international journal of rural management€¦ · decisions (braun et al. 2000; edwards and eggers...

http://irm.sagepub.com

Rural Management International Journal of

DOI: 10.1177/097300520500200102 2006; 2; 29 International Journal of Rural Management

Carmelo Cannarella and Valeria Piccioni Development Networks

Dysfunctions and Sub-optimal Behaviours of Rural

http://irm.sagepub.com/cgi/content/abstract/2/1/29 The online version of this article can be found at:

Published by:

http://www.sagepublications.com

can be found at:International Journal of Rural Management Additional services and information for

http://irm.sagepub.com/cgi/alerts Email Alerts:

http://irm.sagepub.com/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://irm.sagepub.com/cgi/content/refs/2/1/29SAGE Journals Online and HighWire Press platforms):

(this article cites 40 articles hosted on the Citations

distribution.© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized

2008 at PENNSYLVANIA STATE UNIV on February 5,http://irm.sagepub.comDownloaded from

http://irm.sagepub.com/cgi/alerts

http://irm.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://irm.sagepub.com/cgi/content/refs/2/1/29


Dysfunctions and Sub-optimal Behaviours 29

DYSFUNCTIONS AND SUB-OPTIMAL BEHAVIOURS

OF RURAL DEVELOPMENT NETWORKS

Carmelo CannarellaValeria Piccioni

This article describes the dynamics and impacts of local territorial networks’on the base of the articulation of some examples of varieties of well knownpathologies in social systems adjusted to the peculiarities of these local net-works. The potential capability of these pathologies to create profound effectsin inhibiting link formation, to turn positive links into ineffective or negativeones and to enhance the non-linear system behaviour and results, deeplyinfluences the quality of interactions among network agents. The possibilityof providing a correct diagnosis of these network pathologies could provideuseful contributions to alerting about actual and potential possibilities oftheir occurring and in preventing a system collapse caused by a deteriorationin the link value and in the eventual link losses.

INTRODUCTION

Literature provides a wide spectrum of studies about the role and effects of theinterrelations between innovation and processes of change (Antonelli 2003; Geels2004; McAdam 2004; Ottosson and Björk 2004). Nonetheless, the solution tothose difficulties, in stimulating in practice, local level innovation and knowledgediffusion, particularly in rural area affected by stagnation and rigidities, stillremains a problematic task. Even non-marginal rural areas can show static con-ditions and resistances to innovation with direct consequences in terms of agricul-tural and rural economy decline, unemployment, natural resources degradationand erosion of quality of life for local communities. These issues derive fromthe need to concretely empower rural areas, replacing the idea of rural areas aspassive recipients of regulations, information, subsidies and other inputs andtransforming local agents (farmers, rural entrepreneurs, local administrations,associations, etc.) into active subjects within innovative processes designed toidentify renewed economic opportunities on environmentally and sociallysustainable bases. Some basilar operative questions ignite the whole discussion:

International Journal of Rural Management, 2(1), 2006

Sage Publications � New Delhi/Thousand Oaks/LondonDOI: 10.1177/097300520500200102




30 CARMELO CANNARELLA AND VALERIA PICCIONI

why do static rural areas stay static? Why are some rural territories more respon-sive and sensitive to innovation and change than others?

Local rural development is assumed to be composed of complex transforma-tion and adjustment processes based on the utilization and management of actualand potential local human, economic and environmental resources creating area-specific problematic issues (underdevelopment, sustainable or unsustainabledevelopment). In this perspective, local development is likely to be a multi-dimensional process influenced by the variable action of some key dimensionsincluding:

1. Economic factors—firms’ structure, firms’ productivity, composition ofeconomic activities, firms’ competitiveness degree at local, regional, na-tional and international levels, etc.;

2. Technological factors—technology penetration in local products and pro-duction processes and management, etc.;

3. Geographical factors—physical and environmental conditions, geo-economical advantages and disadvantages, etc.;

4. Infrastructural factors—infrastructural quality and quantity (roads, rail-ways, transportation and storage facilities, energy, IT, etc.);

5. Cultural factors—educational levels, mentalities, local histories, social rela-tions, social exclusion and cohesion, etc.;

6. Institutional factors—quality of local administration and public services,presence or proximity to knowledge generators (universities, research cen-tres, etc.) and extension agencies.

On the basis of these territorial patterns, it could be stated that underdevel-opment or unsustainable development can result from inadequacies in one ormore of these dimensions: when one of this dimension tends to fail, it is neces-sary to intervene in the other spheres in order to compensate for the collapsedone and severe development crises can occur if all these dimensions collapsesimultaneously.

Institutional factors play a critical role because the institutional setting, andthe related concepts of institutional strength and institutional thickness, can actas anchor and dynamo stimulating collaborations and synergies to attain commongoals to solve development inadequacies. Local institutions play the anchor roleif they are able to tie up the (key) agents in the area by creating versatile andintensive collaboration relationships among them and play the generator role iftheir activities generate replies to development inadequacies and inefficiencies.In this perspective, the capability of a rural area to act as a ‘network of networks’(at economic, social and institutional levels), operating within an environmentaland cultural context becomes the key factor to support an effective rural develop-ment strategy to cope with stagnation and rigidities. Nonetheless, the creation,development, enforcement and management of such territorial networks, require





local agents to have prompt operative responses based on more intensive manage-ment and better global skills but many territorial agents (firms, farms, localadministrations, associations, individuals and other organizations) find mount-ing difficulties in taking on any extra burden (Arthur 1990; Beer et al. 2003;Curran 1994; Reese 1997). The creation and strengthening of local rural net-works cannot be ‘imposed by law’, rather requires some prerequisites linked tofour basic intrinsic capabilities:

1. Individuation—capability of the network sponsors and supporters to iden-tify relevant agents and appropriate resources to be involved within thesystem;

2. Involvement—agents’ capabilities to be engaged in the network and to en-gage further agents;

3. Integration—agents’ capabilities to be functionally and effectively struc-tured within the network;

4. Cooperation—agents’ capabilities to work together and concretely act withinthe network.

Unfortunately, these capabilities are not always immediately evident in an inertrural area: more often than not they are rather tacit and latent requiring someform of ignition, mainly linked to knowledge and innovation circulation to de-velop appropriate and adequate skills as a precondition to making these capabil-ities more explicit (Leeuwis and Van den Ban 2004; Rogers 1979; Romney 1989).The creation of pilot territorial systems within which information and knowledgecan flow, be shared, put in action and become more productive is likely to be anessential prerequisite to transforming Knowledge Centres (Small and MediumEnterprises [SMEs], research centres, local development institutions, innovationsupporting agencies, individuals, associations, etc.) into Knowledge Networksthrough the improvement of the interconnections (edges of the system) amongthese different local nodes (vertices of the system) of expertise.

When these local networks work properly, facilitating knowledge circulationamong agents and inducing imitative aggregation phenomena, even simpleinitiatives promoted with limited financial resources are likely to boost innovationand growth, producing deep impacts in the area influencing methods, practices,techniques and, above all, mentalities. Empirical evidence often confirms that amassive presence of poorly coordinated public intervention subjects, duplicationsand the creation of a cloud of local offices, agencies and task forces, the lack ofor inadequate cooperation among local agents can create mismanaged knowledgeand information overflows wasting huge amounts of investments and frequentlyfailing in stimulating local development, innovation-spreading and the solu-tion of inertia and static economic and social conditions (Cannarella and Piccioni2005a).





Territorial networks are here defined as stable and long term formal/informalsystematic relation collaborative webs, more concerned with ‘local public goods’,among differentiated agents sharing a common information and knowledgesource on the base of a more or less conscious agreement. Local public goodsdiffer from other public goods for the limited geographical reach of their bene-fits conveyed, being a class of public goods connected to regional and nationalgoods whose production requires a cross-area collective action (that can engageneighbouring territories also from different administrative regions) for the presenceof cross-area problems and cross-area externalities (Bloise et al. 2002; Champsaur,Roberts and Rosenthal 1975; Milleron 1972; Rege 2004; Thomson 1999).

Territorial networks can contribute to achieving and managing these localgoods influencing local business development and capacity building actions.In the first case, local networks can become strategic tools to support the develop-ment and growth of the involved firms/farms and to stimulate a generation ofnew enterprises around them (production districts). In the second case, they canfuel learning processes, enhancing the skills and expertise of the agents involvedin the network, disseminating knowledge and innovation and improving theagents’ capabilities to properly select options and choices and make well-informeddecisions (Braun et al. 2000; Edwards and Eggers 2004; Hall et al. 2001).

In the real world, local territorial networks can often show non-linear behav-iours, they could not operate as expected and their results can highly differ fromthe initial goals for the occurrence of perturbations, pathologies and internal/external pressures. The aim of this article is to identify some theoretical founda-tions about these abnormalities useful to (a) confirm the connections betweeninefficient local rural development and dysfunctional territorial networks;(b) develop realistic collaboration networks; (c) strengthen these networks andevaluate their effective impacts and (d ) evaluate and improve the integrationcapabilities of some focal subjects in the network itself. The identification ofsome indicators about the more frequent network pathologies can provide con-crete contributions to improve the operational dynamics of these networks andthe related strategies and tools.

The present article is based on a survey carried out within the research activitiesof the project ‘Development Dynamics and Increases in Competitiveness ofRural Areas’ (DICRA) resulting from an agreement between the Research Teamon Development and Innovative Processes at the Institute of Chemical Methods(IMC) of the National Research Council of Italy (CNR) and the Municipalityof Vitorchiano (a 4000-inhabitants village in the province of Viterbo, about100 km north of Rome in central Italy). The rural area selected for the project ischaracterized by traditional agricultural activities, small-scale industries and miningactivities being exposed to consistent migration flows from urban centres(Viterbo, 55,000 inhab. distance: 7 km–Rome 4m. inhab. distance: 100 km). Inthe case study rural area, an analysis of the context highlighted a global weakness





of the agents to recognize and receive benefits and opportunities from innovationdue to the action of some structural factors such as:

1. Demographic factors—high incidence of elderly population and rapidgrowing ageing of farmers;

2. Job structure—scarce presence of young farmers;3. Productive transition—economic and social decline of agriculture;4. Residential flows—deurbanization and resettlement of rural areas for resi-

dential use.

The widespread presence of older farmers and a shared inclination among localagents to act separately are intuitively a source of serious pressures towardsroutines, static and repetitive conditions which represent an actual and potentialcause of opposition to innovation whose role is particularly important in thelight of an increasingly competitive and integrated international economic envir-onment. For the purposes of this article, we have investigated about the presencein the area of territorial networks: we have extrapolated five examples of localsystems, identified according to the previously cited definition, from a largenumber of formal/informal networks (or declared as such) deriving from formalprogrammes, agreements and synergies among public entities (at local, regional,national and EU levels), private partners (SMEs, producers’ associations andunions, research centres, etc.) directed to stimulate local development and innov-ation diffusion. For the present survey, short questionnaires and brief interviewswere adopted but frequent informal talks with local people represented theprivileged tool to obtain updated independent information about these topics.Of course this approach is likely to be affected by a certain degree of subjectivitybut in the identification and organization of those factors at the base of a network’sbehaviour, individual and collective ‘prejudices’ can play a not secondary rolecompared to the more conventional aspects of the issue. Given the characteristicsof this survey, a rigorously orthodox statistical approach to the resulting infor-mation was not adopted: rather, using the data to articulate some pragmatic andempirical observations.

LOCAL TERRITORIAL NETWORKS: A CONCEPTUAL FRAMEWORK

The creation, identification and development of a dynamic cohesive relationnetwork among agents in a rural area depend on the action of a large number ofvariables and local, specific peculiarities which characterize, for example, theagents involved, the territory where they operate, their interrelations, the outputresulting from their interrelations, the expectation/frustration balance, imitationand external contacts, trust/mistrust balances, etc. (Holt 2002; Lundstedt andMoss 1989; Von Zedtwitz et al. 2003). Useful interpretative keys can derive





from modeling these networks describing critical phenomena occurring in theircreation and development and in improving their performances when acting asengines capable of stimulating and boosting local development.

Literature provides a wide range of theoretical models for the architectureand properties of these networks (Albert and Baràbasi 2002; Barrat and Weigt2000; Bollobàs 1998, 2001; Bouchaud and Potters 2000; Goh et al. 2002; Kickert1997; Mantegna and Stanley 2000; Newman 2003). Even though these modelshave been developed for the analysis of phenomena distinctive of other discip-lines, they however show interesting analogies (and useful interpretative keys)with the present study. In their simplest form, these systems can be defined as acollection of points or vertices (N) which are connected by a variety of lines oredges. Vertices may represent different types of ‘locations’ and in their applica-tions vertices and edges can outline various types of business relations, infra-structural grids (that is, transport networks or internet web) and human and socialinteractions. In literature, many studies have been conducted about these systemsusing the graph theory, which poses many mathematical problems, but includesmany relevant considerations about the evaluation of these networks’ properties:for instance, defining whether the system is connected or disconnected is oneof the most important properties to look at (Figure 1) because the definition ofan analytical model to describe and evaluate the connectivity status of the sys-tem is essential to understanding whether all the components of the system areadequately involved in the processes, thus determining the diffusion degree,through the edges, of eventual impacts and results among vertices (Barabàsi 2002).

Figure 1A Connected 6-node, 6-edge Network





Other crucial interpretative keys can arise while evaluating local vertex properties.For example, the number (k) of edges attached to a vertex (degree of a vertex) orthe description of variety of degrees in the network summarized in the degreedistribution (K ). This property is extremely important for the description anddevelopment of dominant hubs within the network with a large number ofpaths passing through these hubs (Vertex V2 in Figure 1). These dynamics areessentially expressed by ‘power laws’ which could be synthetically biased in the‘Pareto law’: these systems, in fact, are not formed by purely random interrelationsbut show some ‘preferential attachment’ degrees according to which new edgestend to attach preferentially to vertices with large degree k. Simulations confirmedthat power laws come from networks where there are hubs or vertices withmore edges attached than the majority of vertices: to a large extent, the processesleading to the formation of stable connected networks require some type of‘preferential attachment’ (Jeong et al. 2003; Vazquez 2003).

For the present study, the dynamics of some representative sets of models ofnetworks evolving due to aggregation processes and the effects of aggregationof edges in presence of hubs play a critical role also in consideration of the ef-fects, for this analysis, of percolation and Positive Word of Mouth (PWOM)/Negative Word of Mouth (NWOM) phenomena in enhancing this hub’s aggre-gation power and as partial explanation of an initiative’s success/failure as a modelfor imitative actions (Albert and Barabàsi 2000; Bianconi and Barabàsi 2001;Callaway et al. 2000; Dorogovtsev and Mendes 2002; Herr et al. 1991; Marquisand Filiatrault 2002; Proykova and Staufer 2002; Solomon et al. 2000). Anothercrucial interpretative key is provided by the fact that the system behaviour de-pends to a large extent on the strength of its interactions. Local territorial networkagents’ size, structure, mission and scope are inadequate to explain their ‘pro-perties’ while operating within the system. The relationships occurring in thenetwork often drive to some intuitive interpretations: the stronger the inter-actions, the better the agents are held together and the more energy it takes todisrupt the network assemblage. The connection between interaction strengthand agents’ rate of escape from the network can become a useful indicator aboutthe inner system cohesion forces contributing to explain many related socio-economic phenomena. In this way, links’ strength indicates not only the qualityof socio-economic interactions among networked agents but also the ‘wealth’and the volatility of the system environment.

The quantification of the strength in these network links, widely discussedin theory (Albert et al. 2000; Dorogovtsev and Mendes 2000; Jain and Krishna2001; Marsden 1990; Townley et al. 2003), is extremely difficult to detect inempirical analyses requiring reliable data not always operationally available. Also,the implementation of theoretical models in the real world is a particularly diffi-cult task.

For the purposes of this study, a rate λ of existing link decay and a rate γ oftransmutation of strong links into weak connections are introduced throughout





a scale ranging from one to five. These networks are examined through someempirically observable evidences and symptoms identifying three strength levels,described in Table 1.

Table 1Links’ Strength in Local Networks

Condition Features Parameters

Integration

Condensation

Amalgamation

High link decay andtransmutation degree(λ and γ = 5 or 4)

Medium link decay andtransmutation degree(λ and γ = 3)

Low link decay andtransmutation degree(λ and γ = 2 or 1)

Knowledge flows through hierarchicalstandards. Agents utilise limited informationand ideas from the context. Communicationis not an issue. Knowledge flows within verylimited spheres. Low coordination andagents’ attitude in ‘mixing’ their activities isscarce.

Ideas are shared among groups of specialists.Agents are involved in stable discussiongroups. The network shows an attitude toincrease the number of edges among agents.Medium coordination and agents’ magni-tude in ‘mixing’ their activities is substantial.

The network holds melting pot mechanismsin which agents work together showingrelevant coordination and significant magni-tude in ‘mixing’ their activities. Ideas, infor-mation and knowledge freely flow within thenetwork. All the agents (and the subjectswithin the agents) contribute to the gener-ation and improve knowledge. Knowledgeflows transparently and all the agents (andthe subjects within the agents) contribute tothe evaluation of the network’s results andimpacts and optimise adjustment processesrelated to innovative cycles. The networkdevelops an advanced communicationstructure.

The passage from simpler integration levels to more advanced ones can drive tothe creation of complex knowledge networks whose evolution largely dependson the capability of the agents involved to overcome the conventional limits oftechnological transfer (Burt 1987). Technological transfer is in fact mainly basedon the dynamics of two operational sides: an innovation (knowledge) supply(that is, research centres) and an innovation (knowledge) demand (that is, enter-prises). Even the most evolved models (knowledge loops) are linked to a clear





separation between these two sides. On the contrary, within a knowledge net-work, this distinction is likely to be rather fluid. All the agents involved aresimultaneously knowledge producers and beneficiaries: knowledge networksdo not respond only to the need to generate increases in the volume of knowledgebut rather to make agents’ knowledge (explicit/tacit) more productive. Therefore,these networks contribute to putting knowledge into concrete action becauseoverall knowledge is tested upon meeting other agents’ knowledge and expertiseand, thanks to continuous adaptation processes, can be constantly transformedand potentially improved (Barabàsi et al. 2002; Clegg et al. 2005; Greve 2005;Herlocker et al. 2004; Sugawara and Lesser 1998). For the purpose of the presentstudy, networks are classified according to some parameters:

1 Nature

(a) Formal networks

These networks derive from explicit agreements among agents which clearlydefine ‘the rules of the game’, responsibilities, network’s aims and scope, agents’roles, etc. Examples of these networks are those driving from regional rural de-velopment plans, EU scientific collaborations, etc. For this reason formal net-works are likely to be more reliable but rather bureaucratic, having to adhere tonorms and regulations, less flexible and malleable in case of changing conditions.

(b) Informal networks

These networks derive from verbal or tacit agreements among agents and forthis reason do not have a well-structured architecture and require a tacit reiter-ation of these unexpressed ‘successful’ agreements. These networks, not beinglinked to formal agreements, are less bureaucratic, rather flexible (but in theoryweaker than the formal ones) showing on the one hand more volatility but onthe other hand a higher degree adjustment in case of changing situations.

2 Configuration

(a) Hierarchical networks

For the present study, the network hierarchical structure is given by the agents’‘substitution’ degree (Z): the lower this value, the harder it is to replace thisagent, for the system existence, with an already involved agent or new networkentrants: Za1 > Za2 > Za3. This condition does not necessarily coincide withthe role of hubs because hubs are characterized by a large quantity of edges whilean agent’s substitution degree is linked to the quality of its edges. Examples of





agents having a low substitution degree are public institutions providing financialresources for network creation and development and the achievement of relatedgoals. If an agent shows a high Z level, it is less influential and can easily be re-placed and this substitution will scarcely undermine the network’s activitiesand existence.

(b) Peer-to-peer networks

In these networks, all the involved agents show the same substitution degree:Za1 = Za2 = Za3. In this case, the network’s fate is not linked to a dominantagent behaviour. Nonetheless, the network performance and existence can beput at risk by the behaviour of any agent.

Another critical factor for the networks’ classification is linked to the presenceof the earlier mentioned network generators. In these networks, the relationsystem acts as a skeleton, but for their existence and survival, they always requirea certain sort of ‘nourishment’ which circulates within a parallel ‘vascular system’.This parallel system originated from the generator (which doesn’t necessarilyplay the role of hub) and not always coincides with the relation network. Examplesof these type of systems are those created on the basis of public (local, regional,national or EU) funding in which these economic resources are the energy forthe entire network. Hence, different parallel systems can exist within the samenetwork (economic system, information system, coercion system, etc.). On thebasis of these considerations, local networks can show different perspectives,that can be identified as:

(a) monodimensional—when these parallel systems coincide in the samenetwork;

(b) bidimensional—in the case of two not-coinciding parallel systems withinthe same network;

(c) tridimensional or multidimensional—in the case of three or more not-coinciding parallel systems within the same network.

On the basis of these premises, the characteristics of the present study samplenetworks are resumed in Table 2.

Local territorial networks are essentially ‘anthropological networks’ made ofagents (organizations) whose links indicate a relationship which is not just asimple function of physical distance among agents (Degenne 1999; Kilduf 2003;Scott 2000; Wasserman and Faust 1994; Watts et al. 2002). For these graphs,vertices of a network have no position associated with them. Unlike infrastruc-tural webs (transportations, energy grids, water or gas distribution systems), inthe case of random anthropological networks, the connections among verticesare made with a very limited reference to notions of relative position of vertices:the random graph of Figure 1 thus has no structure other than its connections





A B C D E

3 (a

loc

al p

ublic

adm

inis

trat

ion,

a r

esea

rch

grou

p, a

n E

U-r

egio

nal

info

rmat

ion

cent

refo

r ru

ral d

evel

opm

ent)

4 (a

loc

al p

ublic

adm

inis

trat

ion,

a r

esea

rch

grou

p, a

coop

erat

ive,

a pr

oduc

ers’

asso

ciat

ion)

10 (

a lo

cal p

ublic

adm

inis

trat

ion,

a r

esea

rch

grou

p, a

coop

erat

ive,

a pr

oduc

ers’

asso

ciat

ion,

6 oi

l pro

duce

rs)

7 (2

loca

l pub

lic a

dmin

istr

atio

ns, a

reg

iona

lad

min

istr

atio

n, 2

rese

arch

inst

itute

s, a

coop

-er

ativ

e, a

pro

duce

rs’ a

ssoc

iatio

n)

3 (a

res

earc

h in

stitu

te, a

pro

duce

rs’ a

ssoc

ia-

tion,

a r

egio

nal a

dmin

istr

atio

n)

Tab

le 2

Loc

al N

etw

orks

’ Ch

arac

teri

stic

s an

d Fe

atur

es

Num

ber

and

Netw

ork

Type

of A

gent

sM

issio

nN

atur

eC

onfig

urat

ion

Pers

pecti

ve

Bid

imen

sion

al

Mon

odim

ensi

onal

Mon

odim

ensi

onal

Bid

imen

sion

al

Mon

odim

ensi

onal

Cre

atio

n of

an

info

rmat

ion

ser-

vice

abo

ut E

U p

rogr

amm

es fo

rlo

cal r

ural

dev

elop

men

t

Cre

atio

n of

a s

hop

for

the

pro-

mot

ion

of lo

cal r

ural

pro

duct

s

Impr

ovem

ent

and

prom

otio

nof

loca

l oil

prod

uctio

n

Rea

lizat

ion

of p

ilot

plan

t fo

rliv

esto

ck an

d ag

ricu

ltura

l was

tes

trea

tmen

t in

orga

nic

agri

cultu

re

Des

ign

and

impl

emen

tatio

n of

trai

ning

cou

rse

for

farm

ers

Info

rmal

net

wor

k

Info

rmal

net

wor

k

Info

rmal

net

wor

k

Form

al n

etw

ork

Form

al n

etw

ork

Hie

rarc

hica

l

Peer

-to-

peer

Hie

rarc

hica

l

Hie

rarc

hica

l

Hie

rarc

hica

l





and the positions of its vertices in some real space are not an issue. Yet, randomgraphs can provide only a partial description of local territorial networks, because,to some extent, even virtual forms shapes and appearances (rather than spatialfunctions) should be included among the key properties of these networks asindicators of their flexibility and eventual symptoms of the occurring of certainnetwork pathologies.

If we include the factors generated by the previous classification model, theconventional graphic representation for networks is unlikely to be suitable forthe present analysis because local territorial networks cannot be represented bya collection of points connected by a variety of lines with the involved agents’different ‘weight’ (linked for example to their variable ‘substitution’ degrees).

For this reason, we decided to adopt a ‘molecular’ approach in designingsuch systems and links driving to evolve the network presented in Figure 1 insome models of ‘territorial molecules’ as shown in Figure 2.

Figure 2A Network Molecular Visualization





The visualization of these networks through these molecules can better generalizesome typical properties of local systems: for instance, one can say that the distanceof the edge in these molecules, according to the parameters described in Table 1,can sketch out, to some extent, integration, condensation or amalgamation. Inaddition, the molecular shape can outline the network optimal behaviour andthe related modifications in this virtual shape can simplify and exemplify thedescription of some sub-optimal behaviour, called pathologies, causing funda-mental modifications in the characteristics of the system structure (Jensen andLesser 2002).

THEORETICAL FOUNDATIONS OF PATHOLOGIES OF LOCALTERRITORIAL NETWORKS

Several types of system pathologies have been analysed in detail in the literature.Pathological behaviours in specific limited classes of games emerge from theimplementation of game-tree search algorithms (Kaindl 1988; Nau 1983) aswell as many induction algorithms describe overfitting phenomena that havethe tendency to produce models that contain excessive structure (Jensen andCohen 2001; Oates and Jensen 1997). In the present article, we refer to peculiarpathologies occurring in multiagent systems: these peculiar pathologies areadjustments made by local agents (in strategies and behaviours) which are bene-ficial for the single agent but degrade the performance of the overall network.These networks embody continually updated successful tacit/explicit agreements(Boudon 1981) based on formal/informal norms and the regularity of imitativereiterations produced by mutual influences and interactions directed to keepthe networks disruptive/constructive balance under control to grant the develop-ment and survival of the system itself: In more concrete terms, these pathologiestend to violate the explicit/implicit agreements among local agents. The descrip-tion of these pathologies can derive considering the alterations in a networkwhose performance is a multiplicative function of agent performance (Daft 2001;Galbraith 1973). These networks theoretically obey a fundamental assumption(the local improvement assumption) according to which, if the performance ofall other agents remains unchanged, then improvements in local performanceby one or more agents will improve the system performance. In a properlyfunctioning network, changes in quality and quantity of information and decisionmaking processes capable of improving the performance of a single agent, canimprove the performance of the entire system as well. Some sub-optimal networkperformances can also emerge from unintentional agent behaviour or externalinterferences while these pathologies emerge intentionally, violating the localimprovement assumption pushing to a network negative performance due toone or more agents’ behaviour which produces improvements in local vertices’





performances degrading the whole system and damaging the networks’ improve-ment potentials. The most elementary system performance (action) function Acan be expressed as a sum of the agents’ performances (actions) a or:

∑=naA (1)

As previously mentioned, information and knowledge circulation play a criticalrole for these networks, thus requiring, for the description of their dynamics, amore articulated model linked to some basic parameters such as the number ofagents N, the number s of strategies adopted by the agents acting within the sys-tem, the quantity of total information µ available for each agent (this informationis introduced in the system thus acting as system information source) and eachsingle value P of total information µ. Information µ thus results from a certaincombination of a number of P. The information available in a specific time t isexpressed by µ[t] and µ[t] ∈ [0 … P–1]. Each agent performance a is based onthe amount of information available and on the adoption of a specific strategyR: the agent performance is thus expressed by aµR representing a reply to eachsingle value P of the total information µ and assuming the value aµR > 0 (pos-itive action maintaining/improving link connecting nodes), aµR = 0 (inaction)or aµR < 0 (negative action/opposition deteriorating connecting link). Agentshold a set of possible strategies classified on a scale SR on the base of their successin the adoption of their respective single strategies.

A territorial network interacts with the context within which it operates andthe context also expresses more complicated networks of networks. The environ-ment or context E represents a sum of material/non material—tangible/intangibleresources e at a given value p.

∑= ][][][ tptetE (2)

A matrix J can be introduced to define the interaction among the agents and thecontext: J[t]ae can describe the relation of an agent in time t with the resourcese of the context: among these resources, time and financial assets play a criticalrole. The system net action A[t] can be given by:

∑=

=PN

Rae

JtuR tRatNtA

1

][ ][][][ (3)

where nR describes the number of agents adopting the strategy R. Agents willtend to adopt the ‘best’ strategy or the strategy holding a high SR score, useful tooptimize the system performance. The classification of strategies is updated on





the basis of the gratification gR generated by the system according to SR[t+1] =SR[t] + gR[t+1]. The system performance is influenced (and complicated) bythe fact that the agents hold their own specific gratification scales, connected tospecific and personal ideas of ‘success’ and ‘benefit’, which are agent specificand subjective. The dynamics of these gratification subjective microscale deeplyinfluence the possibility to ignite the system and its optimal performance onthe basis of a widespread collaborative trust as well as the possibility to generatesystem-positive net actions (impacts) from which further gratifications can beachieved.

In theory, the aim of this network is to mobilize and increase informationµ[t] to generate positive impact (A[t] > 0), optimizing not only the choices andactions of each single agent but also those of the system as a whole during thecourse of time. The progressive gratification deriving from the system can beexpressed by the function:

[ ]aeAt

R tJtRaftg ][][]1[ ][µ=+ (4)

which implies that for each agent, the progressive gratification deriving fromthe strategy R depends on the actions the agent has adopted on the basis of acertain amount of information µ[t] available and on the system response; suchgratification is thus linked to the system potential in producing, in the agent’svisions, operational gratifying responses.

‘Time’ represents a critical factor in knowledge circulation and informationproduction and consequently in the maturation of the system performancesand responses. The adoption of an agent’s action aµ[t]R is essentially influencedby ‘experience’ or a combination between the results of strategies previouslyadopted within the network (endogenous factors) and other experiences (exo-genous factors). This sort of agent’s ‘history’ is maturing in a time step prior tothe generation of information in time t. The agent’s actions are thus influencedalso by the information available in t – 1 or µ[t – 1] e SR[t – 1] driving to aµ[t-1]Rand a system net action A[t – 1]. Agents do not rationally adopt actions fromperiod t – 1 to t unless a certain number of results of the previous period areknown (experience) with a related gratification/frustration degree. Also, at thesame pace, the involved links tend to become obsolete. This mechanism of re-sults/effects emersion contributes to the satisfaction (wealth) of the individualsi who concretely operate in the organizations (agents) involved in the system.Individual wealth W is a critical factor because these agents are not definitively‘impersonal’ and the inclusion of ‘personal satisfaction’ can contribute much todescribing the causes and reasons of system distance for an optimal performance.This wealth W can be expressed by:

[ ] aeiii tJCtpttW ][][][][ +φ= (5)





where, φ[t] describes a certain number of material/immaterial results at a givenvalue p[t] in the time t, C expresses the amount of ‘richness’ deriving fromthese results functional to the exploitation of the context’s resources related tothe modalities the context resources are used by the agents. Increases in wealthW can be described by:

∆Wi[t + 1,t] = Wi[t + 1] – Wi[t] (6)

thus driving to:∆Wi[t + 1,t] = ∆Ci[t + 1,t] + p[t + 1]

(7)∆φi[t + 1,t] + φi[t] ∆p[t + 1,t] φ[t + 1,t]ae

This equation influences the related gratification mechanism gR [t + 1] for eachstrategy according to the capability of aµ[t]R in generating increases or reductionsin individual wealth. The relation between individual wealth and the adoptedstrategy is expressed by:

∆WR[t + 1,t] = φR[t] A[t] (8)

where, φR[t] indicates the quantity of results obtained adopting R during theperiod t. It is important to note that in the time step t actions aµ[t]R are still inprogress: they are fully completed in t + 1. From this derives:

∑−

=

=1

0

][][t

i

tR at µφ (9)

The relation between gR [t + 1] and DWR[t + 1,t] can be thus expressed by theequation:

⎥⎦

⎤⎢⎣

⎡=+ ∑

−

=

1

0

][ ][]1[t

i

tR tAaftg µ

(10)

Equation (10) describes the possibility for an agent to opt for a strategy virtuallycapable of generating the highest satisfaction score. This virtuality is caused bythe fact that the success in the adoption of this strategy can be evaluated only atthe conclusion of period t. This condition contributes to determining highheterogeneity among the agents in the idea of ‘success’ of a given strategy. Thesystem dynamics are thus influenced by the classification of the agents’ strategiesand the modalities through which individuals and agents concerned classifyand maximize their wealth.





Abnormalities in the dynamics of the above functions can suggest the emer-gence of system malfunctions caused by eventual pathologies occurring withinthe network.

DISCUSSION: TYPES AND DYNAMICS OF TERRITORIALNETWORK PATHOLOGIES

Individualism

In individualism, an agent improves its own performance by adversely affectingthe context for other agents: by making greater use of the context’s resourcesnormally shared by all the network agents. The increased demand provokes adisproportionate reduction in the value of the shared resources for all otheragents, with growing transaction costs associated with the increased use. Thispathology of multiagent systems is well-known in literature which provides anumber of examples of the occurrance of this circumstance termed the ‘Tragedyof the Commons’ (TOC) (Hardin 1968; Glance and Huberman 1994; Turner1993; Wellman 1993).

This pathology affects the network through alterations, caused by subjectivepushes in one or more agents, focused on Wi[t]: here we assume that theindividuals concerned (managers) have the power to influence the organization(agent) behaviour. Combining equations (7) and (8), an individual (or a groupof individuals) operating within the involved agent, increases his own personalwealth Wi[t], receiving increased benefits, advantages and a sort of ‘richness’(∆fi[t]p[t] and ∆Ci) thanks to a greater use of J, pushing E<0 and eventuallyA<0. On the basis of equation (10), sub-optimal individual behaviour reverber-ates to the agent performance because it implies an alteration in an agent’sSR[t + 1] = SR[t] + gR[t + 1] provoked by a perverse gR[t + 1]. The agent willthus tend to adopt actions capable of producing increases in individual advantagesWi[t], implying that an optimal system performance (A > 0) is to some extent‘worth’ substantially less than a sub-optimal or adverse one (A < 0) altering thechoices among the agents’ strategies, choosing the strategy with a high SR onlyin function of some Wi[t] increases rather than the optimization of the systemperformance.

The combination of the results of such strategies and the lack of or inefficien-cies in network disincentives and sanctions, stratify experiences pushing theagent towards the adoption of action aµ[t]R, which rewards the reiteration ofadverse behaviours and definitively alters the agent’s SR.

This pathology has been reported in network E: respondents evoked a com-monly perceived sense of ‘partiality’ in the use and distribution of networkresources’ among other agents. In this case, one agent suffered the effects ofparticular external pressures and used the system resources supporting privileged





agents selected on the basis of personal beneficial returns, however, altering thefunctionality of the whole system and highly modifying the impact quality itpotentially would have: the network has been used as a ‘private property’ todevelop and strengthen different forms of public relations and benefits.

Crystallization

Crystallization is one of the most frequent pathologies in over-bureaucratizedorganizations (Corkill and Lander 1998; Hall 1968; Walton 2005). Formal organ-izational configurations tend in fact to be static due to the bureaucracy’s effectswhich structure staff and functions to achieve the declared goals. Staff and man-agers have to cope with living experiences of concrete problems, adaptive atti-tudes, compromises, etc. driving the agent to achieve a sort of equilibriumbetween these two pushes and resist excessive unbalances: empirical evidencesshow that the presence of a gap is to some extent essential for an agent to beflexible enough to cope with changing situations and a continuing evolvingscenario (Cannarella and Piccioni 2005b). Working rigidly to the rules couldcripple the agent and its operative mechanisms won’t work as expected.

Nonetheless, in certain conditions, some systems reward agents for conform-ing to prevailing standard procedures, introducing rigidities, reducing flexibilityand degrading the system performance. This pathology arises when one agentcan improve his performance operating rigidly and other agents can improvetheir performances by conforming to this agent’s beliefs or operational praxes.When crystallization prevails over the whole system, it works as ‘social pressure’and ‘network externality’, in particular when it becomes an incentive to conform.If the local performance associated with a set of beliefs depends partially on theextent to which they are shared by other agents, then the initial state of allagents can determine which belief system prevails. This can lead to lower systemperformances when rigidities initially characterize one agent’s behaviour whichcan easily cast upon the network and be adopted by all other agents.

Crystallization thus occurs when too rigid a behaviour comes to be commonlyshared by all the network agents and its advantages in the short run derive fromits capability to contribute to avoiding system sunk costs and mitigating transitionproblems towards changed conditions. These short-run benefits can howeverbecome disadvantages in the long run and the minimization of short-term costsmay preclude the maximization of long-term returns. Reliability rather thanvariability is the fundamental of crystallization and it may stimulate inertia pre-venting changes from occurring which may involve loss of stability and outcomestoo far from the neighbourhood of the status quo.

==== ][][3

][2

][1

tN

ttt aaaa µµµµ 0 thus A = 0 (11)





Network D suffered the effects of crystallization due to the presence of a highlybureaucratized agent (with a very low Z value) which imposed its operationalrules on the other agents, consequently ‘freezing’ the network behaviour.

Rotation

In rotation, the system’s agents engage in a series of ultimately circular statetransitions (Etcheson 1989; Gleditsch and Niolstad 1990). These can proceedindefinitely or end in a state with the same performance as the initial state.Rotation usually consumes network resources without producing any relevantcorresponding result. In this survey, two types of network rotation have beenidentified: the former caused by one or more agents’ inaction (having low Zvalue or operating in a peer-to-peer network) and negatively influencing networkdynamics generating ping-ponging conditions based on looping behaviours andpoor system performances also associated with inefficient time consuming andunchanging processes. In this case, the presence of a1

µ[t] = a1µ[t+1] = a1

µ[t+2] =a1

µ[t+3] = 0, heavily influence the degree with which the neighbouring positivelinks can be turned into negative (at an increased rate γ) and eventually decay(with increases in λ). The reiteration of the same scarce commitment in thenetwork reverberates within the whole system causing sub-optimal repetitivenetwork performances (A = 0) and according to equation (10), to a gR[t] =gR[t + 1] = gR[t + 2] = gR[t + 3] = 0 in other agents.

The latter is characterized by a status in which initial information endlesslycycled among agents through a number of network operative meetings, work-shops, etc., never increasing the network knowledge base (Glance and Hogg1995; Hauser et al. 1993). This condition generates looping behaviours basedon revolving information cycles. Being µ[t] = µ[t + 1] = µ[t + 2] = µ[t + 3]…then aµ[t] = aµ[t+1] = aµ[t+2] = aµ[t+3] … According to equation (4), this flatbehaviour will progressively erode the gratification scale, slowing down thesystem dynamics driving to a = 0 and A = 0.

In our survey, Network B was affected by Rotation Type I: the network hasbeen embarked in a series of ineffective repetitive actions due to delaying behav-iour and time consuming in one agent. Rotation Type II was reported in NetworkC whose sub-optimal performance and fate was aggravated by the simultaneouseffects of other pathologies.

Working against

In working against, one or more agents contrast the regular and optimal networkfunctioning (Bachmann 2001; Cross and Guyer 1980; Monasson 1999; Numaoka1995; Sigmund et al. 2002; Smallbone and Lyon 2002). Two sub-species ofworking against have been identified: a first type characterizes networks suffering





the effects of ‘animosity’ among agents (aµ[t] ≤ 0) with negative links amongprospective neighbour nodes; consequently the system may exhibit sharp tran-sition phases due to these deteriorating conditions pushing to negative feedbackloops between the network’s structure and its dynamics (A[t] ≤ 0). This kind ofpathology can also affect conflicting departments of the same agent heavily influ-encing the agent’s performance and the system dynamics in which the agent iseventually involved.

Network A highly suffered the effects of the first type of working against: amounting animosity between two departments of the same agent caused itssub-optimal behaviour which was immediately reflected onto the network whichpassed through a sharp crisis, sudden increases in γ and λ and a final collapse.

In the second type of working against, one or more agents occupy a predom-inant role within the network and this pathology can occur in networks charac-terized by the presence of ‘hubs’. Working against Type II, in our survey, reportedin Network C, which refers to a condition in which some influential agentsenter a network with the role of hub (with a very low Z value), to control infor-mation flows and distribution (and the quality of the information shared) amongother agents, only to push the entire network into inability, stagnation and inactionbecause in this case the network was competing with other systems. For thisreason, the agent’s behaviour can produce a condition of crash and ‘blocking’:given the characteristics of these hubs and their links, other agents cannot, eventhough system performance would improve reducing the influence of the hub,modify the network course, create other states or a more flexible system environ-ment. In this case, with µ[t] ≤ 0 and a1

µ[t] ≤ 0 of the dominant hub, otheragents, the system unable to produce any gratification, will be forced to aµ[t] ≤ 0and consequently A[t] ≤ 0.

CONCLUDING COMMENTS

In general terms, three territorial local surveyed networks examined here haveshown high link decay and transmutation degree (high λ and γ): they have beenclassified, according to the previously described scheme, at a low ‘integration’level while the remaining two networks, showing medium link decay and trans-mutation degree (medium λ and γ) and medium coordination, presented somefeatures of condensation.

The survey’s results have been resumed in Table 3 and the networks’ perform-ances during the time course have been displayed in the time charts of Figure 3.

The empirical evidence emerging from this survey could suggest that weakernetworks and peer-to-peer networks are more exposed and sensitive to systempathologies: in the case of hierarchical or bidimensional/multidimensional net-works, it is essential to identify those agents with lower Z value and generatorsand frequently testify their performances verifying their eventual inclination to





Figure 3Networks Time Charts

one or more pathologies capable of being transmitted to the entire network. Inaddition, it could be stated that local static conditions and territorial inertia canresult not only from inadequacies in local patterns but also from an inadequatecooperation degree among local agents deriving from inefficient and dysfunc-tional territorial networks which systematically violate the local improvementassumption.

Table 3Networks’ Classification

Network Pathology Indicators Impact

A

B

C

D

E

Working AgainstType I

Rotation Type I

Rotation Type II,Working AgainstType II

Crystallization

Individualism

Negative impact. Activation ofNWOM and failure in othersimilar initiatives

No/negative impact of NWOM

No impact

No impact

No impact

Sudden increases in γ and λ.Network crash. A < 0

Network inaction.A[t] = A[t + 1] =A[t + 2] = A[t + 3]…

Network crash. A ≤ 0

Network blocking. A = 0

Underperforming network.A > 0 but J < 0 and E < 0





(Figure 3 continued)






This article described only a few of the potentially larger number of socialpathologies associated with territorial multiagent networks being focused mainlyon the most commonly perceived and frequently occurring systems diseases.






A more ‘qualitative’ approach has been adopted introducing the concept ofnetwork ‘impact’ defined as positive/negative outcomes created as a result ofnetwork activities. This approach can demonstrate that not every territorial net-work operating in a rural area will provide a positive contribution to agents’economic efficiency and overall local welfare. Territorial networks which operatebadly or produce poor performances can waste financial resources and evennegatively influence the local context and climate reverberating and percolatingtheir negative effects to other territorial operative dimensions (perverse contagionand propagation), also eroding the possibility to adopt further similar initiativesin the future for the effects of NWOM.

We have here considered the health of a territorial network as any property ofthe system that could affect the agent’s perception of the system and consequentlyobservability by the agent is a key property defining some functions expressingsystem pathologies essentially observable by the agents. This subjective point ofview surely made the present analysis rather unstable due to those variable reac-tions, in the course of time, occurring within the system related to four macro-behavioral patterns: (a) resistance (the original knowledge context is seen bythe agent in contrast with the new one); (b) change (the new knowledge contextcompletely replace—substitution—the old one); (c) incorporation (the renewedcontext is adapted in the previous one) and (d) transmutation (alteration of theprevious knowledge context through the adoption of the renewed one with thecreation of a further new original knowledge context). Further investigationsare thus required to make the present findings more robust and coherent andmore rigorously statistic and scientific conclusions to be extended and gener-alized. Having these precautions in mind, this article offered the occasion toreflect on the examples of individualism, crystallization, rotation and workingagainst, to be considered as varieties of well known pathologies in social networksdescribing the violation of the local improvement assumption, their potentialcapability to create profound effects in inhibiting link formation, to turn positivelinks into ineffective or negative ones and to enhance non-linear system behav-iour and results.

It could be stated that territorial synergic networks showing inner mechanismscapable of guaranteeing the validity of the local improvement assumption shouldbe defined ‘reliable’ and ‘properly constructed’ because they cannot exhibitpathological behaviours holding this auto-immune system. The possibility toinvestigate the quality of the interactions among networked agents and the imple-mentation of the approach described in this article, could provide useful contri-butions in alerting about actual and potential possibilities of occurring of thesesystem pathologies and preventing a system collapse caused by a deteriorationin the link value and in the eventual link losses. Further analyses are directed toidentify, investigate and record other forms of pathologies in order to identify amore stable method for providing (a) insights to network dynamics and (b) an-alytical tools that interpret and synthesize these insights rationally, operating





within apparently casual contexts. The present combination of these points (andits further analytical improvements) has provided us with a unique view at thetype of strategy needed to change specific variables in desired directions withinthe networks considered, bringing to light some strategic focal points for morestrategic interventions which will allow for long-term growth of these territorialsystems and substantial improvement in their impacts.

Carmelo Cannarella and Valeria Piccioni are at the National Research Council of Italy,Institute of Chemical Methods, Italy.

References

Albert, R. and A.L. Barabasi. 2000. ‘Topology of Evolving Networks: Local Events andUniversality’, Physical Review Letters, 85: 5234–37.

———. 2002. ‘Statistical Mechanics of Complex Networks’, Reviews of Modern Physics,74: 47–97.

Albert R., H. Jeong and A.L. Barabasi. 2000. ‘Attack and Error Tolerance of ComplexNetworks’, Nature, 406: 378–82.

Antonelli, C. 2003. The Economics of Innovation, New Technologies and Structural Change.London: Routledge.

Arthur, B. 1990. ‘Positive Feedbacks in the Economy’, Scientific American. February.Bachmann, R. 2001. ‘Trust, Power and Control in Trans-Organizational Relations’,

Organization Studies 22(2): 337–65.Barabasi, A.L. 2002. Linked: The New Science of Networks. Cambridge, MA: Perseus.Barabasi, A.L., H. Jeong, E. Ravasz, Z. Neda, A. Schuberts and T. Vicsek. 2002. ‘Evolution

of the Social Network of Scientific Collaborations’, Physica A, 311: 590–614.Barrat, A. and M. Weigt. 2000. ‘On the Properties of Small-world Networks’, European

Physical Journal B, 13: 547–60.Beer, A., G. Haughton and A. Maude. 2003. Developing Locally: An International Comparison

of Local and Regional Economic Development. Bristol, UK: The Policy Press.Bianconi, G. and A.L. Barabasi. 2001. ‘Competition and Multiscaling in Evolving Net-

works’, Europhysics Letters, 54: 436–42.Bloise, G., S. Currarini and N. Kikidis. 2002. ‘Inflation, Welfare, and Public Goods’,

Journal of Public Economic Theory, 4(3): 369–86.Bollobàs, B. 1998. Modern Graph Theory. New York: Springer.———. 2001. Random Graphs. New York: Academic Press.Bouchaud, J.P. and M. Potters. 2000. Theory of Financial Risks. Cambridge: Cambridge

University Press.Boudon, R. 1981. Effetti Perversi dell’Azione Sociale (Perverse Effects of Social Action). Milan:

Feltrinelli.Braun, A.R., G. Thiele and M. Fernández. 2000. Farmer Field Schools and Local Agricultural

Research Committees: Complementary Platforms for Integrated Decision-making in SustainableAgriculture. London: Overseas Development Institute, Agricultural Research & Exten-sion, Network Paper No. 105, p. 16.





Burt, R.S. 1987. ‘Social Contagion and Innovation: Cohesion versus StructuralEquivalence’, American Journal of Sociology, 92: 1287–1335.

Callaway, D.S., M.E.J. Newman, S.H. Strogatz and D.J. Watts. 2000. ‘Network Robustnessand Fragility: Percolation on Random Graphs’, Physical Review Letters, 85: 5468–71.

Cannarella, C. and V. Piccioni. 2005a. ‘Knowledge Building in Rural Areas: Experiencesfrom a Research Centre-rural SME Scientific Partnership in Central Italy’, InternationalJournal of Rural Management, 1(1): 25–43.

———. 2005b. ‘Public Organizations and Local Rural Development; An EmpiricalAnalysis’, Electronic Journal of Business Ethics and Organization Studies, 10(2): 16–23.

Champsaur, P.I., D. Roberts and R. Rosenthal. 1975. ‘On Cores of Economies withPublic Goods’, International Economic Review, 16: 751–64.

Clegg, S.R., M. Kornberger and C. Rhodes. 2005. ‘Learning/Becoming/Organizing’ inOrganization, 12 (2): 147–67.

Corkill, D. and S. Lander. 1998. ‘Diversity in Agent Organizations’, Object Magazine,8 (4): 41–47.

Cross, J., and M. Guyer. 1980. Social Traps. Ann Arbor: University of Michigan Press.Curran, J. 1994. Small Firms and Local Economic Networks. London: Sage Publications.Daft, R. 2001. Organization Theory and Design. Ohio: South Western.Degenne, A. 1999. Introducing Social Network. London: Sage Publications.Dorogovtsev S.N. and J.F.F. Mendes. 2000. ‘Scaling Behaviour of Developing and

Decaying Networks’, Europhysics Letters, 52: 33–39.———. 2002. ‘Evolution of Networks’, Advances in Physics, 51: 1079–187.Edwards, W.M. and T.R. Eggers. 2004. ‘Agricultural Management e-School: Extension

Education over the Internet’, American Journal of Agricultural Economics, 86 (3): 778–81.Etcheson, C. 1989. Arms Race Theory: Strategy and Structure of Behavior. New York:

Greenwood.Galbraith, J. 1973. Designing Complex Organizations. Reading, MA: Addison-Wesley.Geels, F.W. 2004. ‘From Sectoral Systems of Innovation to Socio-technical Systems:

Insights about Dynamics and Change from Sociology and Institutional Theory’,Research Policy, 33 (6–7): 897–920.

Glance, N. and B. Huberman. 1994. ‘The Dynamics of Social Dilemmas’, ScientificAmerican, March: 76–81.

Glance, N. and T. Hogg. 1995. ‘Dilemmas in Computational Societies’, Proceedings of theFirst International Conference on Multi-Agent Systems (ICMAS-95), pp. 117–24. MenloPark, California: AAAI Press.

Gleditsch, N. and O. Njolstad. (eds). 1990. Arms Races: Technological and Political Dynamics.London: Sage Publications.

Goh, K.I., E. Oh, H. Jeong, B. Kahng and D. Kim. 2002. ‘Classification of Scale-freeNetworks’, Proceedings of the National Academy of Sciences, 99 (20): 12583–88.

Greve, H.R. 2005. ‘Interorganizational Learning and Heterogeneous Social Structure’,Organization Studies, 26: 1025–47.

Hall, R. 1968. ‘Professionalization and Bureaucratization’, American Sociological Review,33: 92–104.

Hall, A., N. Clark, S. Taylor and V.R. Sulaiman. 2001. ‘Institutional Learning throughTechnical Projects: Horticulture Technology R&D Systems in India’. London,Overseas Development Institute, Agricultural Research & Extension Network PaperNo. 111, p. 12.





Hardin, G. 1968. ‘The Tragedy of the Commons’, Science, 162: 1243–48.Hauser, J., G. Urban and B. Weinberg. 1993. ‘How Consumers Allocate their Time

when Searching for Information’, Journal of Marketing Research, 30: 452–66.Herlocker, J.L., J.A. Konstan, L.G. Terveen and J.T. Riedl. 2004. ‘Evaluating Collaborative

Filtering Recommender Systems’. ACM TOIS, 22 (1): 5–53.Herr, P., F. Kardes and J. Kim. 1991. ‘Effects of Word-of-mouth and Product-attribute

Information on Persuasion: An Accessibility-diagnosticity Perspective’, Journal ofConsumer Research, 17: 454–62.

Holt, K. 2002. Market Oriented Product Innovation. A Key to Survival in the Third Millennium,Kluwer.

Jain, S. and S. Krishna. 2001. ‘A Model for the Emergence of Cooperation, Inter-dependence, and Structure in Evolving Networks’, Proceedings of the National Academyof Sciences, 98: 543–47.

Jensen, D. and V. Lesser. 2002. ‘Social Pathologies of Adaptive Agents’, in M. Barley andH. Guesgen (eds), Symposium vol. IR SS-02-07. Safe Learning Agents: Papers from the2002 AAAI Spring. California: AAAI Press.

Jensen, D. and P. Cohen. 2001. ‘Multiple Comparisons in Induction Algorithms’, MachineLearning, 38 (3): 309–38.

Jeong, H., Z. Neda and A.L. Barabasi. 2003. ‘Measuring Preferential Attachment inEvolving Networks’, Europhysics Letters, 61: 567–72.

Kaindl, H. 1988. ‘Minimaxing: Theory and Practice’, AI Magazine, 9 (3): 69–76.Kickert, W.J.M. 1997. Managing Complex Networks, London: Sage Publications.Kilduf, M. 2003. Social Networks and Organizations, London: Sage Publications.Leeuwis, C. and A.W. Van den Ban. 2004. Communication for Rural Innovation: Rethinking

Agricultural Extension. Oxford: Blackwell.Lundstedt, S.B. and T.H. Moss. 1989. Managing Innovation and Change. Dordrecht: Kluwer.Mantengna, R. and E. Stanley. 2000. Econophysics. Cambridge: Cambridge University

Press.Marquis, M. and P. Filiatrault. 2002. ‘Understanding Complaining Responses through

Consumers’ Self-consciousness Disposition’, Psychology & Marketing, 19 (3): 267–92.Marsden, P.V. 1990. ‘Network Data and Measurement’, Annual Review of Sociology, 16:

435–63.McAdam, R. 2004. ‘Knowledge Creation and Idea Generation: A Critical Quality Per-

spective’, Technovation, 24 (9): 697–705.Milleron, J.C. 1972. ‘Theory of Value with Public Goods: A Survey Article’, Journal of

Economic Theory, 5: 419–77.Monasson, R. 1999. ‘Diffusion, Localization and Dispersion Relations on “Small-world”

Lattices’, European Physical Journal B, 12: 555–67.Nau, D. 1983. ‘Pathology on Game Trees Revisited, and an Alternative to Minimaxing’,

Artificial Intelligence, 21: 221–44.Newman, M. 2003. ‘The Structure and Function of Complex Networks’, SIAM Review,

45: 167.Numaoka, C. 1995. ‘Introducing the Blind Hunger Dilemma: Agents’ Properties and

Performance’, in Victor Lesser (ed.), Proceedings of the First International Conference onMulti-Agent Systems (ICMAS-95) pp. 290–96. California: AAAI Press.





Oates, T. and D. Jensen. 1997. ‘The Effects of Training Set Size on Tree Size’, in DouglasFisher (ed.), Proceedings of the Fourteenth International Conference on Machine Learning,pp. 254–62. San Mateo, CA: Morgan Kaufmann.

Ottosson, S. and E. Björk. 2004. ‘Research on Dynamic Systems—Some Considerations’,Technovation, 24 (11): 863–69.

Proykova, A. and D. Staufer. 2002. ‘Social Percolation and the Influence of Mass Media’,Physica A, 312: 300–04.

Reese, L. 1997. Local Economic Development Policy: The United States and Canada. NewYork: Garland Publishing.

Rege, M. 2004. ‘Social Norms and Private Provision of Public Goods’, Journal of PublicEconomic Theory, 6 (1): 65–77.

Rogers, E.M. 1979. ‘Network Analysis of the Diffusion of Innovations’, in P.W. Hollandand S. Leinhardt (eds), Perspectives on Social Network Research, pp. 137–64. New York:Academic Press.

Romney, A.K. 1989. ‘Quantitative Models, Science and Cumulative Knowledge’, Journalof Quantitative Anthropology, 1: 153–223.

Scott, J. 2000. Social Network Analysis. London: Sage Publications.Sigmund, K., E. Fehr and M. Nowak. 2002. ‘The Economics of Fair Play’, Scientific

American, January: 83–87.Smallbone, D. and F. Lyon. 2002. ‘A Note on Trust, Networks, Social Capital and Entre-

preneurial Behaviour’, H.H. Höhmann and F. Welter (eds), Entrepreneurial Strategiesand Trust: Structure and Evolution of Entrepreneurial Behavioural Patterns in East and WestEuropean Environments—Concepts and Considerations. Forschungsstelle OsteuropaBremen, Arbeitspapiere und Materialien, 37, Bremen.

Solomon, S., G. Weisbuch, L. de Arcangelis, N. Jan and D. Staufer. 2000. ‘Social Per-colation Models’, Physica A, 277: 230–47.

Sugawara, T. and V. Lesser. 1998. ‘Learning to Improve Coordinated Actions in Cooper-ative Distributed Problem-solving Environments’, Machine Learning, 33: 129–53.

Thomson, W. 1999. ‘Economies with Public Goods: An Elementary Geometric Expos-ition’, Journal of Public Economic Theory, 1: 139–76.

Townley, B., D. Cooper and L. Oakes. 2003. ‘Performance Measures and the Rational-ization of Organizations’, Organization Studies, September (24): 1045–71.

Turner, R. 1993. ‘The Tragedy of the Commons and Distributed AI Systems’, TwelfthInternational Workshop on Distributed Artificial Intelligence, Hidden Valley, PA: 379–90.

Vazquez, A. 2003. Growing Networks with Local Rules: Preferential Attachment,Clustering Hierarchy and Degree Correlations’, Physics Review, 67.

Von Zedtwitz, M., G. Haour, T. Khalil and L.A. Lefebvre (eds). 2003. Management ofTechnology: Growth Through Business Innovation and Entrepreneurship. London: Pergamon.

Walton, E.J. 2005. ‘The Persistence of Bureaucracy: A Meta-analysis of Weber’s Modelof Bureaucratic Control’, Organization Studies, 26: 569–600.

Wasserman, S. and K. Faust. 1994. Social Network Analysis: Methods and Applications.Cambridge: Cambridge University Press.

Watts, D.J., P.S. Dodds and M.E.J. Newman. 2002. ‘Identity and Search in SocialNetworks’, Science, 296: 1302–05.

Wellman, M. 1993. ‘A Market-oriented Programming Environment and its Applicationto Distributed Multicommodity Flow Problems’, Journal of AI Research, 1: 1–23.




international journal of rural management€¦ · decisions (braun et al. 2000; edwards and eggers...

Documents