2. Brand capital as a complex adaptive system

This article develops a systems-theoretic approach to brand capital. In this section, we explain why brand capital is a complex adaptive system, as a prelude to examining how iconic brands can crystallize into institutions that orient and coordinate human action. In earlier work, we extensively developed a complex systems approach to capital that draws upon Lachmann's (Reference Lachmann1956) micro-level process analysis of capital and its structure (Endres and Harper, Reference Endres and Harper2012, Reference Endres and Harper2013; Harper and Endres, Reference Harper and Endres2010, Reference Harper and Endres2012; Harper, Reference Harper2014). Owing to space constraints, we limit our discussion here to adumbrating those aspects of the approach that are most germane to brand phenomena.

Brands are capital. For our purposes, capital is a very comprehensive and inclusive entity. Following Fisher (Reference Fisher1906), we define capital broadly as any combination of resources used to produce a flow of services over time. Our definition of capital provides ample scope for including all types of resources – both material and intangible – in the formation of brand capital. It is not the nature of a brand in itself but how it is used that determines whether it is capital. In short, a brand qualifies as capital if it is used in a production plan – if it is employed for productive ends in the eyes of the brand user. Like other forms of capital, brand capital exists for the sake of, and derives its value from, the service streams to which it gives rise. For example, entrepreneurs use brands as market-making devices to create value and to derive profit, whereas consumers use brands as capital goods to derive psychic income (e.g. enhanced self-image and stronger group affiliation) and lifestyle benefits (Earl, Reference Earl1986). It is these services streams, and not the individual capital goods (individual brands) per se, which are the ultimate objects of people's desires. A brand renders a service when its use brings about a change desired by the actor (e.g. attracting new customers) or when its use prevents an undesirable change (e.g. prevents the loss of regular customers).

To characterize brands as capital has several important implications. In the first place, a capital-theoretic approach highlights the structural aspects of brand networks. Brand capital is not an amorphous, aggregated stock. Rather brand capital displays functional, spatial and temporal organization that depends upon, but is not reducible to, individual plans and purposes. The approach emphasizes the building of structures and the architecture of connections – the intended and realized complementarities – between individual brands and among brand-capital combinations. Combinations of brands are a crucial building block in the connective structure of brand capital and production processes. Like any capital good, a brand only becomes productive once it is connected to, and combined with, other capital goods (including other brands) in a production plan (Lachmann, Reference Lachmann1956). Brands interact with one another, and the way entrepreneurs and other participants combine brands can have real effects on the services yielded by, and the profitability of, brand combinations. At any moment in time the set of connections is incomplete, so that gaps always exist in the brand-capital structure. These gaps represent potential opportunities for building new brands and extending existing ones. Creating a new brand naturally incorporates combinatorial acts of resource integration by brand entrepreneurs and co-creative brand users – what we call ‘recombinant capital formation’.

Furthermore, treating brands as a capital structure promises to enhance our appreciation that brand-capital structures are also complex adaptive systems. As such, brand structures possess several typical properties that are characteristic of complex adaptive systems in general:

1. 1. Complexity: Brand structures are made up of many heterogeneous elements with connections among them. For instance, Google's brand portfolio includes many varied brands standing in various degrees of complementarity to one another, such as Google Search, Google Maps, YouTube, Android, Verily (life sciences) and Calico (bio-tech R & D). Brand structures are particularly complex because they involve the interaction of human minds and the rich interplay of human understanding and imagination.

2. 2. Adaptability: there is plasticity in the connections among brands as a result of learning from experience. They have the potential for forging new connections (both new types of connections and new tokens of existing types) and severing old ones. Brand capital can undergo structural change, both in the internal configuration of its components and in its external connections with entities in its environment. For example, in 2015, Google's top managers changed the connective structure among the firm's brands by reorganizing them into a more efficient modular structure within the parent company, Alphabet.

3. 3. Openness: Boundaries of brands are partial or somewhat permeable. Brand networks are maintained in the midst of a continual flow of ideas, information and materials from the external environment and throughout the brand network. Corporate marketers cannot exclude co-creative consumers from participating in brand-building processes. Many actors, including those with no legal rights of access and use, can employ brands to derive service streams and to create new brand combinations.

4. 4. Stratified modularity: brands display nested, structural patterning at more than one level, with systems containing subsystems. For example, the Alphabet brand portfolio contains two separate brand modules: the renewed and slimmer Google brand portfolio and the ‘Other Bets’ portfolio. In turn, each of these two brand modules contains many individual brands. Consequently, each brand is both a system and an element of a larger system in which it is embedded by a particular connective structure.

5. 5. Recursive generativity: there exists potential to produce countless brand combinations by applying combinatorial rules to a menu of basic units. For example, the corporate Google brand can be combined with a wide range of product descriptors to generate new brand combinations, such as Google Search, Google Maps, Google Play, Google Glass and Google Fiber.

6. 6. Self-organizing character (polycentricity): the economy-wide network of brands (i.e. brand capital at the macro-level) is a spontaneous-order phenomenon generated by invisible-hand processes – it is the unintended social consequence of numerous actions of numerous decentralized agents pursuing their own interests under a set of general institutional rules (e.g. laws of property and contract). The assembly and coordination of the brand network is not the result of intentional human design or of external central control.

7. 7. Emergence: Brands have systemic properties that cannot be reduced to the plans of specific participants involved in the brand-building process. Brands produce effects (e.g. service streams) that are qualitatively different from what their constituent microelements can produce in isolation. For example, the Google brand subsumes trademarked elements within its boundaries; the brand is larger than, and qualitatively different from, the trademarked words and symbols that identify the particular source of goods and services offered by Google.

8. 8. Knowledge-generation: Brand-building and market-making processes generate classification schemes that are meaningful to the participants. Entrepreneurs and co-creative agents try to make sense of market phenomena through the use of particular classification systems for registering phenomena, grouping them into categories and forming connections among these categories. They adjust their classifications to changing circumstances (e.g. changes in human wants and service streams) in order to continue registering and classifying changes in the external environment effectively. For example, the creation of the Google brand helped to pioneer a new category of information-related services – namely a new way of organizing and accessing information on the Internet by using an online search engine.

9. 9. Robust-yet-fragile nature: The properties of a brand structure can be invariant with respect to one set of perturbations and yet the structure can be fragile for different properties and/or perturbations (Alderson and Doyle, Reference Alderson and Doyle2010). In spite of its dominance in the general online search services market, Google is still vulnerable to a major corporate scandal, ICT systems failures, security breaches, cyber-attacks and regulatory antitrust challenges.

As point 8 above suggests, the CAS perspective helps us to unpack the growth-of-knowledge aspects of brand building, market making and market shaping. As a kind of capital formation, the ongoing creation of brand capital is a social learning process. Knowledge is a not an accumulated stock of information but a structure of connections, and these connections change as knowledge grows in the brand-building process. Adjustments in people's classificatory schemes (knowledge structures) affect the uses to which existing brands are put in the short run and affect the kinds of brands that are created and scrapped in the longer run.Footnote ² Indeed, every brand and every brand structure is an experiment to extract meaning from the market. A brand is a tentative solution to a perceived market opportunity – a perceived gap in the existing structure of brand capital. Brand building is a series of tentative discoveries of these gaps. A new brand is based upon a set of entrepreneurial conjectures about the most urgent of the as yet unsolved problems of consumers and about the complementary bundles of brand characteristics that will satisfy these latent demands (Harper, Reference Harper1996; Loasby, Reference Loasby1999). All beliefs and expectations about brands are fallible and subject to refutation at any time. Through a process of trial and error-elimination, entrepreneurs try to discover the constellation of demand for the brand – including rates of first-time sales, replacement sales and repeat sales. End-users are the ultimate arbiters of whether a brand succeeds or not. It is they who exercise ultimate control over whether a brand dies or thrives. The history of marketing is rife with examples of how weak uptake and ultimate rejection by consumers have led to brand failures, such as Microsoft's Zune digital media player, Dell's Web PC, Heinz All Natural Cleaning Vinegar and Bic underwear (see Haig, Reference Haig2003).

From the growth-of-knowledge perspective, the ongoing adaptation of brand-capital structures means that brands are perpetually incomplete capital goods that are only a transient stage in an unending process of transformation. Brands (or at least certain aspects of them) are continually in the process of being formed, adjusted and creatively destroyed, and these changes are a vital part of market processes (Chamberlin, Reference Chamberlin1953: 3). Brands are always open-ended, yet-to-be-fleshed-out projects (‘half-baked cakes’) from the forward-looking planning perspective of prospective brand users. Indeed, they remain incomplete throughout their lifetime, and agents can never know the boundary of any brand ex ante. As market conditions and knowledge change, entrepreneurs continually reshape and remodel brands. Brand managers and other marketing specialists play a major role in the processes of ‘qualification and requalification’ of brands – a term of art that leading economic sociologists use to refer to how calculative agents continually (re)classify, (re)position, and (re)evaluate brands by comparing and relating them to other brands on the market (Callon et al., Reference Callon, Méadel and Rabeharisoa2002). Ongoing requalification constantly threatens the routine attachments of consumers to favourite brands as competitors jostle to capture and ‘attach’ new consumers for their own advantage by ‘detaching’ them from rivals’ offerings. Detachment occurs by repositioning one's brand in a way that grabs consumers’ attention, thereby prompting them to re-evaluate both their habitual purchases and the qualities they attribute to brands, so that they end up forming new attachments to one's own brand network. The process is iterative, requiring trial and error-elimination and sequential adjustments in the combinations of characteristics that define brands. Moreover, in service economies, ICTs facilitate more active participation by brand users and allow richer, more sophisticated processes for repositioning and re-evaluating brands. Digital technologies extend the social networks in which brand users interact as they share their evaluations and lessons learnt from trialling brands. Furthermore, in digital ICT environments, networked users are increasingly able to edit, adapt and reuse brands and turn them over to new uses, thereby changing their functional capabilities, attributes and how they are categorized. Hence, the set of uses to which brands can be put is prone to continual changes and contestation arising from distributed user practices, so that brand structures and product classifications are forever in a state of flux and new gaps are always being created in the web of linkages among brands.

Furthermore, the growth-of-knowledge perspective highlights that brands differ from individual products. Expressed in our terminology, brands are dynamic entities that persist through time and have lives, identities and personalities, whereas products are static, episodic phenomena that tentatively stabilize a combination of characteristics (Callon et al., Reference Callon, Méadel and Rabeharisoa2002: 197–200). Although brands were originally developed to protect products, today the situation is reversed: strong products serve to protect brands (Haig, Reference Haig2003: 7). Nowadays series of products (e.g. successive generations of iPhones, iPads, iMacs) linked into a brand architecture contribute to the perpetuation and continued value of the brand (e.g. Apple). Of course, brand capital cannot exist independently of actual goods and services. To function as capital, brands must always be instantiated or realized in physical structures of some sort: sets of concrete goods and services that deliver the brand experience to users, material symbols and devices in the external world (e.g. brand names, logos, slogans, product packaging), and also inner representations in the mind-brains of market participants, which are no less physical because they are mental. Brand capital is at least partly constituted by people's beliefs about the causal connections between the services yielded by particular brand types and brand tokens (e.g. concrete instances of the brand, particular products), on the one hand, and the satisfaction of human needs, on the other.

Indeed, the CAS approach to brand building entails a very different ontology of brand capital to the one advanced in Klein and Leffler's (Reference Klein and Leffler1981) seminal article on how brand names assure contractual performance. A brief comparison is warranted given the enormous influence of the latter. First, whereas we focus upon the interpretive-coordinative dimension of brands, they focus upon the incentive-alignment dimension. In their article, brand capital is a bilateral contract-enforcement mechanism between a firm-seller and a buyer that assures that firms will supply high quality output in the absence of a third-party enforcer. Having made extensive firm-specific investments (e.g. advertising), the firm has major incentives to maintain quality control in production and distribution. This quality-guaranteeing mechanism resides completely on the supply side, while the demand side comprises nothing but passive receptors of the firm's product. In contrast, we recognize that co-creative consumers also make sunk investments of time, attention and human capital in brand building (e.g. by creating their own websites and user-generated content devoted to the brand). Second, whereas our analytical focus is upon the structure of connections (complementarities) among brands and how they change, the focus of Klein and Leffler is upon elements (e.g. individual agents), supplemented by some heroic assumptions about connections (e.g. perfect, costless intercommunication among consumers). Third, Klein and Leffler's model is end-state economics that treats brands as fully formed entities, perfectly calibrated and synchronized to the requirements of equilibrium. Their neglect of brand-building processes goes hand in hand with ignoring (or assuming away) the growth of knowledge and consumer learning processes. Their model assumes that consumers already know the identity of firms and are aware of the existence of the relevant brands at the outset (1981: 617). Consequently, the model fails to capture one of the most important entrepreneurial functions of branding: grabbing the attention of consumers who do not know about the brand and making them aware of the brand's existence and desirable attributes (Kirzner, Reference Kirzner1973).

Finally, the CAS perspective offers promise for explaining the institutional aspects of brand-building processes. Indeed, the complexity approach both requires and depends on the inclusion of these institutional features. It sheds light on the specific institutional conditions of the emergence and change of brand capital. In particular, it treats brands as internal social institutions (in Lachmann's Reference Lachmann1971 sense) that originate endogenously through the market process itself and emerge from within the outer framework of fundamental external institutions, such as the legal system of property rights. The CAS perspective also highlights how branding processes can generate recognizable, recurrent patterns of interaction (e.g. between sellers and buyers) in the ever-changing flux of markets. These regular patterns of behaviour can themselves become building blocks for more complex patterns and innovative forms of interaction (e.g. brand communities). Robust brand-related patterns (e.g. iconic brands) can also crystallize into institutions that serve as orientation points for recombinant brand innovation. Indeed, recombinant brand innovation at the level of individual entrepreneurial plans requires relative stability at the level of institutions – namely, the more abstract rules of the game of contracting and branding. We now turn to the issue of brands as institutions in more detail.

3. Brands as institutions

The system of branding as a whole – both as a familiar form of economic organization and a particular social technology – is an economic institution. Institutions can be emergent outcomes of brand building within complex adaptive systems. The system of branding has been institutionalized in different ways in different historical, social, and cultural contexts, and it is not limited to modern capitalist economies (see, for example, Eckhardt and Bengtsson, Reference Eckhardt and Bengtsson2010). In addition, individual brands (e.g. McDonalds, Starbucks), which are constitutive elements of branding systems, may themselves become institutions or acquire an institution-like character over time. Brand networks are interpersonal rule systems (often tacitly held) that structure market interactions and social life. They can affect how we talk, how we think, how we feel and how we act. They tend to have a normative component that prescribes and enables certain forms of conduct and social organization (e.g. corporate brand management systems and franchising agreements), and that discourages or prohibits other patterns of behaviour incompatible with the brand's core values (e.g. brand extensions that tarnish the brand). Brand networks establish important distinctions between what actions and things are acceptable, authorized and authentic and what are unacceptable, illegitimate or fake. Agents can internalize these rules and classifications with the result that brands can elicit habits and routines that sustain rule-following behaviour. These rule sets also have an expectational dimension in that recurring patterns of observed behaviour come to be experienced by agents as expected modes of conduct. Brands thereby help economic agents predict the behaviour of those with whom they interact. They give order to the market process and increase the stability of buyer–seller relationships. Although a new brand might begin life as part of an individual entrepreneurial plan, as brands replicate and diffuse through markets, they can become common elements of numerous plans of entrepreneurs, suppliers, distributors, customers and other users.

Although strong brands operate as real constraints on economic participants, they are still softer than many constraints because there remains considerable scope for brand users to break the rules. The legal rights to brands are never absolute; they are always partial and incomplete. For instance, a trademark is only the right to the chance of obtaining some (but not all) of the future services of various attributes of a brand (Fisher, Reference Fisher1906: 22). Many brand users lack legal rights of use but they nevertheless have the power to put brands to novel unauthorized uses and to draw the fruits and profits from that use (that is, they can enjoy uncompensated services of brand capital). Institutional rules (including de jure and de facto property rights) affect who really gets to select the uses of brands, who precisely gets to modify and develop brand capital, and what decisions agents can make to transform, combine, reconfigure and scrap brands.

Brands can be relatively stable over extended periods of time. Some brands possess remarkable longevity – Coca-Cola dates from 1886, Colgate from 1806 and Quaker Oats from 1877. Brands are also becoming omnipresent. As brands are increasingly marketed worldwide, global brands become a common vocabulary for consumers all over the world. Brands such as Coca-Cola and McDonalds are ubiquitous, appearing in nearly every part of the world at any given time. Some brands (e.g. Xerox, Kleenex) may become so widely disseminated and institutionalized that the brand name becomes the conventional generic term in common parlance for an entire category of goods and services to which the brand belongs, with the result that the brand risks losing trademark protection (so-called ‘brand genericide’). In this case, the widespread linguistic practices of consumers and retailers transform what had been a proper name (a unique denotation of one particular brand) into a common noun (a general term for a class of goods or services or any of its members).

The relative stability and ubiquity of brands enables them to play an important orientative function. Brands help us make sense of our environment. Indeed, brands are habitually used constructs in agents’ ways of thinking, perceiving, feeling and acting. We all have a mental model of the major brands we consume, their interrelationships and their roles. We are typically unaware of this internal model until we have to search for alternatives because our favoured brand is out of stock or discontinued or because we move to a region where familiar brands are not marketed. For a large number of economic actors, brands serve as orientation points that guide their actions and interactions in processes of economic exchange and production. Brands have become ‘ideological referents that shape cultural rituals, economic activities, and social norms’ (Schroeder, Reference Schroeder2009: 124). They are ‘nodal points of orientation’ (Lachmann, Reference Lachmann1971) that provide a source of stability in the typical forms and features of economic plans, market transactions and market relationships. Brands are like tentatively anchored buoys that help participants to navigate the turbulent seas of market reality. Brands attaining this specific institutional form serve as ‘common signposts’ (Lachmann, Reference Lachmann1971) that guide and coordinate market behaviour. They become foci in the network of market relationships. They facilitate the process of attuning expectations and enhance the mutual adaptation of plans. Although there is variation in brand meanings across agents, the structure of an established brand is nevertheless common and stable enough to be interpretable by (and meaningful to) many groups so as to make it a recognizable scheme of orientation – so as to enable it to serve as a common reference point.

Brands are repositories of knowledge that can aid economic interactions. Like other institutions (Foss and Garzarelli, Reference Foss and Garzarelli2007), brands are ultimately knowledge capital. Brand names and trademarks are shorthand ways of conveying useful information – they are ‘compact identifiers’ that reduce consumers’ search costs, a category of transaction costs (Posner, Reference Posner2005: 62). For example, the two golden arches of the McDonald's logo act as visual landmark for the brand. Consumers find it less costly to search for a trademarked brand name (designating the desired offering of a particular producer) than to investigate and measure the attributes of all goods to determine which is the one they want (Landes and Posner, Reference Landes and Posner1987: 269). Brands thereby also economize on the costs of cognition. But this is only half the story. The mutual knowledge generated through recurring interactions in building brand relationships also creates additional transaction value and new service streams (Zajac and Olsen, Reference Zajac and Olsen1993). Furthermore, brands create connections and information flows between producers and customers without the need for direct transactions between them. They enable producers to ‘reach over the shoulder of the retailer’ (and across the distributor's web site) straight to the consumer so that producers can market their goods in their own name to the customer without the need for direct exchange or even direct contact (Schechter, Reference Schechter1927: 818).

4. Recombinant brand innovation by entrepreneur-producers

This section places the recombinant brand-building activities of entrepreneur-producers in the foreground while leaving other market participants in the background. That is, as a first step, we make the heuristic assumption that all other market participants (including consumers) are passive agents who lack entrepreneurial initiative and simply react to opportunities placed before them. (We relax this assumption in section 5.) In this section we also focus mainly upon the deliberate construction of brand architecture under a common form of business ownership, such as the firm.

At the micro-level, entrepreneur-producers (hereafter, simply ‘entrepreneurs’) create individual brands that comprise multiple elements over and above just brand names and trade names. Indeed, they create emergent ‘brand complexes’ that combine novel brand names with distinctive brand logos, unique packaging and product designs, and persuasive marketing communications (especially advertising). For example, the Coca-Cola soft drink is an amalgam of elements, including the brand name (‘Coca-Cola’), the distinctive typographical features comprising the flowing handwriting of the logo (this typeface was originally adapted from the elegant Spencerian script that was the dominant form of penmanship in the United States in the middle of the 19^th century), the shape of the bottle, the red-and-white colour combination and advertising slogans (e.g. ‘Taste the feeling’).

With respect to brand-naming practices, entrepreneurial experimentation is giving rise to the creation of new product-name elements and wide variation in combinatorial patterns as entrepreneurs pursue more and more product differentiation to distinguish their offerings from similar products in the market place. Names of branded products can include one or more brand names in their structure (e.g. the ‘Gillette® Fusion® Proshield™ Razor with Flexball™ Technology’). Entrepreneurs (and the specialists they employ) make use of natural language elements and conventional number sequences, and glue these components together to generate complex symbolic strings. Their naming practices can generate branded product names that contain several kinds of parts. An example is ‘PaperMate Sharpwriter pencils #2’, where ‘PaperMate’ is the brand, ‘Sharpwriter’ is the product group, ‘pencils’ is the product description, and ‘#2’ is the variant.

5. Generative networks as the institutional locus of brand building

The preceding section mainly focused upon recombinant innovation by entrepreneur-producers that involved brands under one form of ownership at the enterprise level. Whereas the previous section implicitly assumed that consumers were only ‘brand takers’ (passive economizers who take brands as given), this section relaxes this heuristic assumption and portrays consumers as an important class of ‘brand makers’ – alert, active agents who co-create and modify the qualities of brands. Accordingly, this section augments our perspective by emphasizing the open architecture of brand networks in which many different kinds of actors can co-create, use and combine brands. As a result of such co-creation, realized complementarities among brands can diverge greatly from intended complementarities: brand-capital combinations actually implemented can turn out to be very different from what entrepreneur-producers originally planned. Furthermore, the focus now is upon self-generating (i.e. spontaneous) brand networks, which can be significantly more complex than intentionally designed corporate brand architectures.

In this section we argue that generative networks (and not firms) are the real locus of entrepreneurial brand building. Brands are produced through a myriad of dynamic interactions and relationships created in generative networks. A generative network has the potential to create new brands, new complementarities among brands, new brand structures and even new generative networks. These networks exhibit creatively constructive dynamics. A generative network is a complex sociotechnical system that is composed of many economic agents and capital goods that are linked by connections of various kinds, such as social rules (including norms) and technological rules. The agents simultaneously or sequentially participate in value creation in response to perceived gaps in the capital structure at various levels.

What is of major importance is that the set of agents that bring brands into existence is not limited to entrepreneurs and the marketing managers they hire: lead users, co-creative customers, suppliers, advertising agencies, distributors, graphic and industrial designers, technologists and a diverse range of other interpretive agents, both inside and outside the firm, are all potentially involved in brand building. Hence, brand building is a polycentric process that involves many different kinds of interpretive actors who contest what the brand means, how it can be used, what income streams it will generate, who it will attract and how it relates to other brands. These networks are generative in the sense that interactions among the agents within the network can create new combinations and new uses of existing combinations. They can assemble resources into novel, potentially more valuable combinations. The network has boundless capacity to produce new brand structures by using existing brand elements. Generative networks are also generative in that they bring about the growth of knowledge – changes in agents’ conjectures (interpretations, expectations) and the revision of production plans.

Generative networks always operate against a backdrop of pre-existing institutional structure. They are supported by both codified rules (e.g. intellectual property law) and informal social norms that sustain social interactions. The impact of legal rules on brand co-creation in generative networks can be seen with reference to virtual communities. In digital gaming and machinima (machine-cinema), for example, players increasingly contribute significant original content that can be enjoyed, adapted, and then distributed by other players. As the game experience improves for all players, game developers are able to profit from these user-generated service streams and the enhancement of their brand capital. Harwood and Garry (Reference Harwood and Garry2014) found that the ambiguity in the legal rights to such user-generated content enhances creativity within the machinima community and is paradoxically beneficial to both game developers and players. Similarly, Watkins et al. (Reference Watkins, Denegri-Knott and Molesworth2016) describe how fragmented legal ownership to digital virtual goods (DVG) can have ambiguous effects in that it not only reduces consumers’ degree of control over DVGs in some circumstances but also leads consumers to continually challenge and chip away at the existing baseline of brand owners’ legal rights. In light of the mounting tension between legal ownership and authorship of co-created content, Linden Lab was the first to formally recognize players’ intellectual property rights to their digital creations in Second Life, a successful Massively Multiplayer Online Game (MMOG) (Herman et al., Reference Herman, Coombe and Kaye2006).

One of the hallmarks of a generative network is that knowledge (capabilities) and economic control rights are highly distributed in social space, and these rights can be disputed and contested by other agents. The upshot is that the creation of new brands (brand innovation) is ‘doubly distributed’, to use Yoo et al.’s (Reference Yoo, Henfridsson and Lyytinen2010: 730) terminology developed in a quite different context. Brand innovation is doubly distributed because: (1) de facto control rights over brand building are distributed across many heterogeneous actors of different kinds, including entrepreneur-producers and co-creative users; and (2) the relevant local knowledge and capabilities for building brands are spread among networks of people of different kinds, such as firms, households and brand communities, and subgroups of each. Co-creative consumers generate and disseminate new knowledge, much of it fragmented and contradictory, through the generative network. They also try out brand combinations that might not even have been contemplated by corporate brand owners. Consequently, brand building cannot be localized to individual agents. Recombinant capacity is distributed at a very micro-level – among networked individuals in digital markets who learn from one another's mixing and matching of brands as they publicize their brand-related experiments online on social media sites.

Of course, the implication of all this is that brand managers have neither the knowledge nor the power to completely control brand-building processes. Because decision and control rights to the benefits from using a particular brand are highly decentralized, managers cannot unilaterally determine the brand's actual reputation in the marketplace: they do not have full control over how the brand will interact with competing brands in the market, nor do they wield ultimate control over how consumers will respond to the brand. Consequently, in generative networks, the brand manager may instead become a ‘brand host’, whose key role is to cultivate a vibrant brand community, relationships, dialogue and collaboration (Christodoulides, Reference Christodoulides2009). Unlike the task of the artisan or engineer who constructs the final product in all its aspects, the brand manager's role resembles that of the gardener who strives to create appropriate conditions that are, on the one hand, more favourable to the emergence of what is desired and, on the other hand, adverse for undesirable classes of occurrences. However, these considerations do not rule out the possibility that managers can protect, influence and revitalize some brands or brand elements under particular circumstances. Even though day-to-day de facto rights of control are distributed throughout the generative network, CEOs and brand managers can still seek to re-assert significant residual rights of control over the brand, especially during a marketplace assault or a media fiasco smearing a brand, in a manner akin to how spin doctors and crisis managers wrestle to keep high-profile celebrities’ and politicians’ reputations intact while under siege. For example, Burberry's top management developed a counterinsurgency strategy to radically reposition Burberry as a global luxury brand after it was ‘hijacked’ by a lower-income, socially marginalized group (pejoratively referred to as ‘chavs’ in Britain), who adopted the brand and Burberry accessories as a conspicuous status symbol. In the face of sharply declining sales in the UK, the company centralized all its design activity, revoked clothing licences and re-emphasized its gabardine trench coat as its core luxury product.

As the above remarks about cultivating brand community indicate, the emergence of generative networks requires elevated levels of openness, heterogeneity, connectivity, interactivity and generativity. Generativity requires heterogeneity and ongoing variation in agents’ capabilities, interpretations and expectations and plasticity in social connections. It also requires common interests (e.g. a particular brand might be the focus of participants’ activities), a recurrent pattern of interactions, social rules enabling communication among agents, and joint action (Lane and Maxfield, Reference Lane and Maxfield1996: 228–9). The myriad fledgling brands being hatched on crowdfunding platforms illustrate these phenomena well. To take just one example, WINGStand, now a trademarked brand name, began when a Stanford student, Daniel Haarburger, conceived a new idea for niftily converting Apple's wireless keyboard into a dock for a tablet by using two small plastic clips. He prototyped the idea using 3D printers, then launched the project on Kickstarter. In less than seven weeks, more than 1,500 participants came together to form a whole community of brand builders that pledged financial support to bring the project to fruition. The crowdfunders participated as resource-owners because they had a latent demand for the proposed product (more than 2,500 units were pre-sold); they intended to become brand users once (and only if) the project was realized.

Generative networks can emerge spontaneously without any conscious design, or they can be set up deliberately by brand users (e.g. LUGNET) or be initiated and sponsored by the corporate brand owner (e.g. Harley Owners Group). Similarly, critics of a brand may deliberately set up unauthorized social network sites and try to highjack the brand (so-called ‘brand-jacking’) or such anti-brand networks may arise more organically, as with ‘We Hate Eastern Airlines’ (WHEAL) in the 1950s. Even if they are consciously created, however, generative networks for or against the brand will acquire an increasingly spontaneous or self-organizing character over time as the network responds to unexpected economic changes.

The socially distributed character of generative networks gives rise to brand structures that have self-ordering properties. Brands, and especially ‘open-source’ brands such as Linux, Wikipedia, and Python, are spontaneous rather than planned structures because they change in ways that their original creators cannot anticipate. Brand creation need not be intentional, the result of foresight or an antecedent goal to make a market. Brands can be an unintended consequence of numerous heterogeneous participants (e.g. members of online brand communities) engaging in activities they simply consider worthwhile. For example, the famous ‘Intel Inside’ logo and branding campaign did not begin life as a deliberate co-branding exercise. Japanese original equipment manufacturers (OEMs) that produced PCs encircled the phrase ‘Intel In It’ on their machines in order to reassure their customers that they used high-quality componentry and especially the highly valued Intel chip (Winer and Dhar, Reference Winer and Dhar2011: 190).

There can be no ongoing value co-creation or division of labour within generative networks without some form of communication, and there can be no communication without a system of signs. Hence, generative networks require that participants be able to use a mutually intelligible sign system (e.g. a natural language) and a communications technology by means of which participants can exchange messages. Participants use a sign system to express their own mental states (i.e. their desires, feelings, beliefs, intentions), to modify the mental states of others (e.g. through persuasive advertising), to indicate facts about referential objects (e.g. to make assertions about the availability and quality of a particular brand), and to critically evaluate statements (e.g. to challenge a seller's description of a branded product in comparative advertising). In short, they require a sign system that can perform expressive, appellative, representational and argumentative functions. The argumentative function requires a sign system whose syntax is capable of representing negation. This function is essential to the capacity for generative networks to bring about the growth of knowledge through trial and error-elimination.

Brand building is akin to an evolving multiparty conversation, an open-ended communicative process that is enhanced by digitization and digital ICT infrastructure. Brand building in generative networks has a dialogic character in that how participants respond to the initiatives of others affects the ongoing process of brand building. One participant takes the initiative by making an offer, and others respond to the initiative, by accepting or rejecting the move. For example, a fashionista posts a photo of a new pair of Manolo Blahnik shoes she is wearing in a blog for others to look at, and others respond to the post by entering their own comments online. This is a conventional behavioural pattern – a dialogue game that regulates how the conversation may unfold. The fashionista's post was a bid to start a dialogue game of a certain type. Conventional joint activities such as this require uptake mechanisms that consist of two types of rules: projection rules that indicate what brand initiatives count as a bid for a joint action, and response rules regarding what replies by other participants count as appropriate for the particular circumstances of time and place (Hulstijn and Maudet, Reference Hulstijn and Maudet2006). Uptake mechanisms might also operate when one participant, say the firm, takes the initiative to ‘extend the brand’ to a new related product category; the way in which other participants, say brand community members, take up the initiative that allows them to modify the course of brand building and brand extension. For example, in 2002 Porsche launched its first non-sports car, the Porsche Cayenne, as a mid-size luxury crossover SUV, but Porsche enthusiasts rejected it and refused to accept it as a genuine Porsche vehicle (Fournier and Lee, Reference Fournier and Lee2009: 5).

Generative networks, such as that described above for WINGStand, can emerge because participants recognize that opportunities for mutual gain can only be seized if they combine with each other in the pursuit of common brand-related goals. (To say that participants have a common interest does not mean that they only have common interests: they may also have conflicting interests in other respects.) Whenever brand users recognize that they and other users have common brand-related interests that can only be achieved by joint action (i.e. co-creation), they tend to be primed to identify themselves as members of a brand community. That is, if brand users recognize that their decision situation exhibits strong interdependence with those of other users, they will tend to adopt a ‘we frame’ (‘We are “Adult Fans of LEGO”’) and to jointly participate in creative brand-building activities with other brand enthusiasts (such as LUGNET, the LEGO User Group Network). For example, loyal fans of the Apple Newton (a precursor of the iPad) had a very salient common interest in keeping the Newton technology viable after Apple discontinued it and stopped providing technical assistance. In order to sustain and enhance use of the device, Apple Newton community members supported themselves with their own patches, software and hardware modifications.

Generative brand-building networks can also form around structures of brands rather than one brand in isolation. In particular, ‘communities of style’ can form on the basis of user connections to a combination of different brands from different suppliers (Pihl, Reference Pihl2014). Fashion bloggers and their readers inhabit these kinds of communities. The concept of style comprises the distinct manner in which consumers present, assemble, combine and adapt goods and services. Community members use a particular style (e.g. ‘young, urban, casual’) as a connecting principle to assemble different brand combinations (e.g. mass-produced H&M and Levi's apparel) and they engage in activities and social interactions that manifest that style.