1. 1
Exploring the Origins of New Transaction Costs in Connected Societies
Andreina Mandelli
2003
Chapter 9 (draft)
.snoitazinagro ni smetsys dna tnemeganam egdelwonk )sde ,tsurTnenoviI ajriM dna iratouH aneeL-ajiaM( nI
.snoitazinagro ni smetsys dna tnemeganam egdelwonk )sde ,tsurTnenoviI ajriM dna iratouH aneeL-ajiaM( nI
.snoitazinagro ni smetsys dna tnemeganam egdelwonk )sde ,tsurTnenoviI ajriM dna iratouH aneeL-ajiaM( nI
.buP puorG aedI
Introduction
There is a considerable amount of literature in management science, which claims that the digital
economy is a frictionless economy, where hierarchies and institutions disappear replaced by
dynamic and self-organized webs of companies and consumers, freely meeting on this new web of
opportunities (Bakos, 1997; Hagel, 1999; Hagel and Singer, 1999; Evans and Wurster, 2000).
The implications of this vision (that here we call the "paradigm of the frictionless Internet economy
and society") are huge. It may influence the way managers build market strategies and manage
organizations, but also the way policy-makers address relevant issues concerned with the so-called
digital divide in the knowledge society (Norris, 2000; Compaine, 2001).
This idea is not new. As Agre (2001) reminds us, conservative legal scholars, back in the seventies
viewed social progress, teleologically, as the progressive reduction of transaction costs, and thus
(they argued) the perfect approximation of ideal markets. But Rullani (1998) warns against this
fundamentalist approach to Internet society, because it simply reintroduces the ideology of the
invisible hand of the market and social darwinism, against any idea of collective action.
In previous works (Mandelli, 2001b) and in this study we have addressed the frictionless vision,
challenging the communication symmetry fallacy, on which is based the idea that the network
economy is automatically eliminating the information and institutional hierarchies (even though we
still believe that the Internet introduces radical changes in the way economic institutions are built
and the way businesses are conducted). We claim that the complexity of our interconnected world,
the evolutionary nature of trust and learning dynamics, and the economics of mediation (the
economics of relationships plus the economics of information infrastructure), play a major role in
both the creation and reduction of these new hierarchies in digital society. The result is complex and
not deterministically driven by network technology.
We challenge the frictionless paradigm, providing primary and secondary research support for the
idea that in the digital society and digital economy there still are cognitive frictions and hierarchies.
There is evidence about price dispersion and the role of hierarchical brands in digital economy;
evidence about the failure of business models based on the frictionless assumptions, data about
economic concentration in the Internet industries and infomediation flows, and evidence about
social frictions and new social transaction costs in building new relationships on digital webs.
Research from different disciplines has already addressed this issue. We try to include these
different empirical works in an unitary interdisciplinary framework (using theories elaborated in
social sciences, information sciences, management sciences and mass communication research) and
provide theoretical explanations for this new idea of the impact of technological networks on
society, looking for sources of hierarchies in the complexity of the new social systems, and in the
economics of information and economics of cognitive and social mediation.
Trust in our study is not seen as an automatic driver of hierarchy reduction in digital society; its role
is more complex, since it contributes to both the reduction of old hierarchies and the formation of

2. 2
new ones. We add original empirical evidence supporting the path-dependent idea of the dynamics
of trust, using survey data on a random stratified sample of 288 Italian Internet users. Trust builds
trust. Bonding social capital drives both new bonding social ties and new bridging (out-of-the-
community) relationships. But social capital building requires not only the investment of prior
social resources; it also asks for network social infrastructures (diffusion of connectivity services),
and for cognitive and time resources. The path-dependent nature of trust dynamics on digital
networks is also suggested by the role of values in influencing communication behavior and
attitudes.
The frictionless paradigm in the network economy and society: information symmetry and network-
based forms of governance
There is a major claim in the literature on digital markets (Evans and Wurster, 2000: Tapscott
(1996); Litan/Rivlin 2001; Bakos 2001): with the emergence of the World Wide Web, a totally new
business environment, approaching the ideal-type of the pure market in microeconomics, is
emerging. “One of the major features of the Internet revolution is its potential to make the whole
economic system, nationally and internationally, more competitive by bringing markets closer to the
economists’ textbook model of perfect competition, characterized by large numbers of buyers and
sellers bidding in a market with perfect information.” (Litan/Rivlin 2001, 315). Lower search costs
in digital markets "will make it easier for buyers to find low-cost sellers, and thus will promote
price competition among sellers.” (Bakos 2001, 71).
The Internet as a whole is conceived as one big open market, in which network communication
reduces the transaction costs significantly (especially those related with search costs) and therefore
it creates a new frictionless market, where every economic agent can find what he needs with very
limited costs associated to these cognitive searches and encounters (Evans & Wurster, 2000; Hagel,
1999; Bakos, 1997). It is the web of the "organizational plasticity" and the Electronic Business
Communities (Tapscott et al., 2000), the era of the Communities of consumption, of the empowered
consumer and reverse marketing (Bressler and Grantham, 2000; Hagel, 1999; Hagel and Singer,
1999; Levine, 1999; Kelly, 1997).
In the vision in which transaction costs are eliminated, market-like relationships, but networked by
cooperative spirit, are going to increase their relevance also in the organizational settings. "The
value chains that define a business, the supply chains that define an industry, the customer
relationships and brands that define a franchise, and the organization charts that define hierarchy,
power, and the boundaries of the corporation are all premised on the 'glue' of information. That glue
is progressively melting. The edifices of value chain, supply chain, customers, organization, etc. (all
of which are taken as givens in conventional strategy) are progressively deconstructed, breaking up
into separate entities busily conducting arms-length transactions" (Evans and Wurster, 2000, 5,6).
The Internet economy, according to these authors, is also eliminating the universal trade-off
between richness and reach --- richness being the quality of the information and relationships
available and reach being the number of units of relationship. The trade-off used to be simple but
absolute. Your business strategy either could focus on rich information - customised products and
services tailored to a niche audience - or could reach out to a large market, but with diluted
information that sacrificed richness in favour of a broad appeal. Much of business strategy rests on
this fundamental trade-off. Now, Evans and Wurster (2000) say, the new economics of information
is eliminating the trade-off between richness and reach, blowing apart the foundations of traditional
business strategy. The consequence at the governance level of management is huge. Frictionless
network-based teams and market relationships tend to substitute hierarchical organizations and
economic institutions.
The same intellectual framework is also usually used for predicting the diffusion of a community-
based new social order. Since the Internet will lower the barriers of time and space in
communication, people will build new communities of interest and new communities of emotions

3. 3
much beyond the limits of their local environment (Rheingold, 1993). These communities, for
authors like Hagel (1999) become the basis for the business model of the Internet economy. "The
real opportunity on the Internet is not just doing what you have always done cheaper and faster, but
instead the real opportunity is to rethink at a fundamental level the business models that you employ
on this new platform... Virtual communities actually started as spontaneous social events on
electronic networks ... We believe these spontaneous social events provide the foundation for a very
attractive business model." (Hagel, 1999, 1)
In this vision this aggregation and cooperation dynamics of virtual communities provide the ground
for a significant power shift, from the company to the consumer, which completely reverses the
information flow and the power in brand relationships, due to the new information symmetry and
democracy (almost) automatically brought about by the digital network. "We have a strong belief
that the network is enabling a different kind of market that we characterize as a reverse market. ... In
these reverse market situations, we believe it is about customers finding the right vendors at the
appropriate time" (Hagel, 1999,3).
In order to exploit economically these communities of consumers, Hagel and Singer (1999) suggest
that companies transform themselves in infomediaries, agents who act on behalf of the consumers
for providing them with the information they need, through the collection and the management of
all the consumers' data and profiles. This is the point where Hagel and Singer (1999) argument
meets Evans & Wurster's (2000). For the latter authors the most successful business model on the
Internet is the "affiliation" infomediation model, because the information economy eliminates the
trade-off between reach and richness strategies, between efficiency and efficacy strategies (both at
the consumer and company level) and give competitive advantage to relationships built on trust
instead of short-term self-interest. On the digital networks customers are going to have more and
more power, just because the interconnectedness reduces significantly the information management
costs and generates network externalities. Based on these assumptions the so-called "community
business model" has been widely applied in the Internet projects, both in the Business to Consumer
(BtoC) and the Business to Business (BtoB) area, building business models based on communities
of consumers and independent infomediary marketplaces.
This network-based idea of economic coordination is also applied to inter-firm relationships on the
supply chain (Hagel and Singer, 1999). In its most radical version (Adler, 2001), this network-based
model of inter-firm relationships envisions a future in which the new governance form ultimately
challenges the foundations of the capitalist form of society while simultaneously creating the
foundations of a new, post-capitalist form.
Challenging the frictionless paradigm
The frictionless vision of the network society surely grasps the novelty and disruptive character of
the Internet communication models and their impact on the economy, but we believe it doesn't
come to terms with the complexity of the Internet society, its economics and its changes. We need
a more fine-grained analysis of these changes, trying to understand more precisely the nature of
these new forms of governance and their new hierarchies (Mandelli, 2001b)
We claim that the frictionless paradigm doesn't describe the complexity of the Internet society
because:
1) It's not true that network connections automatically drive information symmetry and power-
shift in relationships;
2) It's not true that network connections automatically create social cooperation and
relationships based on trust;
3) It's not true that relationships based on trust are necessarily non-hierarchical.
Trust, delegation and legitimacy

4. 4
Delegation is a tool for managing complexity in complex systems. In Castelfranchi and Falcone
(1999) trust means delegation. This is why, differently from what it is true for the majority of
scholarly definitions of trust (Castaldo, 2002), we propose to focus on what the trustor misses,
instead of what he chooses. He misses variety. The trustor gives up power, when he gives up the
exploration of potentially better alternatives of cognitive mediation. At the system (or sub-system)
level delegation is the structure of power relationships in the system. So trust can drive cooperation
but this doesn't necessarily mean that trust lowers relationship hierarchies.
Trust is not the opposite of hierarchy because trust is a form of hierarchy (cognitive selection). But
it is also variation. All cognitive hierarchies are at the same time selections (delegation, reduction of
variation) and increase in variation (symbolic value added by cognitive mediations and new
associations). Mediation and delegation are not the same concepts, even though they are connected.
Mediation focuses on the constructivist value added by symbolic interaction in the cognitive and
social encounters. Delegation focuses on selection and complexity reduction.
If we analyze the infomediation role of trust, we can more easily understand its relationship with
delegation and power (Mandelli, 2001). Weber (1919) defined power as "the possibility of imposing
one's will upon the behavior of others." In later studies (French and Raven, 1959), power explicitly
includes the ability to influence values and beliefs. French and Raven (1959) define power as the
ability to influence others to believe, behave, or to value as those in power desire them to or to
strengthen, validate, or confirm present beliefs, behaviors, or values.
But power is not all the same. Since Weber's work on legitimacy (1919) the source of delegation
has been one of the core research questions in social studies. Coercive power is based on the threat
of force. Force is not limited to physical means; it is also social, emotional, political, or economic.
We know that delegation (opposite to coercion) can be based not necessarily on tradition and
charisma, but rather on shared rules and professional specialization. Weber defined authority as the
"means by which dominance was cloaked with legitimacy and the dominated accepted their fate."
Legitimacy can be based on charismatic affect, tradition, or legal/shared rules.
Also according to French and Raven (1959), power manifests itself in several forms; among these
are: expert power, reward power, legitimate power, referent power, and coercive power. In their
stricter definition legitimate power results from one’s being appointed to a position of authority.
Such legitimacy is conferred by others and this legitimacy can be revoked by the original granters.
Referent power is the affective tie to one's community or group. Expert power is based on the idea
that experts are in a better position for selecting the best solutions for the collectivity.
In Morris (2001 "something is legitimate if it is in accord with the norms, values, beliefs, practices
and procedures accepted by a group." (p. 2) In his perspective legitimacy is not concerned with self-
interest. "... the legitimacy of any feature of a social structure is indicated by the fact that it is
supported by those who have nothing to gain from it, even by those who would benefit from some
other structure" (p.5)
These ideas on how human communities legitimate power and delegation are crucial for
understanding how democratic institutions can be built on power asymmetries and delegation.
Representative democracy government mechanisms are based on delegated decisional power. It
may be important also for understanding the legitimacy of coordination of social asymmetry and
infomediation.
According to Schudson (1973) the shared rule and professional type of delegation is the way news
media legitimate their gatekeeping role in society. News values (Gans, 1979) define the
professional and the social responsibility rules of the game. Complexity in this case is managed by
delegated selection; selection is based on negotiated agendas at the policy level and shared social-
responsibility rules (McCombs et al., 1997; Semetko and Mandelli, 1997). In this view cognitive
and cultural hierarchies (information and value delegation) come out from professional delegation
and civic trust. Symbolic coercion (authority non legitimated) comes from the information divide,
based on the differential access to the gatekeeping system, influenced by economic, cognitive, and
social resources (Mc Combs et al., 1997).

5. 5
In our digital society the sources of information are multiplied and more difficult to select. Different
kind of new infomediaries on digital networks have entered the scene (in particular portals and
virtual communities). These higher-level social actors have different agendas and value priorities.
We must understand the logics of these new cognitive policy arenas in digital societies and
economies. This goal is too broad to be addressed in this work. Here we focus on one of its core
components: the role of trust in delegation and cooperation.
The symmetry fallacy, the new transaction costs and the economics of mediation
We argue that the reason why the dominant intellectual framework in Internet economics doesn't
help face the challenges of management in complex digital networks is because it is based on what
we call "the communication symmetry fallacy" (Mandelli, 2001b and 2002b). This is the idea that
technological symmetry between all pairs of nodes in digital networks (the potential total
connectivity provided by the digital infrastructure of networks) create a cognitive and relational
symmetry in the same pairs of nodes. It is the old "communication as transportation" metaphor,
critiqued in mass communication research (Carey, 1989), now applied to the networked world.
As mediations are not simply content channels, relationships are not automatic. They are context-
based and are the outcomes not only of choices, but also of inertia, randomness and learning
(Nelson and Winter, 1982; Dimaggio and Powell, 1991; Macy, 1998; Rullani and Vicari, 1999;
Castelfranchi, 2000). Choices can be instrumentally and not instrumentally driven (Castelfranchi,
1998), but they are not "order for free". They cost technology, information, trust, values, attention
and time resources at the node level, even when the cooperation between those pairs of nodes is
self-organized/ not coordinated centrally (Mandelli, 2002).
Friction-free relationships and friction-free markets are not more common on the digital web than in
traditional communication spaces, because transaction costs at the node-level on digital networks
are not eliminated, they are just changed. For the particular cognitive logistics of the Internet
(Mandelli, 1997; Mandelli, 1998), it can be easier to reach out somebody (once we know where to
look for) and switch from him to another partner, but it can be much more difficult, though, to
establish an efficient and long-lasting relationship, considering the complexity of these dynamic
encounters, their uncertainty, and the cognitive and social resources we have to bring into
(Mandelli, 1998; Jarvenpaa and Leidner, 1998).
The communication symmetry idea doesn't take into account the complexity of social and cognitive
networks we are creating thanks to the technological web, and the individual and system-level need
for new hierarchical (though dynamical and flexible) structures of selection and delegation, made of
different layers of dynamically interconnected cognitive worlds (Mandelli, 2001).
The availability of a technologically frictionless communication channels doesn't increase
automatically information and relationship variety for social mediation. We reduce our access to
diversity in relationships because of cultural inertia (Dimaggio and Powel, 1991) or according to
our need of economizing on cognitive investments (Neuman, 1991; Mandelli, 1997b and 2002).
But we also reduce our access to diversity because of the information asymmetries structured by the
economics of infomediation and content (Neuman, 1991; Mandelli, 1998; Shapiro and Varian,
1998; Vicari, 2001).
But if network society is not the all symmetrically connected society that the new frictionless
paradigm envisioned, it is also very different from the only-locally connected societies of the past.
If societies are communication networks, the structure of the network matters. Digital
communication provides a different infrastructure and structure to our network society. If we study
Internet relationships using the approach suggested by connectionism applied to virtuality (Vicari,
2001), we can recognize new hierarchies at the level of the structural architecture of connections, at
the level of the stock of resources available to the individual nodes in relationships, and at the level
of the connection activation mechanisms (Mandelli, 2001).

6. 6
We can easily grasp this, applied to the so-called digital divide problem in the network society
(Norris, 2001; Dimaggio and Hargittai, 2001; Cawkell, 2001). People have access to the
technological connections of the new social network (they become Internet users) depending on
their social and cognitive resources (technology adoption, education, computer literacy).
Researchers call this the first dimension of digital divide (Norris, 2000). Also, they have access to
the information and the relationships on the network, depending on their ability to reach
(technologically and cognitively) content and other people and being reached by content and other
people. This content and relationship access (the "democratic divide" in Norris, 2000) is limited by
the infomediation/gatekeeping systemi and its business model (figure 1, Mandelli, 2001). A general
portal is less precise than a specialized navigation service. A walled-garden portal like AOL limits
more the options and the diversity of navigation sources than the other types of portals or free
navigation on the World Wide Web. But this access is also limited by users' language, and users'
stock of available tangible (money) and intangible resources (attention, time, cognitive
sophistication, social capital).
Individual
Cognitive
Individual network
Cognitive
network
e
ivid
tic d
ocra
Dem Individual
Cognitive
Infomediary network
Content
Te
ide
Inf
ch
oa
div
div
cc
ess
ide
ch
Te
Individual div
Cognitive ide
network
Individual Content
Cognitive Content
= Technological connection network
= Communication connection
Figure 1 Hierarchies on the network - source: Mandelli, 2001
The infomediation structures on digital networks can be very different (Mandelli, 1999); they can
be very complex (different distributed layers and actors of infomediation, all connected by
hypermedia rings of collaborative services) or very top-down and simple like in the walled-garden
model. They respond to specific functions (trade-off between cognitive efficiency and diversity
access for the user) and have different legitimacy for their gatekeeping role (Mandelli, 2001).
Besides accessing the content differently, Internet users can also interact with content and people
differently, depending on whether the formats of the channel and the medium are designed to allow
interactivity, but also whether they are able and interested in doing it, according to rational and non-
rational rules of activation.
Each elementary and second-level node on the network faces transaction costs (costs of cognitive
mediation) when they activate connections with other nodes and sub-systems. These transaction
costs are made by all the resources invested in the symbolic mediation:

7. 7
• Monetary cost of connection (if the infomediation infrastructure is not based on free-
connection business models)
• Time
• Attention
• Knowledge
• Social capital
• Freedom (privacy)
Knowledge is to be conceived both as cognitive sophistication (education and technological literacy
for individuals) and cognitive frameworks. Social capital is defined in Putnam (1993, 167) as those
"features of social organization, such as trust, norms and networks, that can improve the efficiency
of society by facilitating coordinated actions."
All the intangible resources produce costs, even though knowledge and social capital are subjected
to the law of increasing returns (Shapiro and Varian, 1998), meaning that their value doesn't
decrease with the use. Knowledge and social capital don't devalue with the use, but they require
often huge investments to be built, that is they create sunk costs. Networks still have sunk costs at
the node level, but these sunk costs are different than before; they are mostly based on intangible
investments. We know that sunk costs drive lock-in strategies (Shapiro and Varian, 1998) and this
is one of the major sources of hierarchy on the social and economic networks.
More in general, the new transaction costs have the following sources:
• The cognitive complexity of the increasingly interconnected system;
• The economies of scale (supply side and demand side) and scope in the information
production;
• The tangible and intangible resources required for accessing the content and the
infomediation infrastructure of the network;
• The tangible and intangible resources requested by the relationships with the connected
nodes.
New hierarchies in the network society
Since the spread of the frictionless paradigm in Internet research, few authors have disputed its
major claims from different fields (Shapiro and Varian, 1998; Porter, 2001; Norris, 2000 and 2001;
Mandelli, 1998 and 2001b), even though not in a unitary framework. The core argument of the
frictionless paradigm is the idea that the Internet eliminates transaction costs and communication
asymmetries. If we challenge the idea that the Internet eliminates transaction costs and
asymmetries, we seriously cast doubt on the entire frictionless paradigm.
In this work for economic paradigm we intend (as in Rullani, 1998) a coherent abstract system of
technology, organizational model, type of marketing and labour relationships; "an abstract
framework that theory can build and practice can use ... an intelligent reducer of the social and
natural complexity, which is selected by a cognitive filter that drives it toward the production of
value. ... a kind of collective intelligence." (Rullani, 1998, 30).
We accept the call in Rullani (1998) for challenging the fundamentalist answer to the crisis of
fordism, which mixes a naive mythology of digital revolution with social darwinism. Citing
Habermas he reminds us that this fundamentalism "... which is nurtured by technology and nature
laws, is in search of a program of technological transformation driven by the force of self-organized
systems, with abdication of the political power and the sacrifice of collective action. It is only
apparently a liberal form of post-fordism; actually it is an authoritarian form ... because, without the
old constraints, the new dynamic technostructures and the interests which lead these social
transformations can easily dominate all the others." (p. 27).
For achieving this goal we need to clarify what we mean with the term hierarchy. In this work we
use three ideas of "hierarchies":

8. 8
• The first one is strictly linked to the idea of information asymmetry, the unequal access to
relevant knowledge by relational parties; relevant for this study is the unequal access to both
diversity and richness of information;
• The second one is the idea of power asymmetry in dyadic relationships, based on selection
and delegation (when people explicitly, even though not always because of their willingness,
give up variety options in their content and social encounters, even if they have access to
them);
• The third one is, at the network level, the more complex idea of organizational hierarchy, as
the governance institutional alternative to the market form of coordination, in the
transaction-cost theory approach to organization (Williamson, 1975).
The not-so-frictionless digital economy
Research extensively found that relationships between consumers, relationships between companies
and relationships between companies and consumers in the digital economy don't follow the
patterns predicted by the frictionless model of the Internet economy, both in BtoC and BtoB
markets. As we are going to review in the following paragraphs, there is solid evidence that:
• Inter-firm networking has demonstrated to be constrained by cultural and social frictions;
• The community business model in e-commerce has proved widely unsuccessful both in
BtoC and BtoB;
• Information and organizational concentration has demonstrated to be an important
competitive lever also in the digital economy;
• Price discrimination in digital economy has been found, supporting the idea that brands have
a differentiation role also in the uncertain territories of the Internet markets;
• The establishment of new relationships on digital networks requires the investment of
relevant cognitive and social resources.
These results may be interpreted as the support to the idea that the Internet is "business as usual".
Here we try to bring theoretical support and evidence for the opposite view. The Internet diffusion
does indeed call for a change in business management and marketing paradigms, but in a different
(and more complex) way than it is predicted by the frictionless and reverse-marketing vision.
We claim that in the Internet economy there are different kinds and degrees of digital hierarchies,
because there is not easy and equal access to knowledge, because there still are transaction costs
and social resource scarcity, and because organizational concentration in the content and
infomediation activities still matters. The sources of the new hierarchies are to be found just in the
network nature and economics: the economics of content and gatekeeping (infomediation), the
cognitive logisticsii of the web, the increasing complexity of social networked systems and the
economics of individual cognitive mediation and selection (bounded rationality and bounded
sociability).
Hierarchies in the inter-firm networks
Adler and Kwon (2002) propose to go beyond the dicothomy between the market and the hierarchy
organizational concepts, in the study of post-capitalism. In their vision networks are not just
hybrids, they are a new ideal-typical forms of organizing, with its new corresponding coordination
mechanism: trust.
Several authors predict that dynamical network-based organizational forms will replace hierarchical
organizations (Saxenian, 1994; Fukuyama, 1995 and 1999; Adler, 2001; Adler and Kwon, 2002), in
a world where complexity makes fixed contracts and hierarchical ties incapable to address the
growing need to adapt to environmental changes, and technology offer easier way to build dynamic
connections. We address this issue more in depth in Mandelli (2003). Here we focus on empirical

9. 9
studies that provided evidence challenging the idea that technological networks can easily and
almost automatically transform in network-based coordination.
In the BtoB Internet world, vertical e-marketplaces managed by independent infomediaries were
assumed to become the new network business model (Bakos, 1998; Kaplan and Sawhney, 2000). It
is widely recognized now that the main reasons why these open network-based marketplaces have
failed were their inadequacy to deliver a superior value proposition to the members, and a lack of a
legitimated and trusted system of knowledge-sharing (Grewal et al., 2001; Devine et al., 2001¸
Hoffman et al., 2002). In Grewal et al. (2001) study on a vertical e-marketplace in the jewelry
industry, the researchers applied the motivation-ability framework (Merton, 1957) to inter-firm
cooperation, and found that motivation to cooperate was both efficiency-driven and legitimacy-
driven (as also suggested by the new-institutionalism theoretical framework, see DiMaggio and
Powell, 1983). None of these expectations were met.
The new supply-chain collaboration formulas (in which the incumbent players orchestrate the
networks) have proved more successful (Hoffman et al., 2002): Devine et al. (2001) report that
though buyers can shop for a better price elsewhere, they have found that they are rarely inclined to
do so and prefer private exchanges. Customer relationships built on trust (and supported by
nondisclosure agreements) are essential for knowledge sharing (for example if suppliers have to
monitor a customer’s sales and inventory levels, to forecast product demand, and to assure the
delivery of goods or services as needed). Also McDuffie and Helper (forthcoming), in their study of
global Internet consortia, found that social interaction patterns (different by company history,
country, laws, institutions, geography and resources) have been critical variables in these network
projects. These new networks are not "business as usual" but neither frictionless webs of
relationships. They provide a completely new, more efficient and richer, communication
environment to previous economic communities, consolidated by accumulated knowledge and trust.
One lesson that we can draw from this experience is that managers cannot build communities in
laboratory; they emerge historically from the complex dynamics of relationships in industries and
society. Also Internet business networks are built on frictions. Communities are built on social
capital and willingness to share knowledge, and these two assets are not readily available.
Institutions emerge when they offer to all the economic actors added value compared to the
alternatives. In this case, open community-based value networks failed to deliver higher value to the
involved nodes than it was possible through more hierarchical organizations. Social capital is the
relative advantage of the new hierarchical formats; the cost of building trust was the transaction cost
(and the main reason of failure) of the previous open formulas.
This phenomenon can be better understood using the conceptual tools offered by the research on the
role of trust in building competitive advantage in post-fordist economies (Barney and Hansen, 1994;
Vicari, 1995; Costabile, 2001; Busacca and Castaldo, 1996) and the literature on social inter-firm
networks (Grandori and Soda, 1995; Kogut and Zander, 1996; Koka and Prescott, 2002). Trust, in
the resource-based view of management, is considered one of the intangible assets of firms, which
build on it for achieving differentiation and sustainable competitive advantage. It works as strategic
(and not-easily-replicable) asset just because it is not readily produced.
Also the organizational network literature provides support for the idea that social capital is not an
easy-to-build asset. Koka and Prescott (2002) define social capital, in organizational contexts, in
terms of the information benefits available to a firm due to its strategic alliances. Using longitudinal
data on the population of strategic alliances formed during the period 1980-1994 by firms in the
global steel industry, they provide evidence that social capital yields distinctly different kinds of
information benefits in the form of information volume, information diversity and information
richness. But these benefits are different because the distribution of social capital is not equal.
Networks are not all the same. Not all the members of a network are the same. Some members are
more equal than the others.
The resource-based approach (Barney and Hansen, 1994) has had a significant impact on how we
see the way firms leverage their resources and capabilities, in particular knowledge and social

10. 10
resources, for building competitive advantage. The emergence of intangible resources as a major
component of firm capital, often overshadowing traditional capital (Barney and Hansen, 1994;
Vicari, 1995; Costabile, 2001; Busacca and Castaldo, 1996) has contributed to refocusing strategic
management on organizational advantage (Nahapiet and Goshal, 1998). The firm has been therefore
redefined as “a social community specializing in speed and efficiency in the creation and transfer of
knowledge” (Kogut and Zander, 1996). The trade-off between market and hierarchy, is illuminated
by a new perspective. The reason why organizations emerge is not to leverage tangible resources
anymore, but rather to leverage the intangible unique sources of competitive advantage: knowledge
and social capital. This approach to institution formation can be applied to both firms and networks.
Even in the over-studied Linux case (Browne, 1998; Axelrod and Cohen, 1999; Moon and Sproull,
2000) of open-source and distributed software development, researchers found that the project was
working because it was based on specific and not-generalizable resources: a natural community
with strong commitment (what can be more affectively motivating than the goal to outperfom
Microsoft's operative system?), trust and diffuse expertise.
Also speaking about organizational social ties we can fall in the symmetry fallacy we have already
described for information flows, when we take a technologically deterministic stance.
Technological connections don't necessarily create social connections. Network-based relationships
don' t simply stem from technological connections. They are output of choices, but also socially and
culturally emergent phenomena. Their structures are influenced by the history of prior relationships
and the stock of available social resources. In Adler and Kwon's (2002) perspective "Social capital
is the resource available to actors as a function of their location in the structure of their social
relations" (p. 18)
Networks are not cost-free coordination formats. We can consider social capital as both a cost and a
benefit of relationships (Adler and Kwon, 2002). It is a cost because it requires investments but also
"it needs maintenance" (Adler and Kwon, 2002, 22). Networks have coordination costs, which stem
from the organizational complexity of these new forms of organizing (Gulati and Singh, 1998).
Network formation is a path-dependent, evolutionary process, and trust can help diminish
transaction costs only after a complex and long history of social investments (Lorenzoni and
Lipparini, 1999).
Also, social capital in networks can create frictions and inertia, instead of liberating creativity and
innovation. Local cultural and social forces can hamper the ability of networks to go beyond the
originally local base, in search of optimization coming from the diverse and global reach of the new
technologically connected networks (Powell et al., 2002). Tsai's (2000) study shows that also in
intra-organizational networks, prior network centrality, trustworthiness, and strategic relatedness
significantly affect the rate of new linkage creation and network structure.
One of the examples often used to show the relevance of the network-based model of inter-firm
governance is the "local distretti" form of organizing of the Italian small companies (Saxenian,
1994). These local network-based industrial systems are characterized by flexibility but also by the
ability of adaptively learning in a cooperative context. But in these network-based systems
relationships are not symmetrical (Grandori and Neri, 1999; Brown et al., 2002; Dagnino and
Padula, 2002), and cooperative learning is complemented with control and command, reproposing
the old role of the bigger and more powerful firms in the construction of new dynamic networks in
local and global economies (Gottardi, 1998). The structures of networks evolve and change over
time, and they can be designed and managed by network leaders (Lorenzoni and Lipparini, 1999).
Technological diffusion increased the geographical outreach of the networks originally only local
but this also increased the importance of the local leader firms and their role in leading vast
networks of smaller and subordinated firms (Gottardi, 1998). This system integrates competition
and cooperation, but "it is governed by the leaders" (Gottardi, 1998, 142).
Also the cases of wide global networks of smaller firms organized by powerful orchestrators as
Nike and Cisco (Brown et al., 2002) confirm this highly hierarchical option of network-based
governance forms. The hypothesis that networks drive almost automatically cooperative behavior

11. 11
doesn't correspond to empirical evidence. Studying the importance of the relational dimension of
inter-firm exchanges, Wathne et al. (2001) found that interpersonal relationships between buyers
and suppliers serve as a switching barrier but are considerably less important than both firm level
switching costs and marketing variables.
Gulati and Singh (1998) found that previous social ties reduce coordination costs and the need for
hierarchical control, but they also found that reciprocal interdependence (we can call it complexity
of the system) increases need for coordination hierarchy.
In short, networks are not simply a more democratic and socially rich alternative to hierarchies,
coordinated through trust instead of price or authority. They include elements of markets and
hierarchies, and are coordinated through different control mechanisms: prices, trust and authority.
Control mechanisms require investments at the node level and at the social level. The real novelty
of networks doesn't seem to be the differential degree of hierarchy, but the differential degree of
interconnection (complexity) and flexibility (dynamism of connections) of the systems.
This is consistent with the recent call of researchers (Dagnino and Padula, 2002) for studying the
interconnected (and not dichotomical) role of competition and cooperation in networks. They
suggest rebalancing the focus of research attention, in order to study the "variable-positive-sum
game" and the potentially "unfair" nature of the network mutual exchanges (Dagnino and Padula,
2002).
The concentration in the content and infomediary industry: the new gatekeepers
According to Shapiro & Varian (1998), the economic impact of the Internet and corporate computer
networks is similar to that of earlier networks such as the railways, telephones and bank machines.
The authors claim that ''information rules'' are simply more radical versions of the rules that have
always applied to high-fixed-cost, low-marginal-cost industries. Information industries in general --
Media companies, Internet companies, software companies-- all have in common the combination
of high ''fixed costs'', low ''marginal costs'' and ''network externalities'' (the more people use the
network service the more people will find it worth using it). This generates huge economies of scale
at the supply side (economies of scale in the production and distribution of information - the so-
called first-copy economies in Neumann, 1991), and at the demand level (benefits for the individual
members of the network are dependent on the size of the network itself), as well as huge economies
of scope (the more the same customer buys from you the less is the unitary cost of your relationship
with her).
The rules of the information economy, plus the scarcity of the attention resources at the node level
(their bounded rationality), drive market concentration, not distributed market power. This
information economy creates not only emergent information hierarchies but also explicit strategies
for controlling the navigation options and information diversity accessible by customers. Price
discrimination, product versioningiii and lock-in strategiesiv (though hurting the interests of the
customers because they reduce their access to diversity), generate higher profits for the company.
The world of the information economy, from this standpoint, is made by huge corporations and
trapped customers. The result can be a second-level digital divide: a "democratic divide" (Norris,
2000).
"Small foundations and educational institutions simply can not compete with the vast production
budgets of major commercial portals like Yahoo and Excite. Further, if people want to find such
non-commercial spaces, they may find it difficult due to search engines that steer people to the
continually growing number of commercial web sites." (Neuman et al. 1999). This drive toward
concentration is accelerated by the more diffused business model on the net: the advertising-based
business model. Since the infomediaries (portals of all sorts) need advertising revenues, they tend to
build big audiences and this - along with the economies of scale - makes the competition for shares
of web traffic very intense. On the other side, the economies of scope and the business models
based on affiliation commercial agreements drive the development of the so-called "walled-garden"

12. 12
model of the information assortment. This means that the big portals tend to reduce the number of
information sources accessible and to control the rules of search results in order to respond to the
expectations of commercial partners (big manufacturing firms, retailing and information brands on
the net).
The resulted concentration of what we call the "activated navigation options" on the World Wide
Web is clear from the following chart (figure 2). It refers to the distribution of the total attention
invested by the users on the web in Europe, measured in monthly usage minutes.
Number of web usage minutes per month, by
number of companies to which usage is
attributable - Europe (Jupiter, 2001)
120
100
80 50% of all minutes
60
60 % of all
40 minutes
20
0
March 1999 March 2000 March 2001
Figure 2 Concentration in European Internet infomediation
Online consumers respond to overload of information by dedicating their attention to a very limited
fraction of online shops. They select and give up diversity. Netratings in September 2002 calculated
that the average Internet user accesses only 49 domains in a month. We know that the number of
available domains on the web were more than 36 million already in 2001 (source: Coley
consultancy, 2002)
This doesn't sound new for students of mass communication. From media economics we know that
information economies drive concentration. This is true for the interactive networks as it was true in
the old content and information industries (Neuman, 1991).
We also add another reason, often disregarded, why we should be aware of the potential increase of
economic concentration, specific to the era of the Internet; we first studied this phenomenon applied
to marketing (Mandelli, 1998). The Internet not only decreases connection costs. It also decreases
coordination costs. Digital technologies not only foster the efficiency of cooperation processes
(knowledge conversion through electronic socialization), they also increase the efficiency of the
explicit and codified forms of knowledge conversion (Mandelli, 1998): knowledge externalization
and re-combination using Nonaka and Takeuchi's (1995) words. It is like to say that, if it is true that
the Internet technology reduces the transaction costs concerned with the distant and dynamic
cooperation (therefore fostering the market-type coordination of the interdependencies), it is also
true that it reduces the cost of organizational hierarchy-based coordination based on structures and
standards of infomediation. The trade-off that generates new institutions doesn't disappear. It moves
at a new and lower absolute cost level.
This helps also explain why the predicted disintermediation on the Internet is not in sight. Carr
(2000) writes that like many of the early assumptions about electronic commerce, this one has
proved "laughably wrong." It is now becoming clear that business is undergoing precisely the
opposite phenomenon - what Carr (2000) calls "hypermediation". Transactions over the Web
involve all sorts of intermediaries, not just the familiar wholesalers and retailers, but content
providers, affiliate sites, search engines, portals, Internet service providers, software makers, and
many others (Mandelli, 1998). And it is these middle men that are capturing most of the profit
(Carr, 2000). You could easily imagine this outcome of the Internet changes, if you studied the

13. 13
different but contemporary changes in transaction costs and coordination added value for both
distributed-direct relationships and for hierarchical and intermediated links brought about by digital
network technologies (Mandelli, 1998).
Brands as hierarchies
The frictionless vision of the digital economy tends to focus on the elimination of the information
frictions, and predict a cooperative, dynamic, network-based coordination, without studying in more
depth the real role of trust. In this vision, trust is seen as automatically linked to cooperation and
symmetry in relationships, whereas we propose to consider the role of trust also in building
relational hierarchies, when power in the relationship is asymmetric, as it happens in brand
relationships.
Shapiro and Varian argue against the idea that brands will be replaced by perfect competition.
"Visionaries tell us that the Internet will soon deliver us into that most glorious form of capitalism,
the friction-free economy ... we agree that the Internet will make shopping easier than ever, but
much of the talk about friction is fiction. (Shapiro and Varian, 1998, 142) Frictions are embedded in
buying history and switching costs. They write: "You don' t have to drive to the store to order a new
computer, but your choices for the future will still be hemmed in by the selections you made in the
past. Like it or not, in the information age, buyers typically must bear costs when they switch from
one information system to another. ... When the costs of switching from one brand of technology to
another are substantial, users face lock-in. ... Lock-in can be a source of enormous headaches, or
substantial profits, depending on whether you are the one stuck in the locked room or the one in
possession of the key to the door." (Shapiro and Varian, 1998, 143)
Porter (2001) starts from another perspective. He doesn ' t study the specific rules of the information
economy, even though he ends up admitting that the Internet can increase competition and
challenge competitive advantages based on traditional sources of market differentiation. But what
he doesn't accept is the idea that in this new turbolent economic environment firms do not have
opportunities for differentiation. Unique products and unique brands will have the same power (if
not more) as they had in the past.
Research efforts have found support for the idea that there are lock-ins and information hierarchies
at the brand-consumer relationship level, in digital economy. According to Reicheld and Schefter
(2000), who studied consumer behavior in this new marketspace “price does not rule the web,
loyalty does” (p. 106). Several studies have tested the hypothesis of perfect price competition on
the Internet (Smith et al., 1999; Brynjolfsson and Smith, 2000; Pan, Ratchford and Shankar 2001;
Chen and Hitt, 2001; Johnson et al., 2000; Latzer and Schmitz, 2001; Ancarani and Shankar, 2002).
They found that price discrimination and the role of brands in market relationships were very
similar on the Internet as in the traditional markets. "... several studies find significant price
dispersion in Internet markets. This price dispersion may be explained by heterogeneity in retailer-
specific factors such as branding and trust, retailer efforts to build consumer lock-in, and various
retailer price discrimination strategies" (Smith et al., 1999, 26). Lack of trasparency might be not
the only source of price dispersion; people might be easily willing to spend more money for the
same product, even though perfectly informed about the alternatives, if they invest in a branded
relationship, trusting the company about future differences. Brynjolfsson and Smith (2000) write:
"In light of both existing theory and the earlier results on price levels and price changes, the
dispersion in posted prices is surprisingly high. ... At the same time dispersion in weighted prices is
lower on the Internet than in conventional outlets — reflecting a dominance among certain heavily
branded retailers. Given these findings, we analyse potential sources for the high degree of price
dispersion on the Internet. We conclude the Internet price dispersion may arise from two different
sources of retailer heterogeneity: heterogeneity in customer awareness, and heterogeneity in retailer
branding and trust. We also note that, far from being equalized, these differences among sellers may
be amplified on the Internet as compared to conventional channels." (Brynjolfsson and Smith, 2000)

14. 14
These results are particularly important because they come from studies on products considered
homogeneous, like books and CDs (we believe that the same study applied to more complex
products would have found even higher price dispersion). But this also may support the idea that in
a relationship economy the source for differentiation doesn' t come necessarily from the core
product, but rather from the service that surrounds it (Mandelli, 1998). Research results from Lynch
and Ariely (2000) and Shankar, Rangaswamy, and Pusateri (2001) support this hypothesis, since
they found lower price sensitivity on the Internet when more information is offered. Rajgopal et al.
(2000) found that the quality of user experience can build a relationship competitive advantage. In
an experience economy (Pine at al., 1999) what we buy is not the product but the complex
experience associated to this buying behavior. If so, it is even more difficult to predict a perfect
competition in the Internet markets, since an interactive environment makes the task of building
complex services and experiences around simple products easier than before (Mandelli, 1998).
Amazon doesn' t compete on books. It competes on the quality of their services, and on the
sophistication of the user experience on their web-site (ease of navigation, personalization services,
feedback from the readers, etc.).
Latzer and Schmitz (2001) give the following examples for providing evidence for lower than
expected market transparency:
1) Search engines cover only a small fraction of web-sites (0,03% according to Bergman,
2001) and e-commerce companies have means to manipulate the perception of the search
results (Sullivan, 2001);
2) This is the reason why the traffic is very concentrated (they provide data for the Austrian
market similar to those provided in the previous paragraph of this chapter;
3) Consumers tend to search very few shopping-sites and the fraction of shoppers that stop
their search after the first site visited is high (Johnson et al, 2000);
4) The most important criterion for consumer choice in BTOC e-commerce is brand name not
prices (they provide primary evidence from a survey conducted on a national sample of
Austrian users and meta-analytical research data supporting this hypothesis).
Brynjolfsson and Smith (2000) warn that unexpected (and contrary to frictionless hypothesis)
results could be due to the immaturity of the Internet markets and easily change in the future. We
tend to not agree. Even though it could happen that shopping bots and distributed intelligence in the
market will diffuse and make prices more transparentv, it is not likely that brands will become less
important, since the source for differentiation and cognitive hierarchy (the need to shortcut in
complexity and the need to trust somebody in the light of an uncertain future) doesn't.
Trust in brands is the major driver of infomediation selection on the web, along with trust in social
ties. Using data from the World Internet Project surveyvi on a sample of Italian users, we found
that - when asked what influences their choice of information sources on the Internet - 36% of them
answered that they choose information sources based on their reputation; 34% use the same source
they use offline; 31% is influenced by word-of-mouth. These data confirm the importance of both
cognitive and social hierarchies (delegations) in managing complexity on information-overloaded
digital networks.
Relationship changes in digital economy
Even though we have found support for the idea that there are information, power and organization
hierarchies in digital markets, we don't think this should be interpreted as the proof that the Internet
didn't disrupt the traditional way of doing business. Digital marketing is not a reverse marketing,
but it is neither traditional marketing.
Porter's contribution (2001) to the debate helps give up simplistic ideas about the irrelevance of
differentiation in the new network economy, and re-focus on the importance of making the right
strategic choices when facing the relevant economic trade-offs. But these trade-offs are different
than before. We need to understand how we can create this uniqueness in the new economic and

15. 15
technological context. What is dangerous from a theoretical and practical standpoint is the risk that
we fall in a sterile contraposition between two equally useless views of the Internet economy: "the
frictionless idea" and the "business as usual" mind-set.
We also find that Shapiro & Varian gave an invaluable contribution to the study of the information
economy rules, but they perhaps focused too much on the information side of the digital economy
and overlooked the other side of the economics of mediation: the relationship side. The network
economy is an infomediation/content economy but also a relationship economy. There are new
relationship costs and benefits to consider, and new trade-offs to evaluate. A relationship economy
perspective could help explain for example why network externalities do not always operate in large
communities. Network externalities don't depend on the size of the network (number of members
functionally connected) but on the size of the participating members (Mandelli, 2001, Grewal et al.,
2001). We claim that information rules are not enough to guide action in the digital economy,
because they stress the importance of forced relationships (lock-ins) in order to build economies of
scope, without considering the importance of active and socially rich participation, in order to
leverage cooperation economies. Also, Information rules don't focus the attention on variables,
which we believe is critical in a network economy: social exchanges, with their costs and their
benefits. We call for an integration of the study of the content-side of the network economy with the
study of its relationship-side, in the complexity of a all-connectable (not necessarily all-connected)
world.
The study of the role of brands in the digital economy can help highlight some of the important
characteristics of this relationship economy. Brands in traditional marketing is the managerial tool
for building differentiation in the relationships with the customers, in order to justify price
premiums and build attitudinal and behavioral loyalty (Busacca and Castaldo, 1996; Rust et al.,
2000; Busacca, 2000; Chaudhuri and Holbrook, 2001; Costabile, 2001), in exchange for their role
as cognitive selectors (shortcuts) and uncertainty reducers (trust). This "management of complexity"
role of brands is built around their ability to offer efficient relevant information during the customer
search activity, and constant value (functional and affective) in the relationship experience. The
result is a superior economic return for the firm. Brand trust and brand affect jointly determine
purchase loyalty and attitudinal loyalty. Purchase loyalty leads to greater market share, and
attitudinal loyalty leads to price premium for the brand. (Chaudhuri and Holbrook, 2001).
On the Internet this value-appropriation role of brands is built one-to-one, in terms of micro-
segmentation, versioning and dynamic pricing based on the individual value equations of the
customers. But value-appropriation cannot forget the value-creation side of relationships. Price
discrimination based on the one-to-one perceived value of the same product by different consumers
tries to exploit the dynamic adjustment power of the economic web, but it falls short of the most
important asset: legitimacy. Value is not an absolute matter. It is dynamic and relational (it is
adaptive). And when we speak about trust and brands in digital economy we don't refer only to the
idea that trust and brands can build long-lasting loyal relationships (Reicheld and Schefter, 2000;
Urban, Sultan and Qualls, 2001; Shankar, Smith and Rangaswamy, 2002). We are speaking about
the different nature of delegation and selection of variation that works online, compared to the
traditional one, and we speak about the legitimacy of this selection following the neo-institutionalist
approach (DiMaggio and Powell, 1983). We can consider the so-called "case of dynamic pricing" at
Amazon.com. This case is widely used in public relations studies, because it is a good example of
how companies on the Internet can easily fall in viral communication crises. We are, instead, here
more interested in the reason why this crisis exploded.
Amazon was found to be selling the same DVD movies for different prices to different customers. It
was a test of the dynamic pricing philosophy (which states that customers should be charged
according to their perceived value of the products and their means), built on the information about
the buying habits of 23 million consumers. But customers evaluate and talk. One of them realized
that if he excluded the cookie software on his computer, which identified him as a regular Amazon
customer, the price of the DVD of Julie Taymor's "Titus" fell from $24.49 to $22.74. Chatting on

16. 16
the web site DVDTalk.com, he diffused the news. The reaction of other Amazon customers, also in
other newsgroups, was powerful. They were particularly distressed by the idea that Amazon
charged higher prices to loyal customers (The Washington Post, September, 27, 2000).
It is out of doubt that Amazon has strong brand power (power symmetry not just information
symmetry) in the relationship with its customers, but this was not enough for letting them exploit
the economic leverage of dynamic price discrimination. Customers, who accepted to pay a higher
price when they decided to delegate variation reduction to the brand company in exchange for better
service and future reassurance, refused to pay a higher price when they perceived that this price
discrimination was not legitimated.
Research indicates that capabilities and resources create competitive advantage when they are
valuable, rare, and imperfectly imitable (Barney, 1986). This is the case of brands also in digital
markets, but in these more complex and connected environments this competitive leverage must be
nurtured with more intelligence, collaboration and transparency. Brands must be more intelligent
(because through the learning network with business partners and consumers they can enrich their
cognitive control of the market), and collaborative (because the values associated to the brand must
be socially negotiated and mediated by the emergent and self-organized meaning that the periphery
assigns to them). Collaborative marketing (Mandelli, 1998) doesn't mean that brands are not
hierarchies anymore. It means that brand hierarchies are different: more legitimated and at the same
time more able to cope with complexity.
But brands and trust are not cost-free. If we analyse this in the framework of the economics of
mediation, we understand that brands still are valuable intangible resources for firms that they can
invest in cognitive economic exchanges. Brands still play the role of signaling trustability, driving
customer selection and customer loyalty, which activate the economies of scope at the node level
(one-to-one), and therefore generating value for the brand companies. But for their formation they
require investments of tangible and intangible capital. Brands are built investing in product quality,
quality of communication and quality of customer experience. They are sunk costs for firms.
Also for consumers, brands drive both benefits and costs, both value capture and value invested. For
consumers they work as trust-based shortcuts (benefits connected to the reduction of complexity
and transaction costs for search and evaluation of different options), value associations (benefits
connected to the symbolic area of consumption), but also information and power hierarchy
(reduction and sacrifice of diversity).
So there still are brand trade-offs, both for firms and for customers. But these trade-offs are
evolutionary and negotiated at a more complex level than before. They are a matter of policy
decisions, and policy decisions as we know (Lowi, 1964) are complex outcomes of planned and
everyday negotiations in local, culturally and socially embedded policy arenas.
Internet economy cannot be frictionless, because it' s a complex economy and a complex economy
lives of delegation (even though not always vertical delegation), which is a form of hierarchy. But
hierarchies are not all based on domination and coercion. Selection in a simple and authoritarian
environment may be a matter of low transparency or trapping strategies; selection in a complex and
connected environment is a matter of delegation and cooperation, and this is true also for brand
hierarchies (figure 3).

17. 17
LEGITIMACY
Low High
Low Free
navigation
EFFICIENCY
Walled-garden
portals Professional
media/
Traditional Collaborative shopping bots
branding and intelligent
High branding
Figure 3 Efficiency and legitimacy in Internet infomediation: the role of collaborative branding
Source: elaboration from Mandelli, 2001
So the attention moves to the sources, dynamics and results of social delegation and cooperation.
Trust, Delegation and bounded sociability
Also independently from the new economic governance debate, academics have called for more
systematic research into the role of trust in business relations, observing that: "It is clear that
research on trust needs to advance beyond a catch-all residual in the unexplained random error"
(Koza & Lewin, 1998,85). But this is not an easy task, since there is not clarity around the concept
of trust itself. Regarding this gap, Mutti (1987: 224, cited in Castaldo, 2002, 2) writes that "the
number of meanings attributed to the idea of trust in social analysis is disconcerting. Certainly this
deplorable state of things is the product of a general theoretical negligence. It is almost as if, due to
some strange self-reflecting mechanism, social science has ended up losing its own trust in the
possibility of considering trust in a significant way".
The notion of trust is widely used in different social science disciplines (sociology, economics,
strategic management, organization, marketing, psychology) but often with different meanings.
Castaldo (2002) has recently meta-researched the trust literature, and concluded that there is a lack
of a clear and shared definition of the trust concept, even within the business research community.
He did a meta-analysis of the different (70) definitions of trust, statistically clustering them, and
proposing a multilevel construct-frame. He considered the construct profile (belief, attitude,
willingness), the subjects involved in the relation (person, firms, institution), their profile (honest,
competent, benevolent, committed), the role of risk, opportunism and vulnerability, and the trusting
behaviors. Even though not all the studies hypothesized a direct causal relationship among these
different variables, many placed them in a logical sequence: "This sequence often regards trust as
the expectation, belief (and so on) that a subject with specific characteristics (honesty, benevolence,
competencies, and so on) will perform actions designed to produce positive results in the future for
the trustor, in situations of consistent perceived risk." (Castaldo, 2002,8) So trust is mostly
conceptualized as the heuristics (beliefs that drives action) that help rationally bounded agents take
risky decisions. This heuristics in the definitions examined is more or less based on self-interest
(contrasted to altruistic motivations) and based either on the evaluation of the personality
characteristics of the trustee or on their rational motivation to act positively for the trustor. But

18. 18
basically almost all the authors consider trust as a cognitive tool or state, which we use to guide
action in "the shadow of the future" (Axelrod, 1984).vii
Castaldo (2002)'s content analysis found out that the words most used in the different definitions
were: "action, will, expectation, belief". Very interestingly, cooperation was one of the least used
terms.
These definitions have in our research framework important limitations:
• They do not analyse the difference between bonding and bridging trust (trust in friends and
trust in strangers);
• They define trust as a state, a static phenomenon;
• They don't consider the economics of trust, (there is no account for the costs of the invested
resources in the process);
• They are only focused on the rational-choice dimension of trust and don' t consider the
unthinking, emergent evolutionary component of trust dynamics.
This is why Castelfranchi and Falcone' s (1998 and 1999) model of trust dynamics is helpful. We
are particularly interested in
• Their idea of trust as a dynamic process (trust builds trust),
• Their conceptualization of trust as cooperation (trust requires the investment of social
resources),
• Their idea of trust as the antecedent of dynamic delegation (dynamic reduction of
autonomy)
• Their conceptualizations of trust as complex mediation (external trust builds internal trust).
"Delegation necessarily is an action, a result of a decision, and it too creates and is a (social)
relation among x, y, and z. There may be trust without delegation: - either the level of trust is not
sufficient to delegate, or
- the level of trust would be sufficient but there are other reasons preventing delegation (for
example prohibitions). So, trust is normally necessary for delegation, but is not sufficient:
delegation requires a richer decision. There may be delegation without trust: these are exceptional
cases in which either the delegating agent is not free (coercive delegation) or he has no information
and no alternative to delegating, so that he must just make a trial (blind delegation)." (Castelfranchi
and Falcone, 1998, 3)
So trust drives cooperation but this doesn't necessarily mean that trust in relationships lowers
hierarchies.
Castelfranchi identifies two fundamental problems (Castelfranchi, 1990), studying social interaction
in multi-agent systems:
• The Sociality Problem: “why should autonomous agents enter into social interactions?”
(Castelfranchi, 1990: page 49)
• The Adoption Problem: “how can an agent get his problem to become social, i.e., get it adopted by
other agents?” (Castelfranchi, 1990: page 49).
Castelfranchi (1990) suggests that "dependence" is the informal answer to the sociality problem and
power is the answer to the adoption problem. A ‘lack of power’ concerning their own goals (their
inability to achieve them by themselves alone, which make other agents have "power over" them)
make agents dependent on other agents; these agents' ‘power to influence’ someone else leads to
goal adoption.
According to this conceptualization, trust is not the opposite of hierarchy because trust is a form of
hierarchy (cognitive selection). All cognitive hierarchies are at the same time selections (delegation,
reduction of variation) and increase in diversity and freedom (symbolic value added by cognitive
mediations). Mediation and delegation are not the same concepts, even though they are connected.
Mediation, in our way of using this concept, focuses on the constructivist value added by symbolic
interaction in cognitive and social encounters (cognitive and social associations). Delegation
focuses on selection and complexity reduction.

19. 19
Hierarchies are not all equal, since they are based on different forms of legitimacy of the mediation
structures. (Weber, 1919) The concept of social delegation to media and legitimated gatekeepers in
mass communication research can fruitfully be associated to the concept of external trust in
environments or third parties, who mediate the dynamics of delegation in Castelfranchi and
Falcone, 1999 (figure 4).
Mental states Action
Core trust
Internal core trust Delegation
Evaluation of y
before relying on it.
Beliefs about y’s competence,
self-confidence, motives for
cooperating.
willingness of
cooperating, and persistence.
External core trust
trust in the environment
Reliance
The decision of
relying on y
Figure 4 Internal and external sources of trust
Source: elaboration from Castelfranchi and Falcone, 1999
Trust dynamics
Castelfranchi and Falcone (1999) propose a model of social engagement and cooperation, which
include different views of goals and values driving collaboration. They may be instrumental and not
instrumental. This model describes a circular link between trust (the beliefs - the "state of mind"),
cooperative actions (engagement) and social effects (consequences). Trust builds trust and "... has a
history". Also, trust is not only based on our beliefs about people and their willingness/abilities to
cooperate in the present and the future (internal trust), but also is based on beliefs about the
trustworthiness of the environment and third parties (external trust).
Relational quality and trust formation are processes, guided by cooperation experience and
available knowledge. "The interpretation of these experiences is a complex, multi-dimensional
problem that is a function of the number, frequency, and gravity of their interactions; the difference
between actual and expected outcomes; the nature of any transgressions; the intentions or
motivations attributed to a partner's behavior; and any advance warning and/or post facto
explanation of its actions by the partner. Furthermore, they may affect the partners' willingness to
rely on each other in terms of any one of the three elements that influence performance--
organizational capacity, technical prowess, or integrity." (Arino et al., 2001).
Trust comes out from relationships and it is built dynamically in the process. "Trust builds trust", as
Castelfranchi puts it. Trust drives delegation but this is a dynamic delegation. If we focus only on
trust as the belief about future actions of the trustee, we tend to study the characteristics of the
trustee (trustability and reputation); we don't focus on what is relevant at the trustor level. And it is
not enough to analyse the characteristics of the trustor's personality (the psycological and cultural
variables which influence the tendence to trust somebody). We also need to focus on the process
that builds this resource.

20. 20
In an early study on cooperation in distant-education settings (Mandelli, 1995) we found that
individual cooperation on the net, when not supported by previous knowledge and personal trust,
tends to be fragile. The same result was found in a study by Jarvenpaa and Leidner (1998) on
global virtual teams whose members were separated by location and culture. They concluded that
"... global virtual teams may experience a form of ‘swift’ trust but such trust appears to be very
fragile and temporal." (p. 1)
Also in the organization literature social capital has been found as a pre-condition for the
development of new alliances and greater trust. Networks are seen as repositories "... of information
on availability, competencies and reliability of prospective partners. ... The more the emerging
networks internalize information about potential partners, the more organizations resort to that
network for cues on their future alliance decisions, which are thus more likely to be embedded in
the emerging network. These new alliances, in turn, further increase the informational value of the
network, enhancing its effect on subsequent alliance formation" (Gulati and Gargiulo, 1999, 1440-
1441).
Trust is not a one-shot state of the relationship. It is a dynamical and iterative process: "... nothing is
set in stone. All collaborations start with a set of givens between the partners such as who they are,
what has been said about them, and what their prior experiences have been with each other. While
initial conditions are inherited, it would be wrong for management to assume that perceptions based
on these givens are immutable. From the moment negotiations commence through the start-up and
into the operation of the alliance, the relationship is a living entity subject to considerable growth
and evolution, one that will be shaped by the partners' behavior." (Arino et al., 2001)
Bounded sociability and Internet social capital
In order to test the idea that trust is not a costless resource, available in a frictionless social
environment, we have included Castelfranchi' s concept of trust in the model we used to study the
impact of the Internet on sociability (Mandelli, 2002). We report the most important findings from
our study and elaborate on it, because they are relevant in order to help answer our research
question about the role of trust in building social hierarchies.
We tested the idea that "trust builds trust", and were also interested in the role of cognitive
resources (cognitive energy and time) in this dynamics. Our model rests on the awareness that
cognitive and social resources are finite and their economies influence social investments and
cooperation. Norris (2000b) suggests that social cooperation is influenced by the availability of
cognitive resources (cognitive abilities and time) and by the perception of self-efficacy in social
engagement.
In organizational settings Adler and Kwon (2002) stated that social capital is influenced by
opportunity, motivation and ability, and that structural connections (access) might be not enough for
building social ties; they also require the availability of the relationship and the cognitive
dimensions of social capital. This framework, applied to individual level phenomena, suggests
considering the role of cultural and cognitive resources in the process. Motivation, social capital
and cognitive resources have also proved important predictors of cooperation in virtual
communities of individuals (Kollock and Smith, 1998).
This issue has long been studied by researchers interested in the effects of these new social
aggregations on social engagement (Putnam, 2000; Norris, 2002). As research results (Norris, 2000)
point out, there is a reinforcement effect in Internet sociability; usually actors who become more
active on the web are the most active even offline. The others may be left out of the process. The
reasons for this discouraging effect may be found (Norris, 2000) in:
• Lack of commitment by the individuals;
• Lack of individual cognitive resources (cognitive sophistication);
• Lack of individual social resources (social capital).

21. 21
The economics of cognitive mediation suggest to pay attention not only to the net balance of the
perceived benefits and costs in the cognitive exchanges (Simon, 1972; Neumann, 1991), but also to
the stock of available cognitive and social resources. If they are not available, the individual actors
or the policy-makers of the networks (the managers if we are in a business setting) must invest in
building them.
Applying this framework to individual relationships on the Internet (see figure 5), we hypothesized
that in the digitally all-connected world, access to interactive communication and virtual
community environments can foster the creation of social capital or destroy it, depending on the
original set of cognitive, social and cultural resources available (for the operationalization of all the
variables see Mandelli, 2002).
Sociologists interested in Internet socialization processes helped highlight another important aspect
of virtual social networking: the difference between bridging and bonding social ties. “Bridging
social capital refers to social networks that bring together people of different sorts, and bonding
social capital brings together people of a similar sort. This is an important distinction because the
externalities of groups that are bridging are likely to be positive, while networks that are bonding
(limited within particular social niches) are at greater risk of producing externalities that are
negative.” (Putnam, 2000)
Also Fukuyama (1995), applying the same concept to organizational coordination, claims that only
trust toward the world outside the family fosters innovation and development, while the familistic
culture hinders the ability to extend the richness and diversity of one' s social networks. In a world
where innovation and learning make the competitive difference (Vicari, 1991; Nonaka and
Takeuchi, 1995; Rullani and Vicari, 1999), social networks of independent firms are seen as the
fittest coordination model (McEvily and Zaheer, 1999).
An interesting research question regards the impact of the diffusion of network-based technologies
on these trends. At the individual level, there is the risk that virtual communities lead more toward
bonding sociability, and impede bridging possibilities, because they connect people with the same
interests and lifestyles (Stolle, 1998; Preece, 1999). The Net may consume trust, rather than produce
it, according to Uslaner (2000). In fact the Internet make people connect more with family
members, friends and people with the same interests, than strangers (Cole, 2001). This is consistent
with the so-called balkanization hypothesis, developed by Van Alstyne and Brynjolfsson (1997),
who suggested to " ... examine critically the claim that a global village is the inexorable result of
increased connectivity." Their conclusion comes from the analysis of the constraints posed by limits
of what is known as "bounded rationality" (Simon 1959). "... even in a lifetime, few people have
significant relationships with more than a few thousand others. As long as human information
processing capabilities are bounded, electronic media are unlikely to dramatically change this
total.(Van Alstyne and Brynjolfsson 1997, 3-4).
Contrary to the simple equation between trust and networks, the balkanization hypothesis envisions
a networked world where bonding ties (ties with similar people) will hamper cooperation and in
which variety and innovation at the system level will be lost. We also included this hypotesized
link in our model (figure 5).
In short we tested whether values, individual happiness, cognitive resources, time, trust in people
online, trust in Internet communication environments, and access to people's social network through
the Internet, influence people' s use of the communication cooperative applications on the Internet
(email, chats and virtual communities). Then we tested if all these variables, and bonding offline
social capital (relationships with family and friends) influenced changes in the number of regular
relationships on the Internet.

22. 22
Values Positive
& and
culture negative
Internal trust, Use of changes
External trust interpersonal in
Self-confidence and offline
Individual collaborative and
wellness Diffusion of environments online
Internet access on the Internet social
in people’s
capital
social network
Education, technology
literacy and time
Offline bonding
social capital
Figure 5 A model of trust and social capital dynamics on the Internet
Source: Mandelli, 2002
The data were elaborated using multivariate regression statistical analysis, organized in a path-
analysis sequence. The results confirmed the model hypothesized, but for the direct influence of
time constraint on sociability behaviorviii. The following figure (figure 6) represents the synthesis of
the regression paths, which describe the most relevant regression relationships found (see the
methodological appendix in Mandelli, 2002 for the statistical coefficients and levels of
significance).
Using the
Internet
Values Technological Time used increased
literacy to email regular social
contacts
Internet
self-confidence -
-
Number of Time used
Using the
Made Internet
friends to chat
new increased -
- friends Family contacts with
External online bonding family & friends
trust
Internal ties
Lack of
Alienation trust
- - Internet
Younger access of
age family & friends
Using the
- Internet decreased
Education Southern contacts
local with family
= low significance culture & friends

23. 23
Figure 6 Empirical results on trust and social capital dynamics on the Internet
Source: Mandelli, 2002
The use of the interpersonal communication environments of the Internet (virtual communities and
chats) can drive the creation of both bonding (within the community) and bridging (out-of-the
community) social capital. Using email and group communication people can foster relationships
with family and friends -- but also with people they have never met before (new friends online).
Also the path-dependent nature of the social capital building process is confirmed. There is a
reinforcing link between offline and online social capital. People who have stronger ties with family
members and friends are more likely to increase their regular contacts with them after they start
using the Internet. But they do it, of course, if they can access them through email and other
interpersonal communication tools of the web. The digital access divide may drive a social divide.
Using chatrooms and virtual communities, people can make new friends. But not everybody uses
these collaborative applications. The use of these communication formats (and the consequent
effect on regular relationships) is mediated by age and internal trust (trust in the people they are
going to interact with), that is people who think that is easier to meet other people online than in
face-to-face are more likely to use the synchronous communication environments of the Internet
(chat) and make friends online. Results also confirmed the link between internal and external trust.
External trust (here operationalized as trust in the Internet community environments) helps create
trust in online people.
Even though time stress showed no relationship with sociability outcomes, the data supported the
hypothesis that the availability of cognitive and intangible resources influences cooperation
behavior and sociability phenomena. There was a significant relation between the increase in the
number of regular contacts with people with the same hobbies, the same religious interests and the
same political interests, and the decrease in the number of regular contacts with other social groups.
This, which we call a "switching bonding ties" effect, seems to capture the adaptation to social
changes predicted by the balkanization hypothesis reviewed above, which predicts that the scarcity
of intangible resources and the increased access to people with similar interests drive the reduction
of ties with other social groups.
In short, our study found that the Internet can foster or harm certain categories of social
relationships, and therefore can have both positive and negative effects on the construction of social
capital, depending on specific personal and social pre-conditions Our finding links the risk of
increased society fragmentation not to alienation effects of the Internet, but instead to those
adaptive and evolutionary selection effects necessary for managing the economics of cognitive and
time-based costs of relationships in complex interconnected social networks.
Also culture matters. In our study Italian southern local culture explained different important results
concerned with Internet sociability. Time dedicated to chatting was predicted by southern residence
(also by younger age, lower education, lack of family bonding ties, Internet civic self-confidence
and internal trust). Also the decrease in the number of regular contacts with family and friends was
linked to local residence. This negative sociability result was influenced by lack of Internet
connections of people's family and friends. This is a confirmation of the idea that the first
dimension of digital divide (the differential Internet connectedness of people) has consequences on
the other dimensions of network connection.
In short we found evidence for the idea that:
• Trust is a dynamic process (trust builds trust);
• The formation of trust is constrained by the stock of available cognitive and social
resources;
• The formation of trust is embedded in local culture.
Thus, there seem to be a virtuous circle driving cooperation and the creation of new social capital in
complex social networks -- which describes the adaptive interplay among the stock of social,
technological and cognitive resources with civic engagement, and which may also explain