The Janus-faced genius of cities

Departing from Storper’s (2013) notion of a ‘genius of cities’ but extending the concept from agglomeration economies to inter-city networks and the built environment as foundations of cities’ genius, I argue that cities’ genius is Janus-faced. My contention is that cities’ specific environments not only breed all the ‘good’ innovations that drive innovation and growth but they also generate the ‘bad’ ones, which allow for the development of the means of exploitation. Cities are, as a result of their very properties, key places for the organisation of uneven development.

Introduction 2013; Storper and Scott, 2016;Taylor, 2013;van Meeteren and Bassens, 2016;Walker, 2016). 1 While a number of issues are being discussed controversially, the positive role of cities in innovation and economic development continues to be a leitmotif of current urban studies, as it was in the heyday of modernisation theory. 2 The acknowledgement of cities as the 'mothers of economic development' (Jacobs, 1997), of urbanisation's 'efficiency-generating qualities via agglomeration' (Scott and Storper, 2015: 4), of the 'city as innovation machine' (Florida et al., 2017) and of the 'urban ability to create collaborative brilliance' (Glaeser, 2011: 8) are pervasive. Not even post-colonial scholars critical of the universalisation of ideas originating in the Global North (e.g. Robinson, 2006) have challenged the notion of the 'almost universal positive association' (Brockerhoff and Brennan, 1998: 82) between urbanisation and economic development. It is not surprising, then, that 'development' is one of the terms most frequently associated in urban research with the words 'urban', 'city' or 'cities' (Kirby, 2012). Accordingly, the economic power of cities has become something of an ide´e fixe among development agencies and consulting firms (e.g. World Bank, 2009).
Cities are, thus, 'extraordinary' (Taylor, 2013); they have, according to Storper (2013: 9f), a particular 'genius'. For Storper, this genius arises from agglomeration effects. Cities' 'many overlapping worlds of face-toface contact' provide a 'behavioural context in which different elements of know-how come together', which in turn promotes economic growth. I adopt Storper's notion of a genius of cities but I add two more elements as its foundation, namely inter-city networks and the built environment (section 'The genius of cities'). Moreover, while Storper (2013) positively connotes 'the genius of cities' throughout, I claim that it is Janus-faced, that is, 'looking or acting in opposite or contrasting ways' (Merriam-Webster, 2020). Though I agree that cities are drivers of economic growth, nevertheless I am concerned by the developmentalism often implied in this notion. In a world of poverty, inequality and exploitation, I simply ask myself where is all the misery coming from. Is it really so simple that poverty persists merely because urbanisation does not progress rapidly enough, as the former World Bank economist Ravallion (2007) suggests? Is searching for the causes of poverty really an intellectual dead end, as the ultimate advocate of cities' ingenuity, Jacobs (1970: 121), says, 'because poverty has no causes. Only prosperity has causes' (namely cities)?
Against such developmentalist views, I argue that poverty does have causes, which include as one of the most important ones an exploitation across space. In section 'The ''differentiated growth'' thesis: a critique', I take issue with the common notion that capitalism's uneven landscapes result primarily or even completely from spatially differentiated growth. While geographically divergent levels of economic growth are an important aspect of uneven development (World Bank, 2009), as their main explanation they are insufficient. Inter-regional and inter-national wealth transfers -adding to and nourishing differentiated growth dynamics -have to be taken into account, too. Such relational approaches have long been conceptualised, among other things, as 'metropolitan labour aristocracy' (Emmanuel and Bettelheim, 1970;Cope, 2019), 'unequal exchange' (Emmanuel, 1972;Amin, 1976), the 'appropriation of surplus of the whole world-economy by core areas' (Wallerstein, 1974a: 401;cf. Frank, 1969), the 'geographical transfer of value' (Hadjimichalis, 1984) and relational impoverishment (Elwood et al., 2017). Despite their different theoretical underpinnings, these accounts all elucidate that uneven development at regional and global scales is a socio-spatial process, at least in part the result of deliberately initiated wealth transfers across space, not just selfreinforcing agglomeration dynamics. This draws attention to the actors who set these geographical wealth transfers in motion, to their practices and to the places from where they operate. Astonishingly, though, and contrary to what one might have expected from the great importance attributed to cities for understanding economic development, the city hardly plays any role in 'connectivity-based explanations' (Sheppard, 2016: 158) of uneven development.
Against the background of this failure to view cities as analytical categories for studying the asymmetrical relationships that constitute the capitalist division of labour, I propose to apply knowledge of the extraordinary innovative power of cities (see section 'The genius of cities') to the analysis of uneven development. My claim is that the genius of cities is Janus-faced, working in opposite ways. The unique potential of cities -as agglomeration economies, inter-city networks and the massiveness of the built environment -is the source of both 'good innovations', which bring about efficiency and growth, and 'bad' ones that deprive people around the world of their livelihoods (see section 'The Janus-faced character of the genius of cities'). 3 However, while urban problems have been studied extensively, things are different with the dark side of cities' genius, which has not been explored systematically.
Yet, the idea to conceptualise cities' role in economic development beyond straightforward developmentalism is, of course, not new, as the debates from the heyday of modernisation and dependency theories, which I revisit in the section 'Cities as ''suction pumps''', show. Although the urbantheoretical underpinnings of reflections on cities' 'parasitism' (Hoselitz, 1955), their role in 'internal colonialism' (Gonzalez Casanova, 1965) and polarised development (Friedmann, 1967) and the 'urban bias' of national politics (Lipton, 1977) were usually only hinted at, these debates provide important clues as to why cities should be treated as an analytical category for understanding the production of uneven development.
More abundant, clearer indicators of the Janus-faced character of cities' genius can be found in Fernand Braudel's (1983Braudel's ( , 1984Braudel's ( , 1985 assessment of the rise of capitalism (section 'Fernand Braudel: Cities as aggressive worlds'). Though enthusiastic about cities' energy, Braudel also points to their dark sides; the city of medieval Europe, for example, was 'an aggressive world and an active force for unequal exchange ' (1984: 94). It is equally important that Braudel unequivocally links this aggressiveness of the cities to their very properties, namely agglomeration, external networks and massive physical infrastructure.
Treating cities as 'proactive' (Taylor, 2013: 219) nodes in geographical wealth transfers is based on the theorisation of space 'as an ''active moment''' within accumulation processes (Harvey, 2006(Harvey, [1982: 374; emphasis added). Yet, such thinking has been highly controversial since it became popular in the critical social sciences. In section 'Do cities exploit other cities?', I come back to this debate, arguing that assuming an 'explicit spatial problematic' (Soja, 1980: 219) in uneven development is not a fetishisation of space (cf. Harvey, 2006Harvey, [1982). Rather, because cities are bases of resources, the mobilisation of which is crucial for the exercise of power (Allen, 2003(Allen, , 2005(Allen, , 2010, cities become strategic places for exploitation across space without, however, themselves exploiting other cities or regions. The concluding section discusses some implications of my argument for ongoing debates around specific concepts in urban studies.

The genius of cities
To explain the 'growth-engine' effects of cities, Storper (2013: 9, 227) refers to their particular 'genius'. Agglomeration, the bringing together of many -and many differentface-to-face contacts in a small space creates a specific 'behavioral context', a 'supply architecture' apt for breeding innovation. The resulting '(p)articularly strong forms of local interaction amount to a form of local genius'. While Storper emphasises the local context and thus the site specificity of cities' genius, he also implies that there is some form of extra-local urban virtuosity. While the particular form of the genius loci might vary from city to city, there is, across all local differences, an essence of the extraordinariness of cities, namely agglomeration economies. And indeed, the notion that agglomeration is a fundamental component of economic development is a truism in urban studies, economic geography and beyond that has remained valid at least for the capitalist city since the 18th century (Jacobs, 1970;Krugman, 1996;Marshall, 1890;World Bank, 2009). 4 While agglomerations generate many positive externalities, perhaps the most important one is that cities' density and diversity facilitate and even enforce innovation, because 'learning is a form of interaction' (Storper, 2013: 96).
To agglomeration, I add two more aspects as constituent of cities' genius. First, there are the externalities arising from intercity networks. Following Jacobs' (1970) and Pred's (1977) emphasis on cities' external relations to other cities (rather than to rural hinterlands), Taylor (2004: 1) claims that '(c)onnections are the very raison d' eˆtre of cities'. Inter-city networks, based on complementarity and cooperation rather than on competition, are so essential to understanding cities that Taylor et al. (2010) even speak of a 'central flow theory'. Inter-city networks provide access to information, hitherto unknown skills or goods, capital seeking investment, or simply labour needed for economic expansion. Accordingly, cities' local 'buzz' (Storper and Venables, 2004) requires what Bathelt et al. (2004) have called 'global pipelines' to be set in motion. Moreover, Allen (2005: 221, original emphasis) emphasises that without interaction in inter-city relationships, cities could not exert influence over others -'the networks are themselves constitutive of power'.
Second, the massiveness of cities' built environments is a further source of their genius. As Harvey (1985) has pointed out repeatedly, the built environment is the physical infrastructure for production, circulation and consumption. Cities are 'intense emplacement(s)' (Walker, 2016: 171) of houses, streets, factories, ports, schools, office towers, airports, warehouses, power supply systems, universities, etc. and as such, they provide an enormous, localised complex of human-made resources that serves as the physical foundation for capitalist accumulation (cf. Smith, 2008Smith, [1984).
Taken together, accounts that focus on the externalities resulting from cities' specific features provide a compelling theorisation of cities' extraordinary economic vigour. Agglomeration economies, network externalities and massive built environments create a specific socio-physical setting conducive to growth, the quality of which is 'intrinsically urban in character' (Scott and Storper, 2015: 9; emphasis added). Things happen in cities because of cities' properties. Therefore, cities are not mere stages of development, but critical inputs into it (Massey et al., 1999), which makes them a central analytical category for understanding how -and where -the wealth of nations is produced (Jacobs, 1985).

The 'differentiated growth' thesis: A critique
To a certain extent, cities' economic dynamism explains uneven development at regional and global scales. As they thrive economically, cities produce socio-spatial inequalities: 'Cities are key . to greater wealth creation, but they involve equity or distributional effects: higher average incomes in some places than others' (Storper, 2013: 49). Moreover, because agglomerations' positive effects become selfreinforcing because of processes of circular and cumulative causation (Myrdal, 1957), the social inequalities across space once created will increase over time, even in the absence of other intervening political or economic factors. Jacobs (1970: 167) probably expressed the differentiated growth view most pointedly, exemplifying her vision of uneven development in a city-driven economy with a popcorn machine: '(N)ot all the kernels are popping simultaneously; but all the time corn is popping'.
In Marxist analyses, geographically differentiated growth rates do not represent, as in the Jacobsean view, 'a ''gap'' between more developed and less developed regions' (Smith, 2008(Smith, [1984: 203). Rather, 'leftbehind' regions are actively created by capitalism's polarising development, particularly through capital's 'seesaw movements' (Smith, 2008(Smith, [1984). Yet, many Marxist analyses nevertheless prioritise geographically differentiated growth over exploitation across space as an explanation of uneven development. In that context, the city assumes key analytical importance, similarly to Storper's analysis, although it is argued differently in theoretical terms. Since the concentration and centralisation of capital find their ideal geographical expression within the city, centripetal dynamics -and hence uneven development -are created and reinforced: 'The stock of fixed capital and of consumption fund assets does . provide a solid form of wealth that can be used to produce and consume more wealth' (Harvey, 1985: 145;emphasis added).
While geographically divergent levels of economic growth are an important aspect of uneven development (World Bank, 2009), there are good reasons to reject this notion as the preferred or even sole explanation of uneven development. Relational approaches that focus on the simultaneous and parallel production of wealth and poverty as well as on geographical dimensions of capitalism's antagonisms have long attracted attention from Marxist and non-Marxist theorists (Lenin, 1982(Lenin, [1917; Luxemburg, 1951Luxemburg, [1923; Bukharin, 1929;Myrdal, 1957;Frank, 1969;Emmanuel, 1972;Wallerstein, 1974a;Amin, 1976;Mandel, 1978;Braudel, 1984;Hadjimichalis, 1984;Hadjimichalis, 1984;Hadjimichalis, 1984;Raffer, 1987;Arrighi, 1994;Harvey, 2003;Elwood et al., 2017;Cope, 2019). Despite the different conceptual focuses of these accounts, 5 their common denominator is that wealth transfers across space -inter-regionally or internationally -have historically been and remain a key mechanism for uneven development, adding to and nourishing differentiated growth dynamics. Thus, exploitation does not only happen locally, in the antagonistic relationships between capitalists and workers but also in 'asymmetric connectivities' (Sheppard, 2016) between places and regions within the capitalist division of labour. The reawakening interest in uneven development in economic geography (Barnes and Christophers, 2018;Phelps et al., 2018;Sheppard, 2016) and, in particular, in research on commodity chains (Bair, 2014;Quentin and Campling, 2018;Selwyn, 2019;Suwandi, 2019;Werner, 2018), underscores the insight that asymmetrical and exploitative relations between classes and also regions are 'endemic, not transitory, features of capitalist development' (Peck, 2017: 7270). The uneven distribution of rewards and their geographical transfer are key modi operandi in capitalism.
Surplus transfers across space contribute to regional/global uneven development directly, since the total mass of profit available for re-investment is geographically redistributed, and indirectly, because this redistribution of profit further exacerbates the unevenness resulting from differentiated growth. While in some regions the control over extra surplus allows for accelerated paces of accumulation, regions deprived of their own surplus are forced into slower rhythms. The resulting socio-geographical polarisation has been conceptualised as 'corification' and 'peripheralization' (Arrighi, 1990), or as the spatialisation of class relations on a global scale (Li, 2001). More recently, the cross-border links that create distinct, albeit connected geographies of enrichment and impoverishment have been addressed by the relational poverty literature (Crane et al., 2020;Elwood et al., 2017). Against Jane Jacobs' dictum that poverty has no reasons, this literature stresses the process of impoverishment -actors with different levels of economic, political and cultural power relationally produce poverty.
The Janus-faced character of the genius of cities In many assessments of uneven development, particularly those based on the 'differentiated growth' notion, the city is the central explanatory category. It is where the geographical centralisation of capital finds its ideal expression and where agglomeration economies develop (Harvey, 1985;Jacobs, 1985;Storper, 2013). By contrast, in 'connectivity-based explanations' (Sheppard, 2016: 158) of uneven development, cities have been attributed little, if any, explanatory potential. In the main contributions to the debate on 'unequal exchange' (Emmanuel, 1972), core-periphery relations (Wallerstein, 1974a) or the 'geographical transfer of value' (Hadjimichalis, 1984), the city remains largely absent as an analytical category (but see Frank, 1969; see next section). 6 The same is true for most of the more recent work on the unequal appropriation of surplus along commodity chains (but see Brown et al., 2010;Parnreiter, 2019;van Meeteren and Bassens, 2016) and for the relational poverty literature.
Not paying attention to cities' role in wealth transfers across space is a mistake, however. If such transfers are, as Hadjimichalis (1984: 342) stresses, 'not an automatic process', a simply spatialised version of the sectoral transfer of value between departments with higher and lower organic composition of capital, as analysed by Marx, then questions arise about the agencies, practices and geographies of the organisation of these wealth transfers. And, more specifically, when poverty is 'constituted through multiple, interrelating spatiotemporal processes' (Elwood et al., 2017: 754) at multiple scales, then, of course, it comes to mind to think of cities. In this vein, Parnreiter (2019) pointed out the crucial role of producer service firms and thus of global cities in the practices of the transfer of value. This argument requires, however, theoretical deepening. If wealth transfers across space are vital for accumulation, if they result from deliberate interventions and if specific cities are strategic places for these interventions, then an adequate theorising of cities' role in regional/global uneven development can neither be satisfied with the well-known fact that urbanisation throughout history has implied the spatial concentration of surplus nor with discussing geographically differentiated growth rates. I therefore fully agree with Walker's (2016: 169) criticism of Scott and Storper's (2015) account of the 'Nature of Cities' in which he objects that '(t)hey sidestep entirely the issue of power and exploitation . their theory of cities is almost entirely ''horizontal'' ., and not at all ''vertical'', or about the operation of social hierarchy, corporate profit-making or imperial conquest'. This criticism can be extended to recent urban research more generally, which pays too little attention to the exploitative relations that connect cities to other cities, towns and rural areas (Dunford and Liu, 2017).
To overcome this shortcoming, cities should be conceived of as 'proactive' (Taylor, 2013: 219; emphasis added) nodes in intercity networks and commodity chains, with special attention given to their role in producing and sustaining those uneven relationships that underlie wealth transfers across space. This claim rests upon the assumption that the city -as a specific, spatial configuration, endowed with particular properties that impact on economic action -has something to do with some actors' capabilities to redirect and appropriate surplus flows. Building on Scott and Storper's (2015: 9) useful distinction between issues found in cities and those which are of cities, in the sense of being 'intrinsically urban in character', I contend that the complex organisational know-hows that underlie the exploitative organisation of commodity chains are not only created in cities, but emerge from cities, developed on the basis of their extraordinary social and physical environments. What constitutes the foundations of cities' exceptional capacities for economic development -agglomeration economies, inter-city networks and the massiveness of the built environment -also provides the fertile ground for the development of those 'weapons of domination' (Braudel, 1984: 35) through which uneven development is organised. 7 In other words, I propose that the genius of cities is Janus-faced. While not common in today's efforts to theorise the city, assessments of the dark sides of cities' role in capitalist development are not unknown in urban research. Two strands of the literature, which are worth remembering and invoking in current debates, will be reviewed in the next sections: the debates from the heyday of modernisation and dependency theories about 'parasitic cities' (Hoselitz, 1955), 'internal colonialism' (Gonzalez Casanova, 1965), polarised development (Friedmann, 1967) and an 'urban bias' (Lipton, 1977) of national policies in peripheral countries, and Braudel's (1983Braudel's ( , 1984Braudel's ( , 1985 assessment of the role of cities in the rise of capitalism. All these contributions raised, in one way or another, the issue of whether and how (elites in) poorer countries' capitals or cities in core countries were drawing off resources from cities in global or national peripheries and from rural areas.

Cities as 'suction pumps' 8
From the 1950s to the 1980s, high rates of urbanisation in the de-colonised or decolonising world, an often primate city distribution, pronounced polarisation between urban and rural areas, and widespread urban informality in the big cities raised, as Castells (1977: 39) put it, 'practical, that is to say political' concerns about the relationship between development and urbanisation. Scholars began to question the mainstream modernisation-theoretical view that urbanisation and economic development were interdependent processes in any case. Hoselitz's (1955) discussion of 'Generative and Parasitic Cities' opened the debate. For Hoselitz, the decisive point was not that a city extracted surplus from surrounding regions; as a modernisation theorist, he took that as a prerequisite for economic development. Hoselitz's concern instead focused on how cities used this soaked-up surplus. While he assumes that it is usually multiplied, contributing thereby to overall national growth -hence, the designation as 'generative city' -he also acknowledges that there can be cases where such an evolutionist perspective on urbanisation does not apply. During colonialism, economic development remained limited to a central city without stimulating growth in the entire region, for which reason the major city is 'parasitic'. Importantly, Hoselitz argues that cities become parasites when and because they have functioned as bridgeheads for colonial interests. 9 As the dominant powers' 'gateways to the hinterland which they dominate', main cities have enabled the 'excessive depletion of natural resources, and the exploitation of peasants and other primary producers' (Hoselitz, 1955: 284, 280). Thus, while themselves 'subordinate to the great centres of world trade' (Hoselitz, 1955: 284), colonial cities reproduced the global unevenness on a regional scale, functioning as 'hinges' between the local, regional and global scales of colonial exploitation. Accordingly, 'internal' and 'external' unevenness overlapped and reinforced each other. It is this linking of the regional dimension of exploitation with cross-border surplus transfers that gives Hoselitz's 'parasitic cities' thesis the theoretical acuity that it still has today. Friedmann's (1967Friedmann's ( , 1973 assessment of polarised development is also worth recalling. While sharing modernisation theory's belief that the diffusion of innovation and economic growth works through a country's urban hierarchy, Friedmann breaks away from the belief in automatic trickle-down effects. Contrary to wisdom established by modernisation theory, he contends that core regions -the large cities -may be unwilling to use urban-generated surpluses to develop the whole country. Rather, they 'impose a condition of organized dependency on their peripheries . (they) organize the periphery as a continuing source of supply of raw materials, food stuffs, and semi-processed goods, . (and they) organize the periphery as a set of market areas' (Friedmann, 1967: 23, 30; emphasis added). Friedmann thus thought of socio-spatial polarisation not as a temporary by-product of development but rather as an active process, as the generation of uneven development prompted by a country's core region. He maintained that 'developing countries' typically involve spatial systems that reflect 'a dominant and persisting pattern of nonreciprocal exchange relations among cities and regions' (Friedmann, 1973: 38;emphasis added).
What makes Friedmann's contributions so important for today's urban theorising is that he draws attention to the geographical concentration of tangible and intangible resources in cities that gives elites an edge over other actors. He particularly stresses that spatial proximity makes the powerful more powerful: 'Politicians, bureaucrats and businessmen mingle in exclusive social clubs and the city's top restaurants, send their children to private schools (or the national university), and form tight social networks of their own'. 10 Thus, a 'base of power is solidified within core regions' which allows for political and economic control relations by which dependency is secured (Friedmann, 1973: 18f, 39). Second, Friedmann (1967Friedmann ( : 18, 1973: 26) depicts a country's spatial system as a 'structure of authority-dependency relations' in which a 'hierarchy of urban centres [is] exercising control over both national and regional economies'. Cities' external relations are, thus, not necessarily of mutual benefit, as for example Jacobs posits. Rather, they embody the organisational framework for uneven development which at some point becomes even 'dysfunctional' (Friedmann, 1967: 33) to a country's general economic development.
A third argument worth recalling is Lipton's (1977) 'urban bias' thesis. He argues that poverty in poor countries persists because their governments favour cities through biased resource allocation and price policies. Accordingly, for Lipton (1977: 13; emphasis added) the 'most important class conflict in the poor countries . is not between labour and capital. Nor is it between foreign and national interests. It is between the rural classes and the urban classes'. In addition to stressing that the main conflict in poorer countries is geographical in nature (rather than class-based or between different capital fractions), Lipton also asserted that the specific spatial conditions of the city, namely density, decided who won the battle. Lipton (1977: 13, 62) claims that 'urban classes have been able to ''win'' most of the rounds of the struggle with the countryside' because city-dwellers develop, due to their physical and social closeness, common 'interests, preferences, friends, places of residence and above all perceptions'. For being more articulated than dispersed rural populations, citydwellers can act jointly, as 'threateners, promisers, lobbyists, dinner companions, flatterers, financiers and friends to senior administrators and politicians'.
Within the context of research on dependent urbanisation that has been advanced since the late 1960s by Marxist scholars (Castells, 1977;Harvey, 1973) and those close to world-system analysis (Abu-Lughod, 1977;Roberts, 1978;Timberlake, 1985), a strand has developed that focuses on internal colonialism. This notion goes back to Russian and Italian Marxists who used it to describe regional disparities and the exploitation of peasants by city dwellers, and it was put back on the agenda by the Mexican sociologist Gonzalez Casanova (1965; see also Walton, 1975). Departing from a conceptualisation of colonialism as economic and political relations that ensured the 'domination and exploitation of a total population (with its distinct classes, proprietors, workers) by another population which also has distinct classes (proprietors and workers)', Gonzalez Casanova (1965: 33; emphasis added) suggests that the historical colonial structure 'resembles relations of domination and exploitation typical of the rural-urban structure of traditional society and of underdeveloped countries'. In this internal colonialism, exploitation is organised not only locally, between the owners of the means of production and workers, but also inter-regionally, between the population of a country's 'dominant centre' or 'metropolis' and the rural (and, in the Mexican case, indigenous) population.
Two aspects are particularly noteworthy: Gonzalez Casanova (1965) argues that the ability of the elites to impose colonial relationships within the country comes from a monopolistic control over typical urban economic activities such as trade, credit or communication. And he notes that internal colonialism is organised through an intercity network, in which 'dominant centres' on various geographical scales (down to cities such as Zacapoaxtla, which in the 1960s had some 20,000 inhabitants) organise relations of 'interchange . plainly unfavorable to the Indian communities' (Gonzalez Casanova, 1965: 34).
The notion that uneven development works through inter-city networks has spread in critical urban research. Harvey (1988Harvey ( [1973: 232), for example, contends that in cases of dependent urbanism (such as in Latin America or the 19th-century USA), 'the urban form exists as a channel for the extraction of quantities of surplus from rural and resource hinterland for purposes of shipment to the major metropolitan centres'. For Castells (1977) the criterion for defining a metropolis was its functioning as a unit of control in urban networks that dominated other, smaller units. Well known to this day is also Frank's (1969: 6) wording that the development of underdevelopment is organised through 'a whole chain of constellations of metropoles and satellites (which) relates all parts of the whole system from its metropolitan centre in Europe or the United States to the farthest outpost in the Latin American countryside [.] we find that each of the satellites [.] serves as an instrument to suck capital or economic surplus out of its own satellites and to channel part of this surplus to the world metropolis of which all are satellites '. 11 In the late 1980s and in the 1990s, reflections on cities as nodes in asymmetrical, exploitative relationships across space disappeared from the research agenda. However, such thoughts have found their way into the emerging world city literature (Friedmann and Wolff, 1982;Sassen, 1991) -Friedmann's key role in both debates is striking! Yet, while early world city research was guided by economic geography ideas and the intention to develop a city-centred perspective on global uneven development, as world city research entered the mainstream it moved towards more urban geographic topics (such as the polarisation of labour or real estate markets in cities). Because economic and urban geographic perspectives unfortunately remained unconnected in world city research, it could never establish why and how specific characteristics of certain cities matter to 'how cities and the classes within them achieve control over other regions' (Roberts, 1986: 459). Accordingly, the 'command and control continuum of global city functions' (Robinson, 2002: 547; emphasis added), tying all kinds of cities and places unevenly to global cities, remained as unexplored as the urban foundations of these control relations. More recently, however, this concern has come back to the fore again (Kra¨tke, 2014;Parnreiter, 2019;van Meeteren and Bassens, 2016).

Fernand Braudel: Cities as aggressive worlds
Another important source of inspiration for my pondering about the Janus-faced nature of cities' genius is Braudel's (1983Braudel's ( , 1984Braudel's ( , 1985 account of the rise of capitalism. Based on his conviction that capitalism is always about growth and inequality, and that inequality is always social and spatial inequality, Braudel is not blinded by cities' virtuosity (which he nevertheless admires). Rather, he always addresses both sides. Cities are, says Braudel (1984: 312), 'indispensable to the general process of growth' and therefore 'turning-points, watersheds of human history ' (1985: 479). But they are also producers and profiteers of uneven development - Braudel (1985: 481; emphasis added) contends that cities by definition live on asymmetrical relationships: '(t)he town only exists as a town in relation to a form of life lower than its own'.
Yet, cities do not just live on asymmetrical relationships; they actively create them. To prosper themselves and to prompt economic growth, cities maintain exploitative relations to minor cities and towns, and to the rural hinterlands. They thus form 'semiparasitical systems', 12 which turn them into 'an aggressive world and an active force for unequal exchange' (Braudel, 1984: 312, 94;emphasis added). Importantly, this unequal exchange favoured more than just the few rich in the big urban centres. A 'middle class' also emerged in cities and to a certain extent even the urban poor benefited from cities' uneven relationships to the outside world. Braudel (1985: 484) therefore referred to the city-country relations as a 'version of the class struggle'. While Braudel did not mention the 'aristocracy of labour' debates of the 1960 and 1970s (e.g. Emmanuel and Bettelheim, 1970;Hobsbawm, 1967), the parallelism of his argument to this literature is striking.
For the purpose of this paper, it is of utmost importance that Braudel (1984) insists (albeit without ever formulating an urban theory of his own) that urban elites' capabilities to develop and deploy 'weapons of domination ' (1984: 35) is based on their command over urban resources. He recurrently refers to agglomeration economies, external networks and the built environment as prerequisites for growth, but also for exploitation.
As to agglomeration externalities, Braudel (1985: 481) portrays major cities as 'unusual concentration(s)' of all kinds of different people but also of money, knowledge, goods, services, etc. A consequent advantage held by capitalist agents in big cities -in his case, mainly merchants -was that they were earlier and better informed than their competitors; this was key, because the better informed could quickly switch to the economic sectors or regions where profits were highest. Moreover, the clustering of rich people made investment capital availablein times when a credit system over distance was only beginning, access to large sums of money to make advance payments was everything. Braudel (1983: 416; emphasis added) therefore equates having 'sufficient capital' with having access to 'local sources of credit', which were particularly rich in cities where merchants gathered, where wealthy refugees were welcomed, and where there were enough medium and even smaller incomes that could be tapped, too. In sum, cities were 'high-tension system(s)' that allowed merchants to be 'both sufficiently informed and materially able to choose the sphere of . action' (Braudel, 1983(Braudel, : 400, 1984. Inter-city networks are of eminent importance in Braudel's account of the rise of capitalism, too. In seeing long-distance trade as its cradle, Braudel (1984: 96) claims that '(t)here could be no world economy until there was a dense enough urban network with trade of sufficient volume and regularity to breathe life into a central or core zone'. This life came in the form of new knowledge, skills, goods and services, capital or simply labour, which flowed into interlinked cities. Financial innovations, for example, 'spread from city to city', brought by merchants, who, travelling around, could not, as in the case of the bill of exchange, 'have failed to notice this convenient method of transferring a sum of money to distant parts simply by a piece of paper' (Braudel, 1983: 556). Inter-city networks also helped to control extended trade routes. Were all the long-distance connections operated by headquarters in Europe, asks Braudel (1983: 157), '(o)r were there . subsidiary centres?'. A purely rhetorical question, because merchants needed 'to have associates at strategic points along the trade route, who were a party to one's secrets' (Braudel, 1983: 416). Dominant cities' merchants required bridgeheads (in the plural!) into the territories they exploited. Without 'satellites and the nervous system of urban relay points ' (1985: 150), surplus could not have been syphoned off from other regions. Thus, when Braudel (1984: 96) affirms that 'inter-city links (are) encompassing all underlying activities and obliging them to become part of a ''market economy''', he is actually arguing that inter-city networks have constituted the infrastructure for seizing all 'underlying activities': 'The western town swallowed everything, forced everything to submit to its laws, its demands and its controls', with the consequence that 'capital and surplus wealth is poured' into the most dynamic cities (Braudel, 1983(Braudel, : 27, 1985. Braudel (1984: 48) thus theorises inter-city networks as 'chain(s) of subordinations, each conditioning the others'.
There is, finally, the built environment. When Braudel (1984: 36) talks about 'the construction process of any world economy', this should not be understood figuratively. Countless are his references to marketplaces, exchanges, shops, butcheries and barracks; to streets, avenues and fleets; to ports, docks, pools, basins and quays; and, of course, to the kontors, stores, depots, granaries and warehouses that belonged to Europe's merchants. Physical construction processes thus underlie the building up of a world economy. One example of the strategic importance that Braudel attributes to the built environment must suffice here: merchants' practice of monopolising main assets -according to Braudel, the source of all big profits -depended not only on the mobilisation of political power but also on physical infrastructures. Seventeenth-century Dutch wholesalers, who for Braudel (1983: 418f; emphasis added) were 'masters of the art' of creating monopolies, used as '(p)ractical weapons' their 'great warehouses . The rule was always the same: buy goods directly from the producer for a low price, in return for cash or, better still, advance payments; then put them in store and wait for prices to rise'.

Do cities exploit other cities?
Both the assertion that cities have a certain genius and my claim that this genius is Janus-faced are based on a conceptualisation of space 'as an ''active moment''' in capitalism, as Harvey (2006Harvey ( [1982: 374) put it in the early 1980s (even though he was quite ambiguous on this question). Based on Lefebvre's work (1974), in the 1970s arguments spread in critical social sciences imbuing space with social effectiveness, such that space becomes 'a source of explanation itself' (Soja, 2000: 11;original emphasis). In debates on uneven development, followers of Lefebvre thus began to speak of an 'explicit spatial problematic' (Soja, 1980: 219) in capitalist exploitation. More orthodox Marxists contested this 'spatial turn', objecting that such ideas were fetishising space. A case in point is Harvey's (1988Harvey's ( [1973: 234) critique of Hoselitz, in which he stresses that it is not cities that are parasitic, rather the 'urbanised elite'.
Indeed, cities are no actors. It is never cities but elites within them who appropriate wealth produced by other people living and working in the same city or beyond. So, why suggest that the city is an analytical category for understanding exploitation and uneven development? For exactly the same reasons that we talk about the city when we discuss innovation, efficiency and growth. Where should all the knowledge and practical means to organise uneven development come from, if not from cities? When accumulation is driven by city-related characteristics (Harvey, 1985;Jacobs, 1985;Storper, 2013), why aren't its exploitative practices, too? I contend that the city is an explanans for both: the same urban conditions that bring forth the wealth of nations (Jacobs, 1985) are involved in the production of the 'poverty of nations'. My notion of the Janusfaced nature of the genius of cities is based on the claim that the city as a particular working and living space empowers elites to organise the socio-spatial relations along commodity chains such that wealth flows take on a centripetal form. Importantly, from the Early Modern Times to the present these uneven relationships have benefited not only a handful of urban elite merchants and entrepreneurs but also cities' 'total population', as Gonzalez Casanova (1965) has put it in asserting an 'internal colonialism' and as data on the geography of poverty confirm . Bringing together the reflections on the nature of cities and on uneven development offers an opportunity to conceptualise the city as the link that provides the 'connection between the vertical and horizontal [i.e. geographical -ed.] dimensions of the class struggle' of which Soja (1980: 219) speaks. What is needed, then, is to specify in more detail in which particular manners dominant classes 'depend on urban contexts', as Van Heur and Bassens (2019: 591) have recently held.
While the literature reviewed in the previous two sections gives some indicators of how cities' socio-physical environment provides resources that urban elites exploit to impose exploitative relations across space, Allen's (2005Allen's ( , 2010 account of the 'city-ness of power' is particularly useful for arguing that the genius of cities empowers elites for generative and exploitative practices without risking a fetishisation of space. Allen (2003: 2) insists that power is not 'stored' in cities but, as a 'relational effect of social interaction', constituted through 'interaction and exchange which take place through a complex of networks' (Allen, 2010(Allen, : 2897. Allen adds, however, that exercising power also requires the availability and mobilisation of spatially fixed resources. Power is, thus, 'the product of pooled resources which are used to fuse and modify patterns of interaction at-a-distance' (Allen, 2010(Allen, : 2900. These resources include, first of all, human assets: the right mix of people, their skills and the interactions they generate (all of which are easier to find in agglomerations). Second, Allen (2005: 193) refers to 'the kind of environment [i.e. of a city -ed.], the type of institutions, the flows of information, the symbolic assets and so forth', and, elsewhere (Allen, 2003: 194) to the built environment, namely 'monumental buildings, state ministries, fenced-off housing estates, or the headquarters of global multinationals'. Consequently, despite the fluid nature of power, materiality matters -'you cannot ''dissolve'' such resource bases into a networked morphology of flows' (Allen, 2003: 194). Dominant classes do not operate on the proverbial pinpoint of a needlewithout the availability of a city's resource base, would-be exploiters remain wanna-be exploiters; if they have nothing to mobilise, they remain powerless. In this sense, the (global) bourgeoisie is 'inherently urban', as Taylor contends (2013: 56; emphasis added).

Conclusion
The starting point of this article was my discontent with the one-sided enthusiasm of much of today's urban theorising. While cities are constantly being acclaimed for producing all the 'good' innovations and for being responsible for growth and development, all the 'bad' innovations, which bring about the instruments for exploitation and oppression, seem to come from nowhere.
Contrary to this view, I argue in this paper that cities' genius (cf. Storper, 2013) is Janus-faced in the sense that it works in opposite ways. Agglomeration economies, network externalities and massive built environments can be and, in fact, are mobilised for different purposes -from wealth production to the development of the means of exploitation. I thus add to Peck's (2017: 7271f) request that analysis of uneven development must consider 'interconnections between wealth creation and poverty generation' that urban studies must reflect on which of these interconnections are cityrelated or even city-produced.
The literature reviewed here gives some indicators. For Braudel, cities' specific sociophysical environment provides resources that allow their elites to impose exploitative relations on rural hinterlands, towns and minor cities. Such urban-theoretical references in assessments of uneven development can be found in the literatures from the time of modernisation and dependency theory, too, though in a more loosely scattered manner. Friedmann (1973) and Lipton (1977), for example, point out that physical proximity to one another and access to the tangible and intangible resources concentrated within cities are the foundations of power for those who hold it, while the notion that wealth transfers across space are organised through inter-city networks at different scales can be found in many accounts, from Hoselitz (1955) to Gonzalez Casanova (1965), Friedmann (1967 and Frank (1969). Gonzalez Casanova (1965) finally links the spatial concentration of certain economic activities (such as trade, credit or communication) to elites' ability to control them monopolistically, an idea that is also present in Braudel's work.
While these debates leave us with some suggestive insights into the dark sides of cities' genius, how agglomeration, inter-city networks and the built environment enable urban elites to develop and deploy instruments of exploitation has, to my knowledge, never been explored systematically. Given the space constraints, this paper could not give empirical examples, either. Yet, one of the obvious avenues for examining the Janus-faced character of cities' genius is a systematic analysis of finance's practices that led to the 2007-2008 crisis. A promising way to reveal in which specific ways actors in the financial industry depended on urban contexts (cf. Van Heur and Bassens, 2019) to develop the instruments underlying the subprime system would be to follow clues such as the one from Mould (2016: 159; emphasis added) who remarks that 'subprime mortgages are the very products that this creativity and innovation [i.e. of citiesed.] bring about'. If Mould's assertion is correct (and I think it is), then the instruments that made predatory lending possible are 'intrinsically urban in character', to quote Scott and Storper (2015: 9) again.
My argument has some implications for ongoing debates around specific concepts in urban studies. First, positing an 'urban spatial causality' (Soja, 2010) in the making of uneven development implies to reject the core thesis of the planetary urbanism literature, namely that the city has 'become obsolete as an analytical social science tool' (Brenner and Schmid, 2011: 13). Instead, I follow Scott and Storper (2015: 7) in this regard, who 'assert the status of the city as a concrete, localised, scalar articulation within the space economy as a whole . the specificity of the urban depends not so much on the crude ratio of its internal to external transactions, but on the contrasting qualities of these two sets of transactions and their locational effects'. But does not such a contention suffer from methodological cityism, because it puts 'an overwhelming analytical and empirical focus on the traditional city to the exclusion of other aspects of contemporary urbanisation processes' (Angelo and Wachsmuth, 2015: 16)? No, because my focus is not on 'urbanisation processes in their totality' (Angelo and Wachsmuth, 2015: 21; emphasis added), but rather on specific city-related processes, namely 'agglomeration as process' (Scott and Storper, 2015: 9), inter-city networking as an 'external urban process' (Taylor et al., 2010) and the production of the built environment as part of the 'material process' of grounding accumulation geographically (Harvey, 1985: 22). While the planetary urbanism proponents correctly suggest that 'urbanism as a way of life' has spread out and single features of cities have become available in an increasing number of places, the specific 'spatial intensities of social relations', which according to Massey (1999: 76; emphasis added) constitute 'the city proper', are not available ubiquitously. They emerge only spatially concentrated, where there is a socially and physically dense, diverse and complex environment. Cities are therefore more than just one scalar dimension of capitalism and its uneven development; they remain a central analytical category.
The second implication of my argument for theoretical debates on the nature of cities concerns the literature summarised under the label of Southern urban theory. Concerned with the geographies of knowledge production, it calls for 'parochialising Western knowledge' and for 'a more cosmopolitan engagement with experiences and scholarship elsewhere' (Robinson, 2006: 3). If one does not take this as a plea to privilege the particular over the general but as a request to expand, to correct and to reformulate our understanding of cities based on added knowledge from the South and the East, then this paper shows the usefulness of this concern. Whether from a modernisation-theoretical or a Marxist perspective (as in Hoselitz or Gonzalez Casanova) or in the form of a dense description, as in Braudel, studies of cities and inter-city networks in peripheralised regions in the Global South or in Eastern Europe did much to inform my argument.
Urban scholars have, thus, a lot to offer in economic geographic debates on uneven development across space. Through a critical urban theoretical perspective on surplus extraction along commodity chains, we could begin to develop a 'city-fied' understanding of how the unevenness of capitalism's division of labour is being produced and sustained. This, in turn, would help urban studies to propose a less biased, less developmentalist view on cities' role in capitalism. Cities are not only engines of growth and development, they are an explanans of the enduring poverty and polarisation, too.

Acknowledgements
For helpful and challenging comments on earlier versions of the manuscript thanks are due to the editor, three anonymous reviewers and to David Bassens, Christin Bernhold, Ben Derudder, Stefan Kra¨tke, Ju¨rgen Oßenbru¨gge, Kunibert Raffer, Peter Taylor and Michiel van Meeteren. Their comments have enabled me to make significant improvements in the arguments presented here. The largest single debt is owed to Peter Feldbauer, who sparked my interest in cities and world systems analysis when I was an undergraduate student. The usual disclaimers apply.

Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD
Christof Parnreiter https://orcid.org/0000-0003-3944-2538 Notes 1. Owing to space constraints, it is impossible here to review the various attempts made to define 'the city'. For me, the essence of cities across variations in time and space consists of a specific diversity and density of physical and social features, the bringing together of massive material infrastructure and a large variety of intense social relations in a relatively small space. 2. Again, because of space constraints, I limit myself to the economic dimension of cities' vitality, though cities are being praised for triggering social and political development, too (Hobsbawm, 2005;Sennett, 2005). 3. Certainly, cities do not exist in a vacuum.
City-state relations are critical for how cities develop and for how they operate in national and global contexts (Tilly and Blockmans, 1994). Moreover, one cannot analyse uneven development comprehensively without considering the state. Capitalism requires the networked economic power (cross-border spaces of flows and cities as their nodes) and the territorialised political power in the form of states (Arrighi, 1994). Yet, given the space constraints, I limit myself here to address the role of cities. 4. This line of reasoning goes much further back, in fact. Sun (2012) cites reflections of the Greek philosophers Xenophon and Plato on positive externalities of agglomerations. 5. There is, for example, no consensus on what exactly is being transferred geographically. While Hadjimichalis (1984) speaks, with reference to Marx's law of value, of value transfers, dependency and world-system scholars discuss the geographical transfer of surplus (Frank, 1969;Wallerstein, 1974a). Moreover, while proponents of the 'unequal exchange' theory equate this with 'exploitation by trade' (Raffer, 1987: 6), in a stricter Marxist analysis, the concept of exploitation is reserved for the production sphere. While both discussions are important for theorising the nature of uneven development, they are not of primary importance to my argument regarding the relationship of the genius of cities to the organisation of wealth transfers across space.
6. Even Wallerstein (1974b), who relates the prevalence of a vibrant inter-city network in Europe in the 16th century to core-forming dynamics, does not treat cities as causes but as indications, as 'outcome(s) of a process rather than a process itself' (Taylor, 2018). 7. Although I am referring to the regional/global dimension of uneven development, my argument that the means of appropriating the fruits of other people's work are of urban origin applies to the intra-urban scale of exploitation, too. I cannot elaborate on this argument here because of space constraints, except to say that the instruments of predatory lending would certainly be a case in point. 8. Timberlake (1987: 51). 9. Galtung (1971) also argues that local elites in poor countries serve as bridgeheads of imperialist interests, albeit without giving this idea any geographical dimension. 10. Smith (1976Smith ( [1776: 142f) already pointed out that, because of their physical proximity, urban dwellers have, unlike the rural population, the opportunity to collude -and that they take advantage of this opportunity. 11. Frank happened to be Hoselitz's research assistant in 1955. It is quite conceivable that he was introduced to the notion that cities' parasitism works through metropolissatellite relations on this occasion (I thank Michiel van Meeteren for pointing this out to me). Some years later, Frank was teaching at Mexico's UNAM, at the very same time that Gonzalez Casanova was Head of its School of Political and Social Sciences. 12. Braudel does not refer to Hoselitz's (1955) thesis in his three-volume work.