The constitution of the city and the critique of critical urban theory

The sources and power of agglomeration

In any society marked by what Durkheim (1893) called ‘mechanical solidarity’ there is little in the way of social or economic differentiation: individuals are bound together as a cohesive group by common beliefs and customs but in the absence of an extended division of labour, overall levels of social interaction are low. An example of this kind of society would be a population of largely self-sufficient peasants whose coherence within an inclusive social organism derives primarily from shared traditions and cultural traits. In spatial terms, any society of this kind will tend to form a pattern of dispersed family units or possibly small-scale clusters but is unlikely to be accompanied by significant urbanisation processes.

When individuals become more highly differentiated from one another and more interdependent, much more complex patterns of social organisation and human geography will come into existence. Durkheim described this more complex form of society in terms of ‘organic solidarity’ where individuals are organised in overlapping networks of interaction in which – by comparison with mechanical solidarity – a high degree of impersonality and anonymity prevails. The primary source of organic solidarity, according to Durkheim, is the division of labour where economic production is broken down into many different but functionally interlocking activities. In a more elaborate description, we can recognise (a) a technical or occupational division of labour where workers are differentiated by physical endowments and skills, and (b) a social division of labour comprising specialised units of production, each of which incorporates a coordinated team of workers. The whole structure of individuals and teams is articulated together in a series of relays and circular flows such that the output at one work-station or production site becomes the input at another until final consumption occurs (Smith, 2008 [1776]). In this paper, I shall argue that the division of labour is fundamental to the genesis and durability of cities by reason of complex intermediating processes that revolve around spatial agglomeration and the formation of a dense fabric of interrelated land uses. To be sure, many other forms of social differentiation and interdependence are observable in societies across history and geography, and these must also be accorded due consideration in any putative urban theory, though given the universal imperative of securing the material continuity of individual and social life, special attention must be assigned to the division of labour with its foundations in complex forms of production and work. The main claim to be developed here is that the division of labour plays a paramount role in bringing forth agglomerated forms of human settlement. To put the latter statement in provisional perspective, only consider that at least one and often two members of the vast majority of families in the world’s large cities today are entangled for a considerable proportion of their daily and weekly time in dense webs of production, trade and work that continually change and intermesh with one another.

Individuals caught up in divisions of labour are necessarily and intimately bound together in networks of mutual relationships reflecting the interdependencies that crystallise out between specialised but complementary agents. If, by chance, society were endowed with magic carpets (allowing costless and instantaneous interaction over any geographical extent) any random distribution of these agents would be as viable as any other, and durable settlements or communities bringing them into mutually proximate collectives would be superfluous. In reality, all types of human interaction over geographic space encounter a great variety of impediments and come with many costs related to the resources and time consumed by displacement. For this reason, notably when high-cost transactions per unit of distance are involved, as in many segments of the division of labour, selected groups of interlinked individuals will tend to condense out on the geographic landscape into nodal aggregations where some workable arrangement of interdependency and community can be sustained. Equally, the same aggregations will generally be open to exchanges with an external world by virtue of extended but comparatively low-cost transactions per unit of distance.

Even under conditions where the division of labour is only weakly developed, occasional small spatial clusters may come into being, as, for example, in the case of an administrative body that coordinates collective interests in a society of essentially self-sufficient families. When self-sufficiency begins to break down, additional primitive foci or central places facilitating economic specialisation and exchange over space will tend to appear, as revealed in a number of archaeological studies (e.g. Nakoinz, 2010). In yet more complex societies, with well-developed divisions of labour, incentives leading to spatial agglomeration will become correspondingly more intense. Other kinds of social relationships, too, can play a role in encouraging agglomeration, and specialised settlements may accordingly form on the basis of religious, political and military priorities, though they are likely to be small given that opportunities to reap positive returns of scale and scope will in all probability be limited. Communities endowed with an assortment of production units linked to external markets, however, are much better equipped to overcome internal shackles on their own spatial development, because a deepening localised division of labour will usually generate internal economies of scale and scope, thereby stimulating endogenous growth and opportunities for external trade (Scott, 1988; Storper, 1997). At the same time, incipient clusters are likely to emerge at locations that offer special advantages, such as a resource site, a route intersection, a natural harbour or a river crossing. Locational factors of this type play a role in the spatial seeding of towns and cities and in helping to ground economic growth but in the absence of agglomeration processes they are more liable to persist simply as passive geographic features. Indeed, the logic of the division of labour, agglomeration and urbanisation means that it is entirely possible for cities to flourish at purely arbitrary locations where no natural geographical advantages are available.

Whatever the substantive content of these processes, it is primarily the tightly wrought transactional grid reflecting the division of labour driven forward by production that sets the forces of agglomeration in motion, thus drawing diverse but complementary activities together within nucleated spatial units (Scott, 2012). Nucleation is further enhanced where the agglomerative powers of this transactional grid are compounded by the positive externalities that are invariably secreted within clusters of economic and social activity. Duranton and Puga (2004) categorise externalities in terms of economies of matching (e.g. pairing of needs and opportunities), sharing (e.g. indivisibilities and common assets) and learning (e.g. information exchange and innovation). Further investigation of a varied palette of historical and geographical cases is needed before we can fully assess the interplay between different kinds of transactional systems, agglomeration effects and specific modalities of urban development, but even in the earliest cities circumstantial evidence exists in favour of the view that divisions of labour based in active production systems are a major if not the dominant factor in urban development. The primitive settlement of Çatal Hüyük, which flourished in southeast Anatolia between 7100 and 5700 bc, has little in the way of monumental buildings representing social power or religion but shows evidence of an extraordinarily advanced division of labour supporting a population of between 5000 and 6000 (Mellaart,1967; Todd, 1976). Among other industries, baking, bead-making, bone-carving, leather-making, metal-working, pottery, weaving, dyeing and fashioning obsidian into tools and weapons are strongly represented in the archaeological record of Çatal Hüyük and it appears, too, that much of the output from these industries was exported. Similarly, well-developed systems of manufacturing and trade formed the economic base of the numerous small urban nucleations or oppida (fortified and unfortified) that proliferated across Europe north of the Alps in the late Iron Age (see, for example, Cunliffe and Rowley, 1976; Fichtl, 2005). Of course, it is industrial capitalism, above all, that brings the division of labour and urban agglomeration together in powerfully interwoven association marked by the formation of growth poles generating wave after wave of cities on the landscape. Yet, these cities are never exclusively excrescences of production, work and the division of labour, for as they grow, other spaces devoted to residence, leisure and other social pursuits coalesce out as distinctive sui generis constituents of the city together with matching transactional structures that simultaneously reinforce agglomeration and encourage the detailed spatial differentiation of intra-urban land uses.

The above remarks allude to the vital core of the urbanisation process but a further crucial point must still be made. As Childe (1950) has forcefully pointed out, cities cannot survive in the absence of a surplus food supply, notwithstanding the genetic impulses that flow from the agglomeration process. However, while the existence of a food surplus is emphatically a necessary condition for urbanisation to occur it is far from sufficient, for there is no special reason why the surplus alone will lead specifically to clustering as opposed to some other, arbitrary spatial pattern. It is impossible to feed the mass of citizens without a surplus but it is the transactional logic of complex societies and all that this implies for the organisation of geographic space that accounts for the accretion of an urban population in the first place. I should point out, in passing, that Walker (2016) has taken issue with this formulation and I shall therefore return to the topic of the surplus more fully at a later stage.

The spatiality of the city

Thus far, I have reasoned that the genetics of urban clusters turn centrally on the division of labour, the concomitant formation of a network of transactional relations and hence spatial agglomeration. What, now, of the internal organisation of the nodal entities created in this manner?

At any given stage of development, the city will be nominally constituted as a disparate set of land uses reflecting the division of labour and the living arrangements of an attendant labour force, all drawn together in geographic space around their common centre of gravity. The question we now face is how, in the most general terms, will these land uses be allocated to actual locations? The pertinence of this question is accentuated when we take into consideration the fact that every single unit of land use in the city requires a finite quantum of space in order to function. The total number of units available, moreover, diminishes dramatically as we move towards the city’s centre of gravity. At the outset, then, an endemic tension exists between the demand for space to accommodate each individual unit of land use and the rising scarcity of space as we approach the inner core of the city. This tension typically creates distinctive spatial patterns in the city as some land uses are able to commandeer advantageous central sites (as well as sub-central sites where subsidiary gravitational forces may develop) while others are relegated to more outlying secondary sites. The whole will tend to form a continuous spatial fabric with density levels diminishing as we move away from central and sub-central locations to the extensive margins of the urban area, which, even in their peripherality, are still firmly secured within the gravity field of the city. The resulting forcefield of intra-urban space constitutes what we may refer to as an urban land nexus forming a platform that interacts with layers of more complex human activities as different social formations come and go and as the city enters into its full social and political efflorescence (Scott, 1980; see also Scott and Storper, 2015).

The actual activities that coincide with the skeletal tissue of the urban land nexus will vary greatly in their qualitative character over geography and history, though production and residential spaces will almost always form the minimal integument of the city. Production, for its part, implies the presence of cohorts of workers and consumers who will be accommodated in whatever kinds of residential spaces are apposite given prevailing social conditions. In turn, the spatial extensiveness of the city necessitates communication corridors within the urban land nexus so that complementary activities can interact with one another. These basic components of intra-urban space will be accompanied by and modified by whatever additional social or cultural activities emerge within the prevailing social order. To add to the complications of this situation, particular types of land use will have varying degrees of affinity or aversion to other types so that spatial sorting mechanisms will frequently operate leading to the formation of specialised intra-urban districts and neighbourhoods. As these different crosscurrents ebb and flow, the outer edges of the urban land nexus will come into irregular contact with non-urban land uses so that the city will rarely be circumscribed by a clear linear boundary. The best we can say about this matter for now is that the city will in many instances extend outwards in an indefinite blur. This is an unsatisfactory statement as it stands, but it will be clarified in the subsequent discussion when we come to deal with the issue of planetary urbanisation.

No matter what particular land uses may exist in any given urban centre, the scarcity of advantageous locations in relation to the total amount of space required implies that some degree of contestation for available sites will be endemic within the city. Thus, unless workable mechanisms can be found to ensure that units of intra-urban space are allocated in some socially sanctioned manner there is every reason to suppose that this contestation will be disruptive. Competitive land and property markets regulated by prices represent one such mechanism, authoritarian coordination is another, and any number of possible intermediate arrangements are possible. The main point here is that the innately problematical spatial architecture of the urban land nexus means that its social reproduction through time calls insistently for some overarching agency of stabilisation. Even market mechanisms in the end are rife with failures due to endemic externalities and other vulnerabilities so that supplementary social control becomes imperative. By its nature, then, the urban land nexus is not simply a static geometric construct but is also inherently dynamic, communal and political, and this, as we shall see, becomes vastly more the case as the diversity of its internal land uses increases.

The basic internal dimension of urban materiality thus comprises as its mainstay a web of structured spatial outcomes held together in geographic space by their own gravitational force. I have provided here a diagnostic view of how this system can be understood in terms of separate layers of analysis hinged on the division of labour, agglomeration and the urban land nexus. Need it be said that these individual layers do not occur in history as mutually independent outcomes but only as an evolving amalgam of variegated events and reciprocal relationships? My basic claim is that this approach offers essential insights into how cities come into being, how they develop and how they are configured as social entities contained within but different from society as a whole.

The urban as abstract concept and concrete entity

It is now essential to broach a question that is becoming ever more urgent as the discussion progresses. How, in brief, do we reconcile the abstractness of the theory of agglomeration and the urban land nexus as presented thus far with the fact that every city is different in very complicated ways from every other city? I propose to deal with this query in terms of first principles.

Let us begin by reviewing a celebrated passage from the Grundrisse by Marx (1973 [1857]) where he defines the concept of concrete abstraction. Note that in this passage I have substituted the word ‘city’ for the word ‘population’.

… if I were to begin with the city, this would be a chaotic conception of the whole, and I would then, by means of further determination, move analytically towards ever more simple concepts, from the imagined concrete towards ever thinner abstractions until I had arrived at the simplest determinations. From there the journey would have to be retraced until I had finally arrived at the city again, but this time not as the chaotic conception of a whole, but as a rich totality of many determinations and relations.

In its original context, the passage refers to the notion of surplus value. It outlines a method whereby surplus value can be evaluated as a universal abstraction common to all forms of capital, and that can then be tracked though its concrete instantiations in profits, rents and interest as these quantities are mobilised in various social situations (cf. Little, 1982; Moseley, 2014). Like the notion of surplus value, the model of the urban process put forward here also functions as a concrete abstraction though neither surplus value nor the urban land nexus is ever actually observable in its conceptual purity. The philosophy of critical realism offers a very similar kind of interpretative tactic (Sayer, 2000). As such, the model does not seek to be a mirror of any one city in particular but only a theoretical distillation of the essential internal coherence of urban phenomena in general. The model itself can be embedded in a yet more broadly ranging abstraction representative of geographic space in its role as the foundation and expression of the whole field of social existence (cf. Lefebvre, 1974; Stanek, 2008; see also the later discussion on planetary urbanisation).

On the one hand, then, the city can be apprehended as a universal concrete abstraction in its essence as a mode of spatial integration governed by processes of agglomeration and the logic of the urban land nexus. On the other hand, every city takes on concrete particularity when the heterogeneous ‘objects, spaces, persons and practices’ of actual social life are reflexively interlaced within its spatial fabric as elements of its relational structure. It is precisely this interlacing that makes these things finally urban above and beyond their non-urban identity as social phenomena at large. Concomitantly, an individual factory, retail outlet, house, hospital, transport link or green space cannot be categorised as being urban solely on account of its physical or functional qualities but only by reason of its relational insertion within the framework of the urban land nexus. Correspondingly, any phenomenon that exists within the city but is not organically ingrained in its spatial-cum-transactional armature is only contingently urban. For example, the books currently lying on my desk are certainly located within a city, but they are inert insofar as there is anything specifically urban about their existential condition. In the same way, the safety regulations in force at international airports worldwide have no necessary connection to the urban. Likewise, the familiar phrase, ‘urban education’, may indeed capture something that is inherently urban (e.g. a vocational programme directed to some particular group of urban residents) but to the degree that it refers simply to educational programmes attuned to modern society – inside or outside the city – it is apt to be a misnomer. By extension, basic elements of capitalist society such as social class, accumulation, profit, value and market exchange can be analysed without any reference whatever to urbanisation processes or the city, even though they assuredly resonate reflexively with the city once they are brought into conjunction with the urban land nexus. In a nutshell, we can say quite literally that some objects and occurrences are organically of the city in contrast to others that exist merely in the city as contingent eventualities.

This commentary now provides an opening into the issue of the ambient and exogenous dimensions of urban materiality and their critical role in the actual composition and recomposition of the urban land nexus. The city expresses these dimensions in disparate ways, visible and invisible, both ‘by making tangible the most abstract relationships, juridical, political, ideological’ (Lefebvre, 1970: 36), and via internal urban responses to events, opportunities and pressures coming from the spatial outside. The fact that these influences are always peculiar to specific social formations, with their intrinsic multiformity, means that cities are always amenable, as Buckley and Strauss (2016) aver, to varying pathways of historical and geographical investigation.

Urbanisation versus planetary capitalism

It is widely asserted that the concept of planetary urbanisation was first developed by Henri Lefebvre in his celebrated book La Révolution Urbaine (1970). This and related writings by Lefebvre are replete with inspired insights about the social meaning of cities and space but they are also notoriously cryptic so that it is often difficult to arrive at an accurate summary of his arguments. Still, the very first line of the book announces the hypothesis that ‘society has been completely urbanised’, leading in the text that follows to an extended disquisition on urbanisation, the city and urban society, but one that never quite goes as far as an explicit articulation of planetary urbanisation as it is currently understood. We can, to be sure, find textual evidence in the book for any one or all of the following propositions about the contemporary city: (a) cities are expanding outwards at an accelerating pace; (b) cities are growing everywhere in number and size; (c) the world’s population is increasingly contained in cities; and (d) something designated ‘urban society’ (i.e. ‘the society which results from a process of complete urbanisation’ (p. 7)) is emerging in the wake of what Lefebvre calls ‘industrial society’. Lefebvre also states that this urban or post-industrial society can be understood in terms of a process of ‘implosion-explosion’ meaning that an intensifying centralisation of human society is occurring in cities along with the proliferation of ‘disjoint fragments such as peripheries, suburbs, secondary residences, satellites, etc.’ (p. 24). All that being said, the reader would certainly be hard-pressed to find explicit textual evidence to suggest that Lefebvre is claiming that the current historical moment is one in which the whole Earth is enveloped in an urban membrane, though I should point out that in a much later publication Lefebvre (1989) does invite us to ponder the proposition that at some time in the third millennium urban forms may eventually diffuse to the point where they cover the whole of the globe.

Whatever Lefebvre’s precise meaning may have been in La Révolution Urbaine, his work has most certainly prompted a spate of publications on planetary urbanisation as a more or less accomplished fact. Merrifield (2013) was one of the first to take up Lefebvre’s ideas on this matter but with an emphasis on their largely hypothetical status. Brenner (2013) and Brenner and Schmid (2014, 2015) have been much more forthright in pursuing the idea of planetary urbanisation to its logical end, and they should certainly be given credit for the boldness of their approach. Thus, for Brenner (2013, 2018) the whole planet, including not only human geography but also the Earth’s atmosphere, is subject to the forces of urbanisation; and to ward off any ambiguity, he goes on (p. 103) to incorporate within this synthesis:

… villages and peripheralized regions and agricultural zones, intercontinental transportation corridors, transoceanic shipping lanes, large-scale energy circuits and communications infrastructures, underground landscapes of resource extraction, satellite orbits, and even the biosphere itself.

In this passage we are far from anything resembling the urban land nexus. In this view of the world the urban is generated by ‘capital investment, state regulation, collective consumption, social struggle, etc.’ (Brenner, 2013: 96). Urbanisation is further typified as ‘a process of continual sociospatial restructuring’ (Brenner and Schmid, 2014: 750) and ‘a dynamic, relationally evolving … multiscalar process of sociospatial transformation’ (Brenner and Schmid, 2015: 165–166). These processes are unquestionably intertwined with the dynamics of cities in capitalism (Harvey, 1985) but they are pitched at a macroscopic scale in Brenner and Schmid’s account that severely glosses over the all-important intermediate levels of analysis concerning locational logic and the formation of geographic space where agglomeration and the urban land nexus make their entrance into the field of social reality. In particular, and in the light of the earlier discussion, I submit that Brenner and Schmid offer no meaningful insights into the origins, character and internal organisation of cities despite their claims to have forged a new ‘epistemology’ of the urban (Ruddick et al., 2018; Walker, 2015). Their proposal to dissolve away any necessary link between urbanisation as process and the city as a material outcome is notably suspect on both analytical and etymological grounds. In fact, what they are describing – as I shall argue below – is not urbanisation at all but the geography of capitalist development at the world scale.

City limits

One of the key motivations for the theory of planetary urbanisation is the circumstance that cities are virtually never delimitable by a crisp geographical boundary so that they can never be clearly separated from surrounding areas (Brenner and Schmid, 2014, 2015, 2017). This is actually something of a non-problem, but since Brenner and Schmid see the spatial ambiguity of the outer edges of the city as fundamental to their own approach and a justification for extending the idea of urbanisation to the entire planet, it must be dealt with here at the outset.

A first though by no means conclusive observation is that much of reality is, in point of fact, composed of entities that interpenetrate with their spatial surroundings. There is no linear boundary that separates the Sahara Desert from the territories that encircle and interact with it, yet we never use this circumstance to assert that the desert region must extend indefinitely outwards. Nevertheless, and to anticipate a later argument, we may very well assimilate the Sahara together with its wider geographic context into a higher order of synthesis such as a global system of physiographic or climatic spaces.  In the same way, the Earth is clearly distinguishable from the Sun in spite of depending on the Sun for its heat, light and orbital trajectory. Equally, Earth and Sun are elements of a composite solar system that includes but transcends both. Brenner and Schmid go on to argue that just as there is no identifiable boundary between the city and its exogenous domain, so also must the urbanisation process itself operate across the entire planet. If we were willing to accept an all-encompassing definition of urbanisation as ‘a process of continual sociospatial restructuring’, and the like, the Brenner-Schmid argument could no doubt proceed without much further ado. The point is, however, that this definition overstates the case on the one side and leaves too much unsaid on the other, and I shall now expand upon this judgement in three main thrusts.

First, I have sought to demonstrate already that the genesis and spatial organisation of the city derive in essence from agglomeration processes leading to the formal and functional emergence of the urban land nexus. This theoretical construct provides an explicit problematisation of the city as a peculiar kind of social object. By contrast, formulations such as ‘sociospatial restructuring’ or ‘multiscalar processes of social transformation’ fail entirely to capture the substantively distinctive qualities of the urbanisation process or to say anything about the genesis of cities as observable concentrated nodes of social and economic activity on the geographic landscape.

Second, what I have referred to as the exogenous dimension lies, precisely, beyond the centripetal sway of any individual city and the geographically convergent pattern of land uses that it underpins. The urban outside is constituted as a heterogeneous spatial mosaic subject to locational and spatial forces quite at variance with those that operate in the city. The diamond mines of West Africa, the oil rigs of northeast Saudi Arabia, the fisheries of the North Atlantic, the grouse moors of Highland Scotland, the sheep stations of Australia and the banana plantations of Central America, not to mention whatever remains of traditional rural landscapes, peasant societies and untamed spaces, are no doubt subject to ‘sociospatial restructuring’ and all the rest, but they can never be analysed as instances of urbanisation in any meaningful sense of the term. Instead, they each display critical geographic specificities relating to geology, soil, climate, vegetation, internal regional dynamics and social organisation that differentiate them intrinsically from the city. Brenner and Schmid make much of the relationships that link phenomena such as these to cities all around the world, but while linkage – an intermediate phase of actuality between one kind of entity and another – indicates a relation of complementarity it also denotes separation and difference. Linkage (or for that matter, contiguity) is not synonymous with identity, and a fortiori, it is even less a criterion of ontological integration between the city and the rest of global space so long as the former is marked off from the latter by a unique generating process and an archetypical set of empirical consequences, that is, by an urbanisation process based on the dynamics of agglomeration and a structured, spatially convergent urban land nexus. For the same reasons, we need to take with a very large pinch of salt the view expressed by Angelo and Wachsmuth (2014) and Wachsmuth (2014) that the city is a purely ideological construct and that any attempt to conceptualise it in abstract terms entails the transgression of ‘methodological urbanism’.

Third, as already acknowledged, the outward extension of the forces of agglomeration are rarely or never limited by a sharp boundary though their influence does decline with distance from the city centre. By the same token, the interdependent land uses of the city extend outwards in discontinuous fragments but at density levels that steadily vanish as a function of distance-dependent decay in agglomeration economies (transactional interdependencies and economies of scale plus matching, sharing and learning effects). This state of affairs in no way signifies that a meaningful distinction between the inside and outside of the city is impossible. To veer into another language, the theory of ‘fuzzy spatial sets’ (see, e.g. Altman, 1994; Robinson, 2008) provides us with a very credible method for describing how two different series of phenomena can overlap in geographic space without losing their distinctiveness. No matter how wide the zone of overlap may be, the transition from the inside to the outside is traced out in terms of probabilities that indicate degrees of interpenetration and separation. In brief, any given unit of land use on the landscape can be assessed, notionally at least, as to the extent to which its function, its relationships and its location are intertwined with the agglomerative energies of the city. ¹ There is still no linear break between the inside and the outside but we have no option other than to accept the ambiguity of this situation as an inescapable attribute of the world as it is, and to select a simple probability value as defining a discretionary line of demarcation. It follows that the spatially indistinct edges of urban nucleations cannot warrant the conceptual erasure of the city or its liquefaction into global space (cf. Reddy, 2018). More to the point, the city and its exogenous other differ markedly from one another, not only as distinctive social and spatial realms, and hence as objects of scientific enquiry, but also as sites of political action and contestation (Davidson and Iveson, 2015).

The scope of urban theory

Brenner and Schmid make a perfectly sound argument when they point to the increasing sociospatial restructuring, multiscalar transformation and interlinkage of virtually everything in the modern world, but their use of the term ‘urbanisation’ as a synonym for these processes is singularly inapt. Their argument goes on to invoke concepts of social stratification and differentiation, labour exploitation, surplus value extraction, capital accumulation and creative destruction as essential features of the urbanisation process, and while these concepts are obviously relevant to any understanding of the modern world they simply are too general as they stand to act as surrogates for the urban. If we are to construct a coherent framework that operationalises these concepts with respect to both the city and the exogenous mosaic of heterogenous spaces then we need to move to a higher order of synthesis, one, for example, that turns on notions such as the capitalist world system, or the space-economy of capitalism, or, more inclusively, the spatial system of global capitalism and its articulations with other social formations. It is a fundamental category error to conflate these levels of reality tout court with the urban. Nor is this just a matter of etymological violence, so to speak. As we have seen, the epiphenomenon of the city (relative to the human geography of the globe) demands disciplined descriptive focus that respects its cohesion as an integral object of enquiry in its own right.

By extension, and in the light of the present discussion, the city is always a complex amalgam of social relata but is always considerably less than society as a whole. Even if the world’s population were completely urbanised the city would still not account for the whole of social reality, for only those features that are of the city in the specific sense adduced above can properly be said to be urban. This means in turn that Lefebvre’s notion of an explicitly ‘urban society’ is really a sort of rhetorical trope. The qualifier ‘urban’ in this phrase provides no clues whatever about social stratification, or types of economic institutions, or modalities of authority and subordination, or any other core features that typically define a cohesive social organism. Further, a terminal state of entropy would clearly prevail if ever agglomeration economies were to fade away to a point such that human life was diffused over the whole planet. However, entropy implies the disappearance of cities; it is the opposite of urbanisation in any indicative meaning of the word; it is the anti-urban, the anti-city. In every possible sense of the term, ‘planetary urbanisation’ is an oxymoron.

New directions

In the last decade or so levels of controversy in urban studies have intensified greatly, not only as planetary urbanism has come to the fore but also as other facets of the new critical urbanism have staked out prominent intellectual positions in the literature. A large swath of this literature favours notions of the ordinary and the subaltern as keys to the understanding of urban phenomena, and it typically displays an inclination to treat cities as ‘distinctive and unique, rather than exemplars of any category’ (Robinson, 2006: 171). In this perspective, no city is exceptional and all cities are equally worthy of attention, which, so far as it goes, is an irreproachable judgement. In a somewhat analogous vein, Leitner and Sheppard (2016) criticise what they take to be the ‘monadic’ urban theories developed in North America and Western Europe, and they have suggested instead that legitimate forms of urban knowledge need to be calibrated at the ‘provincial’ level. Running in parallel with these advocacies is a deep concern for the specificities of urbanisation in the Global South with its manifest colonial and neocolonial legacies, not only as a corrective to the relative neglect of cities in this part of the world in scholarly research hitherto but also and very plausibly as a potential source of new conceptual and analytical knowledge (see, e.g. Lawhon et al., 2020; Parnell and Robinson, 2012; Robinson, 2012; Roy, 2009; Roy and Ong, 2011). In addition, the new critical urbanism, particularly in its postcolonial and feminist expressions, is typically committed to politically progressive agendas of urban reform, with an emphasis on issues of social marginality, poverty, patriarchy, race and power. My critique of this body of work is carried out in a spirit that is in sympathy with these positive and progressive orientations while attempting to identify blind spots that, as I argue, seem to stand in the way of key insights into the genesis and dynamics of urbanisation and hence eventually of progressive agendas of social reorganisation.

Geographies of theory

There can be no question about the severe neglect of the Global South in urban theory as it emerged in North America and Western Europe over the 20th century. Fortunately, strenuous efforts to rectify this situation are currently proceeding, especially in the postcolonial literature as more and more studies of cities of the Global South are recorded. This trend has had the salutary effect of initiating a search for new concepts of the urban as a countervailing weight to the dominance of Northern theory hitherto. Conjointly, strains of urban theory emanating from the Global North are coming under increasing scrutiny for signs of Eurocentric bias (e.g. Leitner and Sheppard, 2016; Myers, 1994; Roy, 2011; Sanders, 1992). The usual suspect here is the Chicago School of urban sociology (e.g. Park et al., 1925) with its clearly US-centric view of residential neighbourhood formation, but others can be readily identified, perhaps most clearly in neoclassical urban economics à la Alonso (1964) and Mills (1972) whose models of consumer preferences in the matter of land use and housing tend to reflect purely national concerns. Eurocentrism by these scholars is more a matter of silence than an engaged misreading of the Global South, though their ideas have certainly on a number of occasions been incautiously put to work in places such as India and Latin America. As it happens, the same accounts have serious deficiencies even with respect to their home territory in the United States (Roweis and Scott, 1981). Still, adherents to what I have called the new critical urban theory are correct to assert that knowledge is inescapably contextual in the sense that it always has some relation to human interests understood in the broad sense as forms of consciousness and intentionality emanating from the wider social environment. A century and more of work on the sociology of knowledge, from Marx (1973 [1857]) to Haraway (1988), has fully established this idea as a basic precept. Knowledge produced in one social situation may thus sometimes be incomprehensible or inapplicable or misleading when received in another, especially when it is patently contaminated by parochialism and ethnocentrism. Differentials of social power, as well, often stand in the way of undistorted exchanges of knowledge claims, though active critical challenges to this kind of imbalance are currently prevalent. From these perspectives, the diagnostic appraisals of Eurocentrism and institutionalised intellectual authority that are underway in the literature at the present time actually stand firmly within the long tradition of Enlightenment critique (Pagden, 2013). All of this means, conversely, that we need to be equally on our guard against replacing Northern theoretical prejudices by alternative Southern biases (Alatas, 1993; Lawhon et al., 2020; Mabin, 2014).

Despite the many conceptual insights offered by the new critical urbanists, troublesome questions are raised when we consider the position of postcolonial and affiliated urban scholars in regard to the shape and form of theory. There is every reason to welcome the advocacy by these scholars of a cosmopolitan view that is fully open to cities everywhere (Roy and Ong, 2011), yet some objection is in order about their reticence in the matter of theorising across geographical differences and most especially across the globe. Thus, Leitner and Sheppard (2016: 230–231) state unequivocally that ‘knowledge production is necessarily local’ and that ‘there can be no single urban theory of ubiquitous remit’. They do acknowledge that similarities exist between cities in different parts of the world, but this concession is evidently subservient to the higher imperative of rooting urban analysis in some basic framework of geographic regions or provinces. A more chilling caution against theoretical ventures that range beyond the local is conveyed by Vainer (2014: 54) in a curious echo of geographical determinism when he writes that ‘all knowledge inexorably has a location and, consequently, is not universal’. Now, it follows from the earlier discussion of the socially ambient dimension of urban materiality that cities are invariably subject to differentiating influences originating in specific social formations. The net effect of these influences is to ensure that urban clusters are impressed with localised traits unique to particular moments in geography and history. Leitner and Sheppard therefore have a definite point, because this state of affairs means that many opportunities for conceptual investigation of regional or national groupings of cities will virtually always be present. Yet, that being said, and in the absence of any final demonstration that cities are in all respects irreducibly unique, there can be no warrant from epistemology for the belief that translocal or transprovincial concepts about urbanisation processes must always be spurious. Furthermore, and despite the acknowledged fact that theoretical knowledge is often deeply imbued with local inflections (Livingstone, 2014), we cannot assert the general rule that knowledge formulated in one social or geographical context can never be extended to others. The only really defensible stance in regard to this matter is one of critical openness along with the legitimate demand that any claims advanced must be susceptible in principle to disconfirmability.

In practice, cities can be apprehended from many different perspectives and in diverse analytical registers. On the one hand, rich harvests of knowledge can be obtained from investigation of individual urban centres in all their uniqueness and multivariability, as exemplified by the remarkable monographic studies by Banham (1971) on Los Angeles, Gaillard (1976) on Paris, and De Boeck and Plissart (2005) on Kinshasa. On the other hand, the combinatorial complexity of cities can certainly be disentangled in different ways so that widely ranging theoretical descriptions can be extracted. Every intermediate level of analysis between these two extremes is also susceptible to investigation, including not only provincial groupings but also translocal taxonomic categories such as industrial cities, financial centres, tourist resorts or cultural centres. In a word, no incongruity exists between the fact that every city has a unique personality and the notion that certain substrata of urban reality (such as the urban land nexus) share in common genetic drives. Even so, an obvious proviso is necessary at this point. As illustrated by the disagreements between Ghertner (2014) and Bernt (2016) on gentrification in India, crucial interpretative differences are always liable to arise when data on cities from heterogeneous historical and geographical contexts are brought together. When we consider these data in their raw form, they often exhibit what appear to be strong dissimilarities. More disciplined analytical systematisation, however, will sometimes reveal traces of remarkable equivalences submerged or only partially visible in individual cases. The point can be illustrated in general by reference to the flow dynamics of rivers, or the tendency of profit rates to fluctuate around a common value in disparate sectors in disparate national capitalisms, or the operation of agglomeration processes in the case of cities. There is no reason why ontological realism and theoretical abstraction should not be pertinent to the physical and the social worlds alike.

With the latter remark in mind, ample opportunities clearly exist for fruitful theoretical analysis of urban data across geographic space. Five brief examples may be cited. First, modern cities are all endowed with central business districts marked by broadly similar functions, unusually high levels of density and a common genetics derived from superior levels of accessibility within the urban land nexus as a whole. Second, industrial districts in urban areas across the Global North and South exhibit strikingly similar organisational features despite wide variations in types of output, market power and policy milieu (Oyelaran-Oyeyinka and McCormick, 2007; Scott, 2005; Scott and Garofoli, 2007). Third, basic land-use intensification processes in relation to rent are certainly comparable in Los Angeles and Lagos notwithstanding the fact that they are also differentially shaped by contrasting contextual variables. Fourth, increasing access to private and public transport opportunities is conducive virtually without exception to outward urban sprawl. Fifth, poverty and social distress in Dickensian London ² or today’s large informal settlements of Roma people in Eastern European cities share much in general with the same problems in Manila and São Paulo. Indeed, counterfactual speculation suggests that extensive bidonvilles would in all likelihood arise in American cities if the poor from Africa, Asia and Latin America were allowed unrestricted entry into the country. In addition to these five points, global city-regions on every continent, despite their conspicuous idiosyncrasies, exhibit a number of remarkable similarities ranging from pervasive downtown redevelopment via land-use intensification and aestheticisation to vigorous post-suburban expansion and outward sprawl (Scott, 2019). Moreover, detailed testimony to the spread and socio-political importance of city-regions in the Global South is coming increasingly not only from a few ‘iconic’ cases in the Global North, as Robinson and Roy (2016) aver, but from the Global South itself in research carried out by scholars whose home bases lie in Africa, Asia and Latin America (e.g. Agyemang et al., 2017; Cheruiyot, 2018; Klink, 2001; Pasin and Oner, 2018; Yeh and Cheng, 2020) and I see no reason to dismiss these scholars as just passive dupes of ‘Northern theory’.

A methodological impasse?

The deeply rooted reservations expressed by many critical urbanists about Northern theory, translocal analysis and, above everything else, ‘universal’ concepts, reappear in a very evident resurgence of empiricist and conjunctural approaches to research on the city. This syndrome has been diagnosed as ‘a new particularism’ (Scott and Storper, 2015; see also Smith, 2013), much but not all of which can be traced back to the ordinary cities programme with its clarion call to treat all cities as ‘distinctive and unique’ and its systematic mistrust of ambitious concepts that attempt to comprehend urbanisation as a general process.

This broad retreat from theoretical systematisation has found widespread expression in the rise of comparativist methodologies and their broad adoption by new critical urbanists. Much of this trend has followed in the wake of the influential recommendations of Robinson (2006) who states that whereas every city is an irreducible ‘category of one’ (p. 109) certain kinds of insights may still be obtained by juxtaposing selected cases and ‘thinking across their differences’. Comparativism, as it happens, is a time-hallowed methodology in urban studies with a proven record of usefulness and versatility (Ward, 2010), though with definite intellectual limitations when it involves no more than simply placing empirical observations side-by-side and then inspecting the result for similarities and differences. That said, its analytical power can be enhanced by augmenting the elementary mechanics of raw comparison with various diagnostic extensions (see, e.g. Abadie and Gardeazabal, 2003; Tilly, 1984). It is perhaps for this reason that Robinson (2016) has recently offered a more ambitious statement about comparativist methods by bringing in the notion of ‘elsewhere’, defined not only as other cities but also as whatever abstract notions might appear to be relevant to the case or cases in hand.

In this formulation, analysis can begin anywhere, whether it be an intuition, a prior concept or theory, an individual case, or a set of repeated empirical instances. Analysis then proceeds by means of a to-and-fro involving, on the one hand, the adoption and recursive readjustment of an analytical starting point (empirical or conceptual) and, on the other hand, the selection and reselection of cases for comparative assessment as the study progresses. This brief characterisation does less than justice to Robinson’s extended presentation but her argument in the end boils down to this essential core, with the proviso that her methodological recommendations appear to lean rather heavily on inductive reasoning from empirical cases. Her main points are (a) that ‘we can reimagine comparisons … through elsewhere (another case, a wider context, existing theoretical imaginations, connections to other places)’; (b) that we should think of ‘urban theory as multiplicity’; and (c) that ‘a reformulated urban comparativism insists on keeping open the possibility to draw any urban places, experiences, and events into … conversations about the nature and future of the urban’ (Robinson, 2016: 22–23). In one sense, these statements can be taken as laudable pleas for open-mindedness and intellectual flexibility. In the context of Robinson’s overall argument, however, they lean heavily towards the somewhat less instructive idea that cities are little more than amorphous masses of social matter. Some of these masses may share certain features in common but to the degree that Robinson offers any hints as to what cities might be in substantive reality, it is only to intimate that they are marked by never-ending empirical complexities open to analysis from innumerable standpoints. In an indeterminate world such as this, eclecticism is perhaps the best and only way forward. Robinson’s argument resolutely, and no doubt intentionally, eschews any suggestion that cities may have identifiable archetypical qualities that actually call for inferential, process-oriented investigation. There is, therefore, no recognition that the urban might present the methodologist or the theorist with very special kinds of theoretical and analytical questions beyond the vague quandary of how to bring different empirical cases ‘into conversation’ with one another. Her work thus provides a number of interesting observations on methodology – as well as playing an important role in bringing the cities of the Global South to wide scholarly attention – but in its open-ended syncretism it refrains from and actually inhibits penetrating analysis of the logic and dynamics of cities in general. Many different kinds of urban research programmes are clearly amenable to comparativist methodologies but we must go well beyond comparativism itself if we are to deal with basic issues of the theoretical ramifications of urbanisation as a generalised process above and beyond its immediate impression on the senses as an assortment of diverse cases and unremitting complexities.

For its part, the provincialist approach to urban analysis declares as a matter of principle that any theoretical exercise at any scale of scrutiny above some restricted geographic area is doomed to failure. I have already suggested that the imposition of this constraint on enquiry is unnecessarily self-limiting, but there is another way in which it prematurely curtails the investigative urge. Urban research, sooner or later, must always face the evidence of the real. Obviously, empirical evidence of any sort is usually far from being immune to contestation but it is, in the last resort, a more or less widely accepted court of appeal. Quite simply, the imposition of a priori limits on the geographic range of any theory, urban or otherwise, is gratuitous when more compelling evidence-based means can be found for assessing if and when its testimony remains within or has gone beyond the bounds of its explanatory potential. Likewise, as Peck (2015) has written of the congenital antipathy on the part of post-colonial scholars towards Northern theory, the legitimate point of critical scrutiny in regard to any given set of knowledge claims is not their place of origin but their explanatory successes and failures.

Response to critics

The critique of critical urban theory laid out in the preceding sections is at once an offshoot from the conceptual analysis with which this paper began and a partial attempt to justify the overall theoretical argument. Many of the essential points of the analysis (as contained in Scott and Storper, 2015; Storper and Scott, 2016) have already been subject to extensive critical examination in the literature and therefore – in pursuit of yet further clarification of the claims presented here – I now round out the discussion with replies to some of the leading strictures that have been levelled at our work. Four main points need to be made.

First, Robinson and Roy (2016) allege that our analysis essentially strips away all the diversity and specificity of cities in the Global South. True enough, we postulate a general theoretical model that represents the twofold intellectual construct of agglomeration and the urban land nexus as a universal concrete abstraction, and we claim that this abstraction is a fundamental key to understanding how cities everywhere develop and grow. So far from erasing the complexities of individual cities, however, the analysis is open to a virtual infinity of social, cultural and political declensions even as it establishes an internal spatial dynamic that interacts with these variables to form the living city. This openness is reflected in the ways in which the ambient and exogenous dimensions of the city intersect with the urban land nexus as well as in the formation (and semiotic properties) of what I earlier referred to as climactic land uses. Reasoned disagreement with the analytics of our model is one thing but the claims of Robinson and Roy that our arguments obliterate difference are false. As an aside here, I should mention that Ghertner (2020) is mistaken in thinking that the theory of the urban land nexus is addressable only to situations where private property prevails.

Second, and in much the same vein, Parnell and Pieterse (2016: 236) assert that our work is dismissive of African urban realities, which are, they claim, ‘characterised by economic informality, multiplicity, marginality and dispersion’. Unfortunately, Parnell and Pieterse provide no details as to why or how these variables represent key entry points to an understanding of the internal dynamics of African cities. In order to achieve this goal, the chosen variables must be set in some conceptual context that is capable of showing how they emerge, interact and generate urban entities. Our own theoretical model is quite susceptible to calibration in ways that accommodate these matters (including dispersion, assuming that this term means something like spread out, or, probably more to the point, partially spread out). Hence, even though our model has little or nothing to say explicitly about the substantive details embodied in informality, multiplicity, marginality and dispersion these variables are nevertheless intertwined with the agglomeration process and the urban land nexus through their instantiation within specific centres and their functions as interdependent land uses.

Third, Mould (2016) and Roy (2016) allege that our analysis erases historical difference, forecloses multiple concepts of the urban and is permeated by economism. I have already indicated how our theoretical work accommodates the myriad ways in which different social formations in different historical and geographical contexts come to inhabit the urban land nexus, as it were, and to make it over in endless varieties of realised forms (see also Randolph and Storper, 2020). To be clear, our model insists on the primacy of agglomeration and the urban land nexus as the fundamental elements of the urbanisation process, but this basic framework is also open to modification by other social and cultural forces that leave a definite stamp on the form and function of the city. Moreover, there is no reason whatever why this particular approach necessarily excludes other ways of conceptualising urban realities, and following Addie (2020) and Sullivan (2017) among many others, I acknowledge that complementary and mutually compatible ways of framing problematics of urban life and form can always be constructed. By way of illustration, perspectives on the city as an arena of class struggle, of everyday practices and identities, of gender-based discrimination and of symbolic landscapes all represent highly plausible lines of enquiry and, what is more, are all reconcilable with the notion of urbanisation as an expression of agglomeration processes and the dynamics of the urban land nexus. At the same time, neither Mould nor Roy explain with any clarity what the term ‘economism’ is supposed to mean, and so I am left as best I can to infer that what they really intend to suggest is that we are attempting to account for non-economic aspects of social and cultural life in cities on the basis of economic reasoning. The simple response here is that our model does not seek to account for the substantive social and cultural attributes of cities beyond the fact that they are bound up in reciprocal and non-deterministic relationships with spatially inflected variables such as density, proximity, centrality, convergence, encounter and exclusion (cf. Simmel, 1950 [1903]; Wirth, 1938). If, on the other hand, Mould and Roy are saying that our conception of the city as a phenomenon rooted in agglomeration and the formation of the urban land nexus can be negated by consideration of non-economic social and cultural processes, then the onus is on them to provide a reasoned demonstration as to how this might be the case.

Fourth and last, Walker (2016) takes us to task for under-emphasising the role of the economic surplus in the formation and growth of cities. As I have already suggested, an agricultural surplus is indeed a necessary but far from sufficient condition of urbanisation. More generally, the agricultural surplus makes agglomeration possible but it is not tantamount to the agglomeration process itself. Walker’s notion of the surplus actually goes beyond agricultural products to include goods and services of any kind. He then argues that clustered settlements arise in many instances because elite members of society congregate together in geographic space where they can deploy these surplus quantities in symbolic elaboration and social display. However, in the absence of massive localised economies of scale and scope, it is difficult to see how groups of sybaritic parasites, even when augmented by a coterie of minions, could then blossom into substantial centres of population. Even in the case of early Aztec cities, urban nuclei functioned mainly as market centres ruled over by petty kings. With the passage of time, spectacular imperial and ceremonial developments were integrated into these nuclei and thus became active elements of the city. However, Aztec cities were sustained as viable socially differentiated clusters by significant cohorts of craft producers, service providers, market traders, merchants and a large labouring population including slaves (cf. Smith, 2017). Without any doubt, important social and economic interactions occurred between representatives of the imperial-cum-ceremonial functions of the city and the rest of society, but always in the context of mutual spatial convergence thus forging this peculiar version of agglomeration into a dynamic integrated urban land nexus. Analogous arguments might be advanced for other non-capitalist forms of urbanisation, such as European and Japanese castle towns, for example.

Last things

I have tried in this account to present a theoretical overview of the origins and dynamics of cities by highlighting the processes by which urban nucleations are established in geographic space and how this then leads to the emergence of localised constellations of interrelated and spatially convergent land uses. I have referred to any such constellation as an agglomerated urban land nexus or the internal dimension of urban materiality. I have argued that both a socially ambient and an exogenous dimension also shape the materiality of the city, the former by reason of its decisive effects on the entire social and political spectrum of urban life, the latter by reason of its role as a source and destination of flows into and out of the city. Out of these different structuring dynamics, the city makes its historical and geographical appearance as a unique type of social object and a uniquely constituted site of communal existence and political action.

In the light of this analysis, I have also sought to evaluate a number of theoretical claims that have been advanced by the new critical urbanism and that have exerted considerable influence over the literature of late years. I hasten to add that there is much in this critical urbanism that is insightful, informative and useful. It brings the hitherto much neglected cities of the Global South forcefully into view as subjects of empirical investigation and, notably, as potential fountainheads of new theoretical findings. It offers a rich harvest of insights into questions of identity, culture and ‘the everyday struggles of people, of life as it is lived in relation to the urban’ (Ruddick et al., 2018: 388). Above all, it maintains a clear focus on the predicaments of urban existence for marginalised, impoverished and subjugated populations and on practical and political possibilities for emancipatory reform. For all that, the literature of critical urbanism, as I have sought to demonstrate in this paper, is marked by important lacunae with respect to the logic of urbanisation and the concept of the city, and while its legitimate though ultra-cautious preoccupation with the singularities and contingencies of cities helps to guard against over-hasty transfer of analytical findings and policy insights across different regions, provinces, nations and hemispheres, it also actively inhibits constructive mutual exchanges in these matters. If urban geography is to recover a vigorous analytical base and to serve progressive political objectives, as it must, it needs also to recover a resolute view of cities not only as uniquely complex empirical entities with locally determinate fields of meaning and action but also as domains of scientific investigation, distinct from but embedded in planetary capitalism, and accessible to theoretical enquiry at the highest level of generality and inclusiveness.