and political cities, urban growth in both types originates in a common set of behavioral and built-environmental mechanisms that underlie both the economic and political drivers normally discussed in the urban literature. A focus on these fundamental mechanisms leads to our second answer to the question of “one thing or many?”: the city is one thing amidst a plurality of manifestations of urban life. It is a place of energized crowding (Kostof, 1991, p. 37) that generates growth and change (Smith, 2019). These underlying mechanisms have been explored by several strands of recent work that take the perspective—theoretically grounded and empirically supported—that cities are, and have been, social networks of people embedded in physical space (Fisher, 2009; Hipp et al., 2012; Bettencourt, 2013; Youn et al., 2016). This body of research has revived the prospects for building an analytical framework for understanding the origins and drivers of urbanization operating in cities from the distant past and contemporary urban life, both economic, and political cities.

Our dual answer to the question of “one thing or many?” has implications for understanding cities and urbanism today. Whether one is interested in identifying universal urban traits (Smith et al., 2015), tracing the development of urbanism over time (Mumford, 1961), or using ideas from past cities to inform contemporary practice (Rapoport, 1973; Hakim, 2012), it is important to understand both the continuities and disjunctions between cities today and those of the past. Are the principles of the new urbanism (Congress for the New Urbanism, 1996) based on universal urban realities, or do they only reflect conditions of the very recent past? How can scholars predict whether current principles of urban resilience will play out over long periods of time? We propose that our formulation of cities through the ages can help frame analyses of these and other questions about cities and urban life in the present and future.

In his book Triumph of the City, urban economist Edward Glaeser defines cities as “the absence of physical space between people and companies. They are proximity, density, closeness” (Glaeser, 2011, p. 6). Glaeser's minimalist definition does seem to capture what for many is the essential feature of cities. Yet its very minimalism allows the definition to apply equally to every form of human settlement—from hunter-gatherer camps to cities—in which physical proximity facilities social life. Furthermore, high density is no longer characteristic of all modern cities (Angel, 2012), nor was it characteristic all ancient cities (Fletcher, 2009).

Archaeologist [Cowgill (2004), p. 526] observed that, “It is notoriously difficult to agree on a cross-culturally applicable definition of “the” city, but we cannot do without definitions altogether.…No single criterion, such as sheer size or use of writing, is adequate.” The urban literature reveals two dominant approaches to city definition: a sociological/demographic approach and a functional approach. The most influential definition is that offered by sociologist [(Wirth, 1938), p. 8]: “For sociological purposes a city may be defined as a relatively large, dense, and permanent settlement of socially heterogeneous individuals.” This definition clearly fits contemporary cities, and it is favored by most scholars of urbanism today. But historians, anthropologists, and archaeologists have pointed out that the sociological definition excludes most pre-modern cities from consideration as urban settlements (e.g., Sjoberg, 1960; Fox, 1977; Smith, 2016).

Early urban settlements were prominent and influential within their regional settings, yet their levels of population size, density, and heterogeneity were considerably lower than contemporary western cities. This situation led to the adoption of a “functional” approach to definition, based on work in economic geography (Lloyd and Dicken, 1972). The definition of archaeologist [Trigger (1972), p. 577] is typical: “It is generally agreed that whatever else a city may be it is a unit of settlement which performs specialized functions in relationship to a broad hinterland.” An urban function is an activity or institution that directly affects life and society in a hinterland. The initial functional definition of cities focused on retail economic functions, and central place theory provided concepts and methods for understanding cities as economic central places (Christaller, 1966). Anthropologists then moved beyond economic functions to define cities using other regional impacts such as political administration or religion (Fox, 1977; Marcus, 1983).

While these sociological and functionalist urban definitions have often been opposed to one another and treated as alternative approaches, we wish to highlight an important commonality. Both definitions encompass the idea that social interactions within a delimited space are important drivers of the urbanization process. In the words of architectural historian [Kostof (1991), p. 37], “Cities are places where a certain energized crowding of people takes place.” That is, cities are settings for frequent and intense social interactions, and by implication these interactions have important effects on urban behavior and output (Smith, 2019). The density and social heterogeneity of Wirth's urban concept imply the importance of that energized crowding. And the urban functions of the alternative definition are nothing more than specific kinds of social interactions that generate the influence a city has on its hinterland. This notion of “energized crowding” is also the theoretical foundation of our discussion below of the city as many things at once, or, in effect “the city as one thing.”

Most cities before the Industrial Revolution were political cities and most contemporary cities are economic cities. This dichotomy is based on the nature of economic activity, its prominence in urban dynamics, and its relationship to growth. Virtually all of the literature in urban economics and urban geography focuses on the locational choice of businesses and individuals, and on the positive externalities (due to agglomeration) to the productivity of firms and individuals. This leads contemporary work in urban economics, economic geography, and regional science to have a “modernist” bent applicable mainly to cities whose economies are productive and dynamic. In our scheme most economic cities exist within the capitalist world system, but we think that more research is needed to determine the extent to which our economic cities concept might apply to cities before the modern era. Political cities, on the other hand, are those cities in which political or administrative activities predominate. Economic processes are either suppressed by political forces (as in recent primate cities), or else are simply far less developed than in contemporary cities.

The concepts economic city and political city are Weberian ideal types: pure classifications that will never match the empirical world precisely (Gerth and Mills, 1946, p. 59, 60). They are ends of a continuum. This dichotomy is similar—but not identical—to a number of dichotomous typologies in the urban literature (Table 1). We are proposing a new dichotomy because none of these prior schemes are sufficiently broad to encompass the entire range of historical urbanism.¹

The earliest of these ideal-type city dichotomies was Smith (1979) distinction between government towns and commercial towns (Stull, 1986). The primary activities in the former are administration and rule, which, in Smith's scheme, are considered unproductive labor. Smith describes these workers as “idle, dissolute, and poor” (Stull, 1986, p. 300). Most workers in commercial towns, in contrast, are “industrious, sober, and thriving.” Adam Smith's list of eighteenth-century government towns includes Rome, Madrid, Versailles, Paris, and Edinburgh prior to 1707. Commercial towns included Glasgow, many English towns, and most Dutch towns. He classified other cities of his time as having attributes of both types: London, Lisbon, Copenhagen, and Edinburgh after 1707.

The contrast between the consumer city and producer city originated with Max Weber and other early twentieth century economic historians to contrast the ways in which Classical and medieval cities obtained food from their hinterland. Commercial enterprises in medieval producer cities allowed urbanites to obtain food from farmers through commercial exchange, whereas elites in ancient consumer cities received rural income from rents. Classicists debated which concept best fit Roman cities and towns for decades (e.g., Parkins, 1997), an argument that got tangled up with the primitivist/modernist debate on the ancient economy. Recent work has reached two conclusions: (1) markets were operational in many but not all ancient societies; and, (2) the presence of markets does not turn ancient societies into protocapitalists entities (Morris, 2004; Feinman and Garraty, 2010; Garraty and Stark, 2010).

In a study of variation in Roman cities in Britain, Clarke (1993) evaluates several general models of the spatial organization of preindustrial cities, including Sjoberg (1960), Vance (1971), and Langton (1975). None of these models fit the Roman cities of Britain well, leading Clarke to synthesize various insights of these authors into two models that he calls, “Public urbanism” (Sjoberg's model of class-based spatial zones, with less emphasis on the commercial component) and “Commercial urbanism” (the Medieval city model of Vance, which emphasizes the importance of guilds and the lack of zoning by class).

The next two dichotomies arose in the 1950s in the literature on economic development. Economist Hoselitz (1955) classified cities as generative or parasitic if they have a positive or negative impact on economic growth in their region or country; see also Wrigley (1978). Anthropologists Redfield and Singer (1954) promoted stereotypes of developing nations in their classification of orthogenetic cities (traditional cities where the “moral order” dominates) and heterogenetic cities (modernizing cities where the “technical order” is primary). The dependent/autonomous economy dimension of Richard Fox's functional typology maps onto our political/economic dichotomy quite closely; his regal-ritual, administrative, and colonial cities categories fit into our political category, and his industrial and mercantile cities fit into the economic type. City-state capitals are split between the two categories; Greek, Yoruba, and Aztec examples were political cities, but many post-medieval European city-state capitals were economic cities. We also include primate cities in our political category. These are cities whose size far exceeds other cities in their regional system (Adamic, 2011). Of contemporary primate cities, [Ades and Glaeser (1995), p. 195] state, “political forces, even more than economic factors, drive urban centralization” (we discuss primate cities more fully below).

Our categories political and economic cities—like the other parallel dichotomies listed in Table 1—are Weberian ideal types that are not intended to match precisely any specific city. In fact, these categories may best be viewed as ends of a continuum rather than as a rigid dichotomy. Our purpose is to improve scholarly understanding of the variability among cities, not to categorize specific cities.

The concept of cities as places of intensive social interaction—“energized crowding” in the words of [Kostof (1991), p. 37]—is shared by both of the dominant definitions of cities and urbanism (Smith, 2019). Furthermore, processes of social interaction operate at a deeper or more fundamental level than the economic and political drivers of urban growth reviewed above, and they can be considered as generative forces for the growth of both economic and political cities. Our argument draws on a wide range of social-science scholarship on this topic.

Population size, density and heterogeneity have long been recognized as fundamental properties of urban settlements (Wirth, 1938; Bairoch, 1988; Angel et al., 2016).³ This socioeconomic and cultural heterogeneity in turn facilitates the plethora of interactions (intentional or serendipitous, fleeting, and consequential, anonymous or long-lasting, driven by economic imperatives or sustained by shared ideological commitments) that make urban life sociologically distinct (Jacobs, 1961; Fischer, 1975). It is this crowding—and the creative, inventive and innovative possibilities which it realizes—that in turn generates growth and change (Glaeser, 2011). Storper and [Storper and Venables (2004), p. 31] have analyzed the mechanism through which “energized crowding” occurs. They call this, “the most fundamental aspect of proximity: face-to-face contact.” Although their analysis of the role of face-to-face interaction in urban agglomeration is set firmly within the economic geography approach that ignores political cities and pre-modern societies, at least three of their four “basic functions” of face-to-face contact apply to pre-modern societies: communication technology; trust and incentives in relationships; and screening and socializing. Their fourth function—“rush and motivation”—is more difficult to evaluate for pre-modern contexts.

The perspective that all cities—across eras, geographies, and cultures—share a core of fundamental socioeconomic processes as well as certain predictable quantitative properties has recently coalesced into settlement scaling theory (Bettencourt et al., 2007; Bettencourt, 2013; Bettencourt and Lobo, 2016). Settlement scaling theory draws on insights from urban economics, economic geography, and regional science and shares with these disciplines a common explanation for the existence and development of cities as resulting from the interplay between centripetal and centrifugal “forces” (Colby, 1933; Isard, 1956; Fujita et al., 1999; O'Flaherty, 2005). Population size is arguably among the most important determinants and consequents of socioeconomic development and change in societies before the modern era (Carneiro, 2000; Johnson and Earle, 2000). Such relations are known across the sciences as scaling relations, which relate the macroscopic properties of a system—here a city—to its scale, or size (Barenblatt, 1996, 2003; Brock, 1999). For this reason, the systematic study of such relationships in cities is known as urban scaling or settlement scaling.

What is novel about the scaling framework is that it views cities as integrated socio-economic networks of interactions embedded in physical space, and then derives specific quantitative predictions about the relationship between population size, material output and aereal extent of settlements. The social interactions facilitated by physical proximity—and the lower costs associated with such interactions—drive productivity (Bettencourt et al., 2008, 2013; Bettencourt, 2013). These relationship have been statistically investigated not only for modern metropolitan systems (in the United States, Western Europe, Brazil, Japan, India, China and South Africa) but also for Native American farming villages in North America (before the arrival of the Europeans), Pre-Hispanic Andean and Central Mexican settlements, Ancient Greek and Roman cities, Medieval European cities and towns, and cities in Tudor England (Ortman et al., 2014, 2015, 2016; Cesaretti et al., 2016; Hanson and Ortman, 2017; Ortman and Coffey, 2017; Ossa et al., 2017; Cesaretti et al., under review). The results show a striking similarity in scaling relationships, as predicted by the settlement scaling framework. Across cultures and history, when human societies create permanent settlements, these grow denser, on average, with growing population. In addition, the increases in economic production and outputs are proportionally greater than the increase in population size. This, in turn, indicates that humans in larger settlements live in socioecological settings that facilitate higher rates of social interaction, relative to smaller settlements in a given context.

Social settings of ongoing face-to-face contact generate processes of trust, incentives, and monitoring that are among the key ingredients of successful community organization (Ostrom, 1990; Sampson, 1999; Bowles and Gintis, 2002). Urban planners are much concerned with finding ways that the urban built environment can foster productive face-to-face interaction (Talen, 1999; Brower, 2011). The observation that neighborhoods are universal features—not just of cities, but of almost all types of large human settlements (Smith, 2010a; Smith et al., 2015)—supports the notion that face-to-face interaction is fundamental and important in all kinds of cities, both economic and political.

The city as a facilitator of social learning might ultimately provide one of the most salient sources of continuity among urban forms across space and time. After all, social learning is emblematic of Homo sapiens. The resulting process of cultural adaptation that is responsible for our species' success was facilitated by both population size and social connectedness (Boyd and Richerson, 2005; Henrich, 2015). Urban life is just one manifestation of the way humans have constructed social and physical spaces to exploit their unique abilities to build cumulative culture (Enquist et al., 2008). We should note that not all of the outcomes of energized crowding are positive traits; increased poverty, crime, and social alienation are also products of social interactions in cities.

These observations about the generative role of face-to-face interaction, from a spectrum of academic disciplines, apply equally well to political and economic cities. The urban built environment provides settings for social interaction, whether the parks, cafes, and sidewalks of contemporary cities or the plazas and marketplaces of ancient cities (Stanley et al., 2012). The same processes of communication and exchange take place among the residents of all cities, past and present. These basic interactions are the essence of what is distinctive about cities as human settlements. Their expression in the broad similarities of scaling relationships across time and history has a clear implication: the city as a container for social interactions throughout history and around the world is one thing, not many.

Our analysis has implications for identifying more and less productive paths of comparative research of cities and urbanization. The “city as two things” perspective suggests that analyses of urban growth in economic and political cities are best carried out independently because the processes of growth in the two types of city are quite different. Empirical analyses of the rates and mechanisms of growth in past cities can help identify economic cities before the modern epoch. We argue above that economic growth in Imperial Rome was insufficient to generate the agglomeration economies of modern economic cities. But, given the high level of documentation for Roman urbanism, quantitative analyses of Roman urbanism (e.g., Bowman and Wilson, 2011; Jongman, 2016; Hanson and Ortman, 2017; Hanson et al., 2017) provide models that will contribute to a better understanding of the nature of political and economic cities in the ancient world more generally.

Attempts to apply models of growth devised for contemporary economic cities to ancient cities in the absence of extensive quantitative data (e.g., Algaze, 2008, 2018) may be premature. On the other hand, models of modern primate (political) cities (de Long and Shleifer, 1993; Ades and Glaeser, 1995) would appear to be a productive source of insights for understanding ancient political cities. And given the smaller number of such political cities today and their lack of prominence in the literature on growth theory, perhaps the growing research on urban dynamics in pre-modern political cities might help scholars better understand the primate cities that dominat many developing nations today.

Our second perspective—the city as one thing—has great potential for comparative research that draws cases from, and produces insights for, both pre-modern and contemporary cities. For example scholars are beginning to draw parallels between the roles of neighborhoods in ancient and modern cities (Smith, 2010a; Arnauld et al., 2012; Sampson, 2012; Smith et al., 2015). Our analysis suggests that this topic is a particularly fertile one for comparative analysis, given the importance of social interaction and the built environment in neighborhood dynamics. The success of urban scaling research in identifying the predicted scale relationships in ancient settlement systems (see the sources cited above) provides strong empirical support for this perspective.

Productive research comparing pre-modern and contemporary cities needs to avoid two approaches that have long dominated comparisons of ancient and modern social and economic conditions. On the one hand, many social scientists outside of anthropology simply assume that conditions in the past or in non-western societies were similar to conditions today, and thus economic models can be applied directly without modification. This approach is called “modernism” in Classical studies and “formalism” in economic anthropology. Such work tends to be ignored or dismissed by historians and archaeologists because much of it fits the facts so poorly as to be useless. On the other hand, some scholars (“primitivists” in Classics and “substantivists” in economic anthropology) have portrayed the past (and non-western societies) as so radically different from the present that comparisons are impossible (Polanyi et al., 1957; Finley, 1973). Most scholars now see this as a misguided and highly limiting approach to past societies and economies (Smith, 2004; Wilk and Cliggett, 2007).

Our dual answer to the question posed in this paper's title (“one thing or many”) can help scholars avoid these problems of misleading comparisons. From the perspective of the forces that create cities and generate urban growth, there are two general types of cities: economic and political. The dynamics of growth and operation of these cities are quite different, and facile comparisons between them will do little to illuminate general processes of urbanization. But from the perspective of the way that people interact with one another in delimited urban spaces, all cities are similar; there is only one type of city. The “energized crowding” first identified by [Kostof (1991), p. 37] is one of the universal features of cities, from ancient time to the present (Smith, 2019). The ability of scholars to explain and understand this process of energized crowding may help determine the productivity of comparative urban scholarship in the future, and it may even contribute to the success or failure of cities and urbanization in the future.

JL and MS conducted the research and wrote the article.