GaWC Research Bulletin 317

This Research Bulletin has been published in Global Networks, 12 (4), (2012), 446-466.

doi:10.1111/j.1471-0374.2011.00357.x

Please refer to the published version when quoting the paper

Introduction

Considered as the elementary units of collective commercial agency, in economic geography firms have been largely unproblematised as unitary and coherent actors (Yeung 2003; Maskell 2001; Taylor and Asheim 2001). As Grabher (2004a) suggests, economic geography research at the meso-level on networks has largely been focused at an inter-organizational level. This has particularly been the case in much of the research being undertaken on world city networks, which has to date largely focused on the role of advanced producer services firms and their trans-national office networks (see Sassen 2000 2001; Taylor 2004). It is typical of much world cities research that overlapping social networks, and the individual actors that constitute them, are uncritically subsumed into inter-firm networks. Ettlinger (2003) argues that this top-down strategy excludes the people involved in the daily practices of work, and leads to an ‘ecological fallacy' whereby it is presumed that what holds for firms in networks also holds for individual actors. For Grabher (2002a), the integrity of the firm as the basic analytical unit is being increasingly undercut by organizational practices that are built around projects involving a multiplicity of organizational and personal networks. For Yeung (2005a) there is a need for a relational conception of the firm as social networks in which actors are embedded in ongoing power relations and discursive processes, and as such mapping the power geometries of the firm entails significant analytical attention to the relational power of social actors and their territorial organization. These arguments have important implications for research into the formation of urban networks of world cities. While it is now widely accepted that cities do not have power in themselves, but rather find power in the urban networks in which they are embedded, recent research on relational geographies and on project working highlights that there are actors of network formation other than firms that need to be considered - agents that are crucial for cities to achieve ‘global reach'. For cities to project their power over distance the successful enrolment of these actors in networks is crucial (Smith 2007). From a relational perspective it is social actors, rather than the firm as an abstract entity, that become the key analytical focus (Yeung 2005a).

Social network analysis provides a set of tools that can be employed to empirically assess the centrality and power of cities, through the analysis of social actors and their networks. This paper employs social network analysis to undertake an examination of urban networks based upon the social connections occurring through project work, i.e. relational data linking cities with other cities rather than on attributional data for specific cities. The paper is concerned with the production of music, which provides a particularly revealing focus for research due to the ways in which music production is caught up in multiple layers of networks (Connell and Gibson 2003) involving a wide range of actors, particularly given the rise of new internet technologies enabling enhanced networking over geographical space. Specifically, the paper examines the working flows that occur between recording studios, based in cities across the globe, when they are part of temporary creative projects that are brought together to produce recorded music albums. The end result is a mapping of the power and centrality of cities within the relational urban networks of music production.

To begin, the paper considers the production of musical recordings as market-based temporary project work, with a specific focus on key skilled actors, their personal and professional networks, and the crucial role played by technology in facilitating networking over distance. Following on from this, the paper examines the ways in which power in relational networks has been conceptualized, with specific reference to world cities research, and how power can be measured through social network analysis methodologies. Following a brief discussion of data collection, the paper then describes the results of a social network analysis that attempts to define and map the urban networks formed through creative project working in the recorded music industry, assesses the level of connectedness of cities, and employs a number of measures to determine the power and centrality of cities within networks of production for digital music markets.

Project work in the recorded music industry

For Yeung the ‘firm' is a “messy constellation of multiple identities, contestation of power, and shifting representations” (2005a: 451), and as such monolithic ‘black box' conceptions of this crucial analytical category need to be revised. He argues that there is a need for a relational conception of the firm as social networks in which actors are embedded in ongoing power relations and discursive processes. A relational perspective on economic geography explicitly draws attention to the importance of economic actors and how they act and interact in space (Bathelt and Glückler 2003). Thus from a relational perspective it is social actors, rather than the firm as an abstract entity, that become the key analytical focus (Yeung 2005a). For Yeung (2003) actor networks, ranging from personal/social networks to inter-firm business networks, are an important focus for research in new economic geographies because all kinds of networks are constituted by these heterogeneous associations and relations among actors. Further, Dicken et al. (2001) argue that for human intentionality to take effect it must be mediated through such heterogeneous actor-networks, which are spatially and temporally constructed. Projects represent particular forms of temporal and spatial actor-networks. Projects can be defined as systems of production that are constituted by a variety of different economic, social and cultural agents often with specialized and complementary competencies collaborating over a pre-determined period in order to complete a pre-specified and usually complex task (Lundin and Söderholm 1995), where the complexity of the task necessitates the coordination of multidisciplinary skills that it is not economically efficient to bring together on a permanent basis (Lorenzen and Frederiksen 2005). Such temporary project systems are not a new phenomenon, having always been present in certain industries (Asheim 2002), although the development of innovative new technologies now allow even more flexible arrangements for project-based working (see Christopherson 2002).

Lorenzen and Frederisken (2005) suggest that, in the music industry, product innovation is organized into projects in order to facilitate experimentation and product variety in order deal with demand contingencies arising from ambiguous and changing consumer tastes. They demonstrate that projects are carried out mainly in the market, rather than inside the boundaries of firms, in order for projects to include new and shifting resources and skills, skill holder motivations and deal with tasks that render internal governance and planning inefficient. Whereas much of the literature on projects has focused upon ‘project teams', in which skilled actors are employed within the same firm, in ‘market-based' projects participating skill holders are employed in different firms or may be freelancers, transcending the boundaries of firms (Lorenzen and Frederickson 2005). As DeFillippi and Arthur (1998) assert, fluid project working challenges the idea of core competencies existing as internal resources, and the knowledge base required to produce a recorded musical product is largely external to the record company, and often is not internal to the industry (see for example Power and Jansson 2005). In order to produce successful products on increasingly global markets, record companies must be able to draw on relevant knowledge bases for the relevant part of the value chain in production (Asheim 2002), and draw essential competencies into the firm as individual projects require, including freelance labour. As the recording industry cannot ultimately control what is going to be commercially successful, larger firms often have attempted to monopolize access to the best recording facilities and most talented engineers and producers (Negus 1992). Musical recording in the late 1960s was recentralized in cities and strongly reconnected to the music industry as the new technology demanded considerable investments in studios and skilled personnel that only major record companies could afford (Watson et al. 2009). As Gibson (2005) notes, the best technology was rare, and usually found in cities with thriving music scenes, at studios such as the EMI Abbey Road studios in London and Capitol Studios in Los Angeles. Such studios typically employed permanent staff. However, since the 1970s, new sound recording technologies have broken monopolies (Jones 2002), brought about democratization of the recording process, and undermined the position of many recording studios (see Leyshon 2009). Patterns of project-work in music production have become more flexible, and we have seen the comparatively recent developed of a freelance project-based model for recording. Studios now largely act as an independent service within the contemporary recorded music industry, with many owned and operated by entrepreneurial producers and engineers (Watson et al. 2009). This has had important repercussions for project work in the music industry by increasing the number of studios and number of skilled studio workers available to firms and musicians.

Musical recordings are essentially ‘one-off' projects that bring together, temporarily in space and time, a group of skilled creative workers to undertake a project with a definite end product. However, as personal networks are built, further projects may be undertaken involving recurrent collaboration, with new projects tending to draw on core members of successful prior projects. As projects are repeated over time, project ecologies may emerge, involving a range of different firms and organizations, individual actors, technologies, spaces and places (Grabher 2002a 2002b). These ecologies will form the backdrop to every subsequent project initiated, as new projects find their participants in the ecology (Lorenzen and Frederiksen 2005). Grabher (2002b 2002c) emphasizes that such chains of repeated co-operation are held together, or indeed cut-off, by the reputation members gained, or lost, in previous collaborations. Activities in temporary projects are dominated by individual knowledge embodied in highly mobile project members. As these embodied creative knowledges are for sale on the labour market, any competitor can potentially draw on competencies that have developed (see Lam 2000). Individual skills are transferred between projects as project members typically collaborate simultaneously with a wide range of firms. Networking is then the emblematic practice in projects (Wittel 2001). However, as well as professional networks and communities of practice revolving around firms, projects also involve personal networks that “symptomatically efface the distinction between private and business” (Grabher 2004a: 105). In project-based working it is often personal networks, rather than formal firm contractual networks, that provide the basic social infrastructure for putting together a project team. Previous research on the music industry has highlighted the importance of geographical proximity and face-to-face interaction in the development of personal and social networks and relationships in the music industry, the dynamics of which are built around an informality that blurs the business–social divide (Watson 2008) and transgresses the boundaries of the firm. As such, spatial agglomerations function as potent frameworks of cultural reproduction (Scott 1999). It is important, however, to note that these milieux are not geographically constrained. For Asheim (2002) the continued importance of localized learning can be challenged by the increasing importance of temporary project working. Personal and professional creative networks are increasingly spanning the globe, resulting in geographically far-flung project collaborations (see for example Cole 2008). In the case of the music industry, new technologies that network studios in geographically-distant locations enable musical recording projects to be co-ordinated on a global scale and so allow for projects to draw on pools of creative labour in geographically-dispersed locations. Musical recording projects therefore often span across and connect multiple cluster-based ecologies, and as such this paper provides a geographically wider frame of analysis than those studies focusing on one single ecology in one particular place (for example Grabher 2002a on the advertising industry in London).

Conceptualizing and measuring power in urban networks: social network analysis methodology

This analysis employs two different measures to assess the power of cities in urban networks. The first measure used is Bonacich's power-based centrality measure (see Hanneman and Riddle 2005). In applying this measure to urban networks, centrality and power in the network is a function of the connections of the cities to which a particular city is connected. The more connected the cities to which a particular city is connected to, the more central the city is. The less connected the cities to which a particular city is connected to, the more powerful the city is, and the less connected cities will be more dependent on it. The second measure used is flow betweeness. This measure is based on the proportion of the entire flow between two actors, through all of the pathways connecting them, which occurs on paths of which a given actor is a part. The measure adds up how involved the actor is in all of the flows between all other pairs of actors, as a ratio of the total flow betweeness that does not involve the actor (Hanneman and Riddle 2005). Betweeness centrality is an important indicator of control of information exchange and resource flows within a network (Knoke and Yang 2008), as the measure ascertains the extent to which an agent can play the part of a ‘gatekeeper' with a potential for control over others (Scott 1991). Although they may not necessarily have the most connections to other cities, those cities with a high degree of flow betweeness centrality are considered to be the most important mediators in the urban network. These cities are better situated than other cities as a result of the position that they occupy in the network (Alderson and Beckfield 2004) due to their own and their neighbour's network connections. A core-periphery analysis is also undertaken on the valued data matrices to identify those cities belonging to the core of the network and those which belong to the periphery. The social network analysis presented in this paper was undertaken using the UCI NET software (Borgatti et al. 2002). The network visualizations provided are derived through the embedded NetDraw visualization tool.

Data collection

The projects on which the analysis focuses are recorded popular music albums, defined as a group of audio tracks with a generally consistent track list across the different territories in which it is released. Each album is its own temporary project, consisting not only of firms (record companies), but also localities - recording studios in particular cities, and the professional and personal networks of the musicians and studio producers and engineers – ‘creative labour'. Within these projects, elements of creative labour may be fixed in particular studios, with recordings being transferred digitally, or this labour may be mobile between studios in different cities. It is these movements, of both labour and recordings, which are the connections that form urban networks of musical production within the recorded music industry. Thus, in collecting data for the social network analysis described in the following section of the paper, an event-based strategy has been employed in which network boundaries are drawn by including actors who participate in a defined set of activities occurring in specific times and places (see Knoke and Yang 2008). Each of these events, in this case temporary music industry projects (albums), has their own distinct production network, varyingly dispersed in terms of their geography. An example of a geographically dispersed network is shown in Figure 1, for the album ‘Tonight' by Franz Ferdinand, released on Domino Records/Epic Records in January 2009. The network of recording for this particular album is dispersed across six studios in six cities, including cities in the UK (London, Bristol, Glasgow), the US (Los Angeles, Phoenix) and Canada (Vancouver). By including multiple events (albums) in the network analysis, it is possible to produce a comprehensive and inclusive network, in which many distinct networks overlap with one another.

Figure 1: Example album project network: Franz Ferdinand ‘Tonight' (Domino Records/Epic Records, 2009).

City codes: BR-Bristol; GL-Glasgow; LA- Los Angeles; LN-London; PH-Phoenix; VN-Vancouver.

Databases of recording information for albums, consisting of information on the recording studios used, and the creative labour involved in the recording, were constructed based upon the details given in the credits of albums appearing in the top 10 iTunes download charts, for the UK and US digital music markets, during the first six months of 2009. Not withstanding the ‘crisis' in the music industry that has resulted from the introduction of digital software formats (see Leyshon 2001 2003 2009; Leyshon et al. 2005; also Hughes and Lang 2003), the digital music market is forming an increasingly important part of the global music market. In January 2009 digital platforms accounted for around 20 percent of global recorded music sales, with the digital revenues of international music companies growing by an estimated 25 percent in 2008 to $US3.7 billion (IFPI 2009). iTunes sales charts were chosen for analysis because iTunes is the leading player in the online downloads market, and in 2008 became the largest music retailer in the US. iTunes top-10 music sales charts are published online and are continuously and automatically updated, and are available for most of the major national digital music markets. This allows comparisons to be made between a range of national digital music markets. In this paper, comparisons are made between the UK, US and Australian digital music markets. These three Anglophone markets were selected for analysis primarily due to the availability of the required data in English. An exploratory data collection exercise for a number of non-Anglophone markets including Japan revealed significant difficulties in obtaining the required data such that a full and comprehensive analysis would not have been possible for these markets within the limits of the research.

For reasons of practicality the continuous updates to the charts could not be followed on a constant basis, and therefore the charts were analysed on a weekly basis. Data was sampled between 1^st January 2009 and 31^st June 2009. Only full albums released in this time period and up to one year before, and including newly released material, were included in the sample. EPs (releases containing a smaller number of tracks than a full album), compilations, ‘greatest hits' compilations, and albums originally released over one year before the sampling date, were not included. The final databases contain data on 53 albums from the UK download charts, 52 albums from the US download charts, and 39 albums from the Australian download charts respectively. The data are coded as non-directional, i.e. there is no distinction made between ‘senders' and ‘receivers' in relationships, rather they are considered to involve mutual exchange. The data produce three symmetrical and valued matrices, one for UK networks of production, one for US networks of production, and one for Australian networks of production with the matrices linking 36, 43 and 29 cities across the globe respectively. Inevitably a significant amount of overlap occurs between the three databases.

Urban networks of musical production

Table 1 ranks the top five cities based on the release of albums into the UK digital music market. The figures given are based on the number of albums for which studios in the city were involved in the recording ‘project' expressed as a percentage of the total number of albums captured from chart data. Based upon this, London is shown to be the pre-eminent centre for the output of sales-successful recorded music into the UK digital music market. Studios based in the city were involved in the recording projects for over 50 percent of all the albums captured in the data. Los Angeles and New York, with 38 percent and 36 percent respectively, trail behind London but are far ahead of a second tier of smaller UK, European and US cities. Many other cities with individually smaller levels of output make up a third-tier of production. The dominance of the global city triad of London, New York and Los Angeles in terms of sales-successful output for the UK digital music market is clearly highlighted by these figures. Table2 ranks the top five cities based on output of albums into the US digital music market. In the case of the US digital music market, Los Angeles is shown to be the pre-eminent centre for the output of sales-successful music, with its studios involved in the recording of almost 60 percent of all the albums captured in the data. It is closely followed by New York, with New York studios involved in 46 percent of the albums sampled. Contrasting with the case of the UK digital music market, London is significantly behind both Los Angeles and New York in terms of sales-successful output into the US digital music market, accounting for 25 percent of the albums sampled. These cities are followed in the top 5 cities by two more US cities, Nashville and Portland (MN), accounting for 10 percent and 8 percent respectively. Table 3 provides the same rankings for the Australian digital music market. In this case, New York and London are pre-eminent, with studios in both cities involved in over 40 percent of all the albums captured in the data respectively. They are closely followed by Los Angeles, whose studios are involved in 38 percent of the albums sampled. These are followed by the Australian city of Melbourne. Accounting for 13 percent all the albums captured in the data, it is well behind the global city triad, but ahead of another Australian city, Sydney, which accounts for 8 percent all albums. This data for the US and Australian markets thus also highlights the dominance of the global city triad of London, New York and Los Angeles.

Table 3: Top 5 cities ranked by output of albums; Australian digital music market.

Connectivity in Urban Networks

While this output data is useful in providing a hierarchy of cities based on levels of production, it tells us nothing about networks of production between cities. The data gathered on connectivity, based on the links between cities occurring as part of creative projects, is more informative as to the configuration of urban networks of musical production. The data for connectivity for networks of production for the UK digital music market further highlights the dominance of the triad of London, New York and Los Angeles. Table 4 ranks the top cities based on their total number of connections to other cities. London, New York and Los Angeles dominate the rankings as the three most connected cities, with around three times the number of connections of the fourth placed city, Bristol. All three cities have their highest connectivity to each other, and all of the other cities have their highest connectivity with one or more of these three cities. The strongest link between individual cities is shown to be that between New York and Los Angeles, very closely followed by the connection between London and Los Angeles. The remainder of the list consists of other smaller UK, US and European cities. Figure 2 provides a visual representation of the urban networks formed by these connections. The visualization displays the triad of London, New York and Los Angeles lying at the centre of network, surrounded by a web of less connected cities whose role as music recording centres is articulated through the three highly connected global cities.

Table 4: Top cities ranked by total number of connections, UK digital music market.

Note: Tie strength is based on number of inter-city links; the size of the nodes is based on the total connectivity of the city.

Table 5 ranks the top cities within the urban networks of production for the US digital music market, based on their total number of connections to other cities. The US global city dyad of New York and Los Angeles are shown to dominate the rankings of the most connected cities. Both cities have over twice the number of connections of the third placed city, London. The two cities are shown to have an extremely strong level of connection to each other when compared to the strength of their links with other cities, having around four times more connections with each other than they have with London. Figure 3 provides a visual representation of the urban networks formed by these connections. The visualization displays the dyad of New York and Los Angeles lying at the centre of network of production. Contrasting with the network for the UK digital market shown in Figure, London does not match these two cities in terms of importance at the centre of the network. Table 6 ranks the top cities within the urban networks of production for the Australian digital music market. Mirroring the case for the UK market, the triad of London, New York and Los Angeles are shown to dominate the rankings, with New York marginally ahead of the other two cities. The highest ranked Australian city, Melbourne, has only a fraction of the number of connections of the triad. Figure 4 provides a visual representation of the urban networks formed by these connections. The network diagram displays a very similar configuration to that for the UK digital music market (Figure 2), with the triad of London, New York and Los Angeles lying at the centre of network.

Note: Tie strength is based on number of inter-city links; the size of the nodes is based on the total connectivity of the city.

Table 5: Top cities ranked by total number of connections; US digital music market.

Table 6: Top cities ranked by total number of connections; Australian digital music market.

Centrality and Power in Networks of Production

In the urban network of production for the UK digital music market, Los Angeles, whilst only the third most connected of the cities in terms of total connections, is calculated to have the highest degree of centrality, i.e. has the most connections to other cities with a high degree of connectivity, marginally above both New York and London. Although London accounts for the output of many more albums into the UK digital music market than Los Angeles and New York (52 percent of albums, compared to 38 percent and 36 percent respectively, by this measure it is the least central of the dominant three cities. However, in terms of power in the urban network, i.e. in terms of many cities with low degrees of connectivity being dependant upon the city, London is calculated to be the most powerful city in the network, very closely followed by New York. Los Angeles is the third most powerful city, but is shown to be far less powerful than both London and New York. London is also calculated to be the most important mediating city in the network based upon the flow betweeness centrality measure, significantly more important than New York, which is turn is a significantly more important mediator than Los Angeles. These results, outlined above and summarized in Table 7, are indicative of London 's dominance as the most important city within the urban network of production for the UK digital music market.

In the urban network of production for the US digital music market, New York is shown to score highest on all three measures (Table 7). This is despite having a weaker album output than Los Angeles (involvement in 46 percent of total albums compared to 58 percent), and only a marginally higher number of connections (54 compared to the 53 of Los Angeles). Based on the Bonacich measure, New York is calculated to have the highest degree of centrality, i.e. has the most connections to other cities with a high degree of connectivity, although it shown to be only marginally ahead of Los Angeles. Both cities have much higher centrality rankings than London, which in turn is significantly ahead of the fourth-placed city, Atlanta. New York is also shown to be the city with the most power in the urban network, i.e. in terms of many cities with low degrees of connectivity being dependant upon the city. By this measure, New York is shown to be much more powerful than Los Angeles. Los Angeles is shown to be only marginally ahead of London in terms of power in the network, despite accounting for a much higher output of albums (involvement in 58 percent of total albums compared to 25 percent) and having many more connections (53 compared to the 23 of London). This highlights London 's power over certain weaker cities in the urban network, cities which New York and Los Angeles may have to go through London to access. New York is also calculated to be the most important mediating city in the network based upon the flow betweeness centrality measure, significantly more important than Los Angeles, which is turn is a significantly more important mediator than London. These results are indicative of New York 's dominance within the urban networks of production for the US digital music market.

New York also scores highest on all three centrality measures for the Australian digital music market (Table 7) and therefore is calculated to have the highest degree of centrality, to be the most powerful city, and the most important mediating city in the network in the urban network of digital music production for the Australian market. It is however only very marginally ahead of London in terms of its centrality. Los Angeles comes ahead of London on the Bonacich power measure and flow betweeness centrality measure, despite having a marginally weaker album output, although is below London on the Bonacich centrality measure. One interesting outcome is the score of the Australian city Melbourne on the flow betweeness centrality measure. Melbourne is positioned clearly in fourth, behind the global city triad but well ahead of other cities in the network. This demonstrates that Melbourne occupies an important position within the network as a mediator city, playing the part of a ‘gatekeeper' for access to the Australian music market.

A core-periphery analysis for the networks of production for the UK digital music market gives a core that contains nine of the 36 cities involved in the production of the musical outputs included in this analysis. Along with the three dominant cities of London, New York and Los Angeles, is a second-tier of core cities: Atlanta, Bristol, Dublin, Glasgow, Miami, and Stockholm. These cities have relatively strong ties to the three dominant cities, and to each other, when compared to peripheral cities. The same analysis for the networks of production for the US digital music market gives a core that contains just five of the 43 cities included in the data. New York, Los Angeles and London are present in the core; they are joined by Atlanta and Portland (MN), the only second-tier core cities. All other cities in the network have relatively low connections with the core cities and each other. A core-periphery analysis for the networks of production for the Australian digital music market gives no distinct core or periphery.

Prestigious Studios in Prestigious Cities

As stated previously, the data used in the study is non-directional, in that it does not distinguish between connections to and from a city. Indeed, it is assumed that links between cities involve mutual exchange and communication in both directions. However, although it is not directly measured in the data, there is one particular part of the musical recording process where cities may perhaps be considered ‘senders' and ‘receivers': the mastering of recordings. Here recordings are sent via electronic means, to be mastered in specific studios, which undertake mastering for an unbalanced share of the recordings produced. Thus this key production process plays an important role in concentrating production networks through certain key cities.

In terms of the UK digital music market, the most significant mastering studio is Metropolis Studios based in London, followed by Sterling Sound based in New York (see Table 8). Together, these two mastering studios account for one-third of the total number of albums sampled. In the top five these studios are joined by Bernie Grudman Mastering (Los Angeles), Masterdisk (New York), and Gateway Mastering (Portland, MN). Together these five studios account for 55 percent of the total number of albums sampled. This highlights the concentration of this key process in particular studios in particular cities. In terms of the US digital music market, it is a US-based studio that is prominent. Sterling Sound, based in New York, dominates the list of key mastering studios (Table 9), accounting for 27 percent of albums. It is followed by Bernie Grudman Mastering (Los Angeles) Gateway Mastering (Portland, MN), Marcussen Mastering (Los Angeles) and Metropolis Studios (London). Together these five studios account for 66 percent of the total number of albums sampled, suggesting even greater concentration of the mastering process than that found in the networks of production for UK digital markets. Sterling Sound in New York also dominates the list of key mastering studios for the Australian digital music market (Table 10), accounting for 28 percent of albums. It is followed by Metropolis Studios (London), Bernie Grudman Mastering (Los Angeles), Gateway Mastering (Portland, MN), and The Exchange (London). Together these five studios account for 62 percent of the total number of albums sampled.

Table 8: Top 5 mastering studios in networks of musical production, UK digital music market.

Table 9: Top 5 mastering studios in networks of musical production, US digital music market.

Table 10 : Top 5 mastering studios in networks of musical production, Australian digital music market.

We might consider these select cities, to which a disproportionate amount of recordings are ‘sent' as prestigious cities, because they receive many directed connections. As Alderson and Beckfield (2004) describe, these are the cities that are sought out by other cities, have ties directed to them, and are chosen over others. It is perhaps unsurprising that the three most central and powerful mediating cities as indicated by the centrality measures - London, New York, and Los Angeles - are also the three most prestigious cities based on these connections. There are two central reasons for the concentration of the process in these cities. Firstly, technology is central to the mastering process, and therefore those studios that can afford to invest the latest technology will be most desired by potential clients. However, having the most desired technology is not enough alone. As described previously, the process requires studio engineers with the appropriate level of skill and creativity to employ the technology to best effect. All of the major mastering studios have mastering engineers contracted to them. Clients not only seek to use particular studios, but also to use particular mastering engineers based upon their reputation. For example, Ted Jensen, chief mastering engineer at Sterling Sound in New York, alone accounts for 15 percent of the total number of albums sampled from the US digital market, while mastering engineers John Davis and Tim Young of Metropolis Studios in London, together account for the mastering of almost 20 percent of the total number of albums sampled from the UK digital market. Bob Ludwig of Gateway Mastering in Portland alone accounts for 10 percent of the total number of albums sampled from the US digital market, and 7 percent of those from the UK digital market. The prestigious nature of certain studios, and thus of particular cities, can then be directly attributed to the skilled engineers that are working in the studios and living in the cities. Recent work in economic geography, lead by the work of Richard Florida (2002, 2005), has emphasized how large global cities such as London, New York, and Los Angeles act as magnets for these talented individuals from across the globe, in which many both work and live.

Conclusions

In market-based projects in the music industry, ties between record companies, musicians, and specialized producers and engineers reach out between musically creative cities across the globe, resulting in the development of new relational geographies of creativity. Through social network analysis this paper has provided an exploration and mapping of a sample of urban networks of production within the global recorded music industry. It has emerged from the social network analysis presented in this paper that the spatial agglomerations of music industry firms, studios, and creative labour in particular key cities remain central to music recording process in the age of digital music markets, with outstanding technical studio facilities strongly centralized in particular key cities. This is especially the case for the triad of global cities of New York, Los Angeles, and London, home to very strong concentrations of record companies and recording studios (see Scott 1999; Watson 2008). The path dependence of networks of recording is then intimately embedded in physical infrastructures, material outcomes of economic processes that are localized in certain places and territories and exist over long periods of time (see Bathelt and Glückler 2003).

ACKNOWLEDGEMENTS

I am extremely grateful to both Michael Hoyler and Phil Hubbard of Loughborough University for providing constructive comments on earlier drafts of this paper, and for their continued support for my research. I would like to acknowledge the financial support of the Department of Geography, Loughborough University, UK, towards the doctorate research which informs this paper. My thanks go to three anonymous referees for providing detailed and constructive comments on the paper. An earlier version of this paper was presented at the ‘Global Cities Now? Current Perspectives in Global Urban Studies' conference held by the Urban Geography Research Group of the RGS-IBG at the Centre for Urban Theory, Swansea University, 5^th – 6^th November 2009. I am grateful to all those who provided constructive feedback on the presentation.

REFERENCES

Alderson, A. S., and Beckfield, J. (2004) ‘Power and position in the World City System', American Journal of Sociology, 109 (4), 811-851.

Allen, J. (1997) ‘Economies of power and space'. In: Lee, R and Wills, J (Eds.), Geographies of Economies, Arnold, London, 59-70.

Asheim, B. T. (2002) ‘Temporary organizations and spatial embeddedness of learning and knowledge creation', Geografiska Annaler, 84 B (2), 111-124.

Bathelt, H. and Glückler, J. (2003) ‘Toward a relational economic geography', Journal of Economic Geography, 3, 117-144.

Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002) UCI NET for Windows: Software for Social Network Analysis, Harvard: Analytic Technologies.

Christopherson, S. (2002) ‘Project work in context: regulatory change and the new geography of media', Environment and Planning A, 34, 2003-2015.

Cole, A. (2008) ‘Distant neighbours: the new geography of animated film production in Europe ', Regional Studies, 42 (6), 891-904.

Connell J., and Gibson, C. (2003) Sound tracks: popular music, identity, and place, London: Routledge.

DeFillippi, R. J. and Arthur, M. B. (1998) ‘Paradox in project-based enterprise: the case of film making', California Management Review, 40 (2), 125-139.

Dicken, P., Kelly, P.F., Olds, K. and Yeung, H.W. (2001) ‘Chains and networks, territories and scales: towards a relational framework for analysing the global economy', Global Networks, 1 (2), 89-112.

Ettlinger, N. (2003) ‘Cultural economic geography and a relational and microspace approach to trusts, rationalities, networks and change in collaborative workplaces', Journal of Economic Geography, 3 (2), 145-171.

Florida, R (2002) The Rise of the Creative Class – and How It's Transforming Work, Leisure, Community and Everyday Life, New York, Basic.

Friedmann, J (1978) ‘The spatial organization of power in the development of urban systems'. In: Bourne, L S and Simmons, J W (Eds.) Systems of Cities, Oxford University Press, New York, 328-40.

Gibson, C. (2005) ‘Recording studios: relational spaces of creativity and the city', Built Environment, 31 (3), 192–207.

Grabher, G. (2002a) ‘The project ecology of advertising: task, talents, and teams', Regional Studies, 36, 245–262.

Grabher, G. (2002b) ‘Guest editorial', Environment and Planning A, 34, 1911-1926.

Grabher, G. (2002c) ‘Cool projects, boring institutions: temporary collaboration in social context', Regional Studies, 36, 205-214.

Grabher, G. (2004a) ‘Learning in projects, remembering in networks? Communality, sociality, and connectivity in project ecologies', European Urban and Regional Studies, 11 (2), 103-123.

Grabher, G. (2004b) ‘Temporary architectures of learning: knowledge governance in project ecologies', Organization Studies, 25, 1491-1514.

Hughes, J. and Lang K. R. (2003) ‘If I had a song: the culture of digital community networks and its impact on the music industry'. International Journal on Media Management, 5 (3), 180-189.

IFPI (2009) Digital Music Report 2009: New Business Models for a Changing Environment, International Federation of the Phonographic Industry, January 2009,

Jones, S. (2002) ‘Music that moves: popular music, distribution and network technologies', Cultural Studies, 16 (2), 213.232.

Lam, A. (2000) ‘Tacit knowledge, organizational learning and societal institutions: an integrated framework', Organization Studies, 21 (3), 487-513.

Leyshon, A. (2001) ‘Time-space (and digital) compression: software formats, musical networks, and the reorganization of the music industry', Environment and Planning A, 33 (1), 49–77.

Leyshon, A. (2003) ‘Scary monsters? Software formats, peer-to-peer networks, and the spectre of the gift', Environment and Planning D, 21 (5), 533–558.

Leyshon, A. (2009) ‘The software slump?: digital music, the democratization of technology, and the decline of the recording studio sector within the musical economy', Environment and Planning A, 41, 1309-1331.

Leyshon, A., Webb, P., French, S., Thrift, N., and Crewe, L. (2005) ‘On the reproduction of the musical economy after the Internet', Media, Culture & Society, 27, 177–209.

Lorenzen, M. and Frederiksen, L. (2005) ‘The management of projects and product experimentation: examples from the music industry', European Management Review, 2 (3), 198-211.

Lundin, R. A. and Söderholm, A. (1995) ‘A theory of the temporary organization', Scandinavian Journal of Management, 11, 437-455.

Maskell, P. (2001) ‘The firm in economic geography', Economic Geography, 77 (4), 329-344.

Negus, K. (1992) Producing Pop: Culture and Conflict in the Popular Music Industry, London: Edward Arnold.

Negus, K. (1993) ‘Global harmonies and local discords: transnational policies and practices in the European recording industry', European Journal of Communication, 8 (3) 295-361.

Power, D. and Jansson, J. (2004) ‘The emergence of a post-industrial music economy? Music and ICT synergies in Stockholm, Sweden ', Geoforum, 35, 425-439.

Sassen, S. (2000) Cities in a World Economy, 3^rd Edition, Thousand Oaks, CA: Pine Forge.

Scott, A. (1999) ‘The US recorded music industry: on the relations between organization, location and creativity in the cultural economy', Environment and Planning A, 31, 1965-1984.

Short, J. R., Kim, Y., Kuss, M. and Wells, H. (1996) ‘The dirty little secret of world city research', International Journal of Urban and Regional Research, 20, 697-717.

Smith, R. G. (2007) ‘Postructuralism, power and the global city', in P. J. Taylor, B. Derudder, P. Saey and F. Witlox (eds) Cities in Globalization: Practices, Policies and Theories, London: Routledge, 258-270.

Taylor, M. and Asheim, B. (2001) ‘The concept of the firm in economic geography', Economic Geography, 77 (4), 315-328.

Taylor, P. J. (2004) World City Network: A Global Urban Analysis, London: Routledge.

Taylor, P. J., Walker, D.R.F., Catalano, G. and Hoyler, M. (2002) ‘Diversity and power in the world city network', Cities, 19 (4), 231-241.

Warner, T. (2003) Pop music – technology and creativity: Trevor Horn and the digital revolution, Aldershot, UK: Ashgate.

Watson, A. (2008) ‘Global music city: knowledge and geographical proximity in London 's recorded music industry', Area, 40 (1), 12-23.

Watson, A., Hoyler, M. and Mager, C. (2009) ‘Spaces and networks of musical creativity in the city', Geography Compass, 3 (2), 856-878.

Wittel, A. (2001) ‘Toward a network sociality', Theory, Culture & Society, 18 (6), 51-76.

Yeung, H. W. (2003) ‘Practicing new economic geographies: a methodological examination', Annals of the Association of American Geographers, 93 (2), 442-462.

Yeung, H. W. (2005a) ‘The firm as social networks: an organizational perspective', Growth and Change, 36 (3), 307-328.

Yeung, H. W. (2005b) ‘Rethinking relational economic geography', Transactions of the Institute of British Geographer, 30, 37-51.

NOTE

^*Allan Watson, Department of Geography, Staffordshire University, Stoke-on-Trent, Staffordshire ST4 2DE, UK, a.watson@staffs.ac.uk