Bachmann, Lorenz B. R.: Review of the Agricultural Knowledge System in Fiji - Opportunities and Limitations of Participatory Methods and Platforms to promote Innovation Development -


Chapter 2. Theory review: Agricultural Innovation and Agricultural Knowledge Systems

Any theory can only reflect a part of reality. Depending on the viewpoint, the theory will explain certain aspects very well and fail to explain others. Conceiving a theory only as a perspective on reality, it becomes clear that the perspective ’looking at the front of the house‘ can describe the front very well, but it will fail to tell anything about the back of the house and information about the inside will be limited to what can be seen through the front windows.

In this sense, the researcher‘s intention is not to create a new theory about innovation in agricultural development. Instead, attention is drawn to the possibility to examine reality through several windows (theories) and to combine the findings to form a new understanding. In practice a selection of important theories will be examined. By reviewing their specific strengths and weaknesses, it will be shown what these theories have to offer to interpret innovation and change in agriculture. The underlying assumption is that such multiple perspectives are a better means to interpret the complex processes of development than single positivist<7> models.

This chapter begins with a look at a selection of older and newer theories of agricultural innovation. The main part of this chapter consists of a detailed review of agricultural knowledge systems.

2.1 Comparison of innovation theories

A comprehensive review of innovation theories would be outside the scope of this thesis. Instead, a selection of three theories and their relevancy within the agricultural development process will be examined closer.

Articulated first in the sixties and early seventies by Rogers and others, the ’diffusion of innovations model‘ is probably among the best known concepts. Rogers and Shoemaker (1971, 19) define an innovation as:

“... an idea practice or object perceived as new by an individual. It matters little, so far as human behaviour is concerned, whether or not an idea is ’objectively‘ new as measured by the lapse of time since its first use or discovery. It is the perceived or subjective newness of the idea for the individual that determines his reaction to it. If the idea seems new to the individual, it is an innovation.“

In this context social change is understood as a process including three sequential stages: invention, diffusion and consequences (Rogers and Shoemaker 1971, 38). Technical


change in agriculture is consequently understood as the result of the adoption of technical innovations by farmers. Scientific research is seen as the source of such innovations<8>.

Christoplos and Nitsch (1996, 28) review the diffusion model and describe adopter categories, adoption process and characteristics of innovation as the three main elements:

The Adopter categories classify farmers according to the rate of adoption of a new technology or practise. The first adopters are called innovators. They are followed by early adapters, early majority, late majority and laggards. The categories are associated with certain characteristics. Innovators are presumed to be venturesome, the late majority sceptical and laggards traditional. Earlier adopters are expected to have more education, higher social status and larger and more specialised farms. They are further considered as less dogmatic, less fatalistic, more rational and achievement oriented, and to hold a more favourable attitude toward credit, change, risk, education and science. Furthermore, they participate more in farmer organisations, are more cosmopolitan, have more contacts with outsiders, are aware of new recommendations and exert influence on local opinion. Late adopters on the other hand are characterised as being negative to change, risk and science, and as having little contact with extension services. Several extension methods, in particular the training and visit system (T&V)<9> are implicitly based on the diffusion model, recommend choosing contact farmers in the categories of innovators and early adopters which are sometimes titled as progressive, outstanding or model farmers.

The adoption process describes the stages an individual goes through from the first exposure to an innovation to actually adopting it. The model distinguishes five stages: Awareness stage, interest stage, evaluation stage, trial stage and adoption stage.

Difference in speed of adoption are explained by five characteristics of innovation:

  1. Relative advantage: the degree to which an innovation is perceived as better than the idea it supersedes.
  2. Compatibility: the degree to which an innovation is perceived as consistent with the existing values, past experiences and needs of the receivers.
  3. Complexity: the degree to which an innovation is perceived as relatively difficult to understand and use.
  4. Trialability: the degree to which an innovation may be experimented with on a limited basis.
  5. Observability: the degree to which the results of an innovation are visible to others.

The diffusion model and in particular its practical applications as part of the T&V system or the transfer of technology has been criticised by various authors (cf. Röling 1988a, 207; Engel 1995, 145).


One of the major deficiencies of the model is that the stage of technology generation is omitted in the model. It assumes that a chosen technology<10> is appropriate. If a farmer applies a technology in the correct way, all unfavourable factors (diseases, weather, environment) are expected to be mitigated. The key assumption is that these negative factors can be controlled by proper farm management. However, unlike the conditions in industrial countries where the diffusion model was first developed, farmers in rainfed marginal areas of development countries generally do not have the means to control their environment. They have to adapt their choice of technologies to rapidly changing agro-ecological and socio-economic conditions. “If the rains are late, if the local shop runs out of fertiliser, or if the children become sick and medical expenses are such that the family can no longer afford to buy the recommended inputs, farmers may have to totally change their plans. Such changes do not fit in the planning framework embodied in the diffusion model“ (Christoplos and Nitsch 1996, 31).

Another major flaw of the model is its strong sender perspective that tends to promote elitist tendencies. The characterisation of farmers in the late adopter groups as negative to change, risk and science makes them appear as irrational. Adoption is considered the only rational behaviour. Consequently, extensionists are guided to explain non-adoption as farmer deficiency. “We know what is best for them. It is our task to help them and become knowledgeable.“ The diffusion model is thus impregnated with a strong pro-change, pro-innovation and pro-technology biases that encourage its users to develop an elitist attitude towards their target groups“ (Christoplos and Nitsch 1996, 32).

Considering this heavy critique, one might ask why the model is presented here again. One reason is that the model is still used in many development countries including Fiji<11>. However, critical revisions of the concept are rare. Learning from the above critique, it becomes evident that it is important that ’adoption studies‘ are required already as part of the technology generation process to ensure that a technology is appropriate. Furthermore, a new technology‘s capital requirements and scale of production needs to be assessed in


order to estimate for which group of farmers it might be suitable. Thus, in many cases a diffusion of a certain technology can only be expected to reach a certain specified group of users. Innovations that fit all farmers are rare exceptions. A good monitoring system of extension activities should consequently always be combined with research into the adoption and diffusion processes. For Fiji, experimentation with different farmers groups and different types of technology could show which groups qualify best (or if at all) as ’model farmers‘ to promote extension efforts. This could provide insights into how adopter attributes could be redefined. In this modified way, diffusion thinking could still contribute ideas for organising extension work. That is the main reason why the model is taken up here again.

Another tradition of innovation research is that of ’induced innovation‘. “Farmers are induced, by shifts in relative prices, to search for technical alternatives that save the increasingly scarce factor of production (Hayami and Ruttan 1985, 88)“. The model, however, does not consider technical change as entirely of an induced character. All actors such as farmers, scientists and planners etc. play active roles in responding to exogenous (supply) and endogenous (demand) factors and taking part in the general progress of science and technology. Consequently, the model defines technical change as “... any change in production coefficients resulting from the purposeful resource-using activity directed to the development of new knowledge embodied in designs, materials and organisations“ (Hayami and Ruttan 1985, 86).

The induced innovation school points at the importance of the economically scarce factor for directing innovation processes. It makes clear that innovation processes have to be seen in their specific social and economic context. Innovations have to be economically feasible and reward the user with an economic advantage. Economics have to be seen as a cornerstone of development and innovation processes. However, the tradition also has its limitations. In subsistence agriculture, many decisions can not be determined in monetary terms. Hence farmers do not always behave according to economic rationality and environmental factors all too often remain unconsidered.

A third, recent school of thought, Engel (1995, 146) labels “the network tradition“.

Analysing innovation processes in larger industries, Moss-Kanter (1989) looks at types of co-operations between companies. Pooling, allying and linking (PAL) between companies, is recognised as an important strategy to generate innovation and improve competitiveness. This can also be observed in agriculture, where networking is becoming very popular in recent years. Many organisations and NGOs (AGRECOL, IIED, ILEIA, ITDG, IBSRAM just to name a few) are active around the globe trying to exchange information and cooperate in various fields. Engel (1995, 147) describes the essence of the network tradition as follows:

“It concentrates upon all social interactions relevant to agricultural innovation at a particular point in time within a specific social, economical and ecological context. It assumes that in any given situation a multiplicity of social actors develop and mange interactive relationships in order to improve their practises and develop new ones. The reason that these actors engage in such relationships is perceived interdependence: each is perceived as holding some of the keys to the others‘ projects.“


Networks, thus, build on the different specialised skills that result from the division of labour in agriculture and surrounding sectors. A concept on how these network relations function is proposed by Gremmen (1993) with his ’interplay model‘: practices<12> evolve autonomously in interaction of different social actors. Each can be seen as a competent performance, constraint only by its own defining and rules that emerge by experience. These rules are subject to continues revision by social interaction of the participants in a practice. Knowing as an activity rather than knowledge is crucial. “The central claim of the interplay model is that improvement is primarily an internal achievement of practices themselves. External influences can speed up or slow down the indigenous improvements of a practice“ (Gremmen 1993, 159). Open inter-action between practises must be seen as an external influence on practices. These influences are generally not directed only one-way. In this sense innovation in practices is a result of interaction in practises and not to be seen as a discovery process of only one practise such as science. “Science is often, and mistakenly, seen as the ideal way of advancing knowledge“. In the contrary different practises such as science and technology may be seen as “enmeshed in a symbiontic relationship ... science as one context of inventive activity“(Gremmen, 1993, 116 and 140).

Comparing the three schools of thought (diffusion, induced innovation, networks), all hold some valuable ideas to characterise innovation processes. Therefore, none of the traditions should be discarded, but rather integrated in a more comprehensive conceptualisation of innovation as a result of social processes. In this respect, the network approach appears as a suitable basis for further elaboration.

Analysing various social networks, Engel (1995, 112) identifies five basic configurations of networks with different leadership as the driving force of innovation:

Industry driven configurations

If the market place is the ultimate place where success or failure of agricultural innovation is determined, those social actors who control the marketing or processing of produce lead ahead: marketing boards, traders, retail or food chains. Co-ordination between key actors on the work floor is achieved through co-operative agreements and/or commercial contracts. Innovations are promoted which increase the profit margins of the participants in the product chain. This does not necessarily imply an increase of profitability at farm level: innovation may be required to improve logistics for collecting of produce (e.g. milk collection tanks) or to comply with international standards. Participating farmers are not necessarily large scale or rich, but often some commercial outlook is required. The number of farmers is determined by the market share of leading actors and the current level of on-farm productivity. Increasing the latter is the often the preferred strategy to


increasing the number of producers. Common is that farmers are given production contracts. Generally, agro-commercial establishments actively take part in the transfer of technologies. Research is organised in commodity research programmes. Extension is likewise commodity oriented. Emphasis is laid on the development of appropriate packages. Research and extension services are (partly) company owned and financed in relation or indexed to market sales. Quantities and qualities, on the (world) market, are the main controlling forces, which lead to the development of production oriented technologies and a standardisation of agricultural produce. Competition leads to increasing specialisation of agriculture in sectors and sub sectors (e.g. dairy, cotton coffee, etc.). Even more sophisticated patterns evolve in production chains with specific markets. Market orientation is the strong point of these configurations, while a lack of sensitivity to social differentiation and long term ecological deterioration are often observed as associated negative symptoms.

Policy driven configurations

Where national Government directs the course of innovation, one may speak of a policy led configuration. In such cases, generally, the Government is the main source of finance for research and extension. Commercial or non-governmental actors are relatively weak. The Government imposes its leadership through the implementation of its policies and development projects and programmes. In this role, Governments may be supported to a variable extent by external donors. Technologies are extended via demonstration farmers to the rural constituency. Co-ordination between core actors (e.g. Ministry of Planning, Ministry of Agriculture, etc.) is achieved via the existing hierarchical structures and policies: planning and project approval procedures, technology certification procedures, official mandates of (semi-) government institutions, bureaucratise procedures for budget negotiations, allocation of resources, monitoring and control mechanisms. Industrial actors generally play a secondary role in such configurations. If technologies are believed to be available ’on the shelf‘ and farmers only need to be trained to use them, research may play a minor role by being bypassed, while technologies are promoted via extension services directly . The resulting configuration is characterised by a rigid definition of tasks and limited flexibility. Just like market led innovation, policy led innovation may contribute to social differentiation. However, focus upon national priorities may also include goals such as social welfare and equity or ecological sustainability. Strong point of the configuration is the high sensitivity to policies, while bureaucracy and inefficient use of resources are often reported as a weak point.

Farmer driven configurations

Here, the relationships between farmers‘ leaders, farmers‘ organisations and agricultural policy makers are dominant. However, most case study evidence of ISNAR studies reveal a lack of farmer influence rather than a dominant position. Strong farmer driven configurations seem to be most frequent alongside well developed cash crop industries (e.g. coffee in Columbia, cf. Kaimowitz 1989). Farmers are organised via unions, co-operatives, associations or other functional groups. Co-ordination between actors is probably achieved by standardisation of interests: farmers‘ organisations articulate their needs at different levels, and if possible, influence other actors (e.g. research and extension) to adjust their activities accordingly. The problems associated with (mostly) farmer driven configurations lie in the organisations themselves, their capacity to


effectively make decisions on technology development, their relationships with rank-and -file membership and their representativeness in respect of the variety of farmers‘ interests.

Research and development driven configurations

Here, the course of innovation is within the hands of national and/or international research institutes. Research institutes generally have a relatively free hand to determine their priorities, research approach and the way they co-ordinate with other actors to disseminate results. Two main currents of work may be distinguished: technology push and technology development. The former is geared towards compilation ready-to-disseminate packages to be transferred by extension following the ’linear model‘. In the later, no easy replicability of technological packages is assumed, but the need for local design and adaptation to suit local conditions and resource base of farmers is stressed.

Co-ordination of tasks in research and development led innovation configurations is based on a standardisation of skills and is rooted in the definition of what a competent researcher does: his or her prior qualifications, the accepted research approach (on-farm, on-station, FSR etc.), accepted research protocols (plot designs, data collection methods; etc.) and the type of results expected (new varieties, practical recommendations, etc.). Staff recruitment and training are the institutes most important instruments to maintain the quality of innovative performance. A strong point of research and development driven configurations is their potential to develop ’tailor-made‘ specific technologies that fit the needs of specific groups of farmers. Its weakness lies in the replicability of results, while appropriate technological solutions are developed only small scale, the wider application fails or lags behind.

Donor driven configurations

Foreign or national donor agencies, agricultural projects and demonstration farmers are the core actors in donor driven configurations. Agricultural projects serve as ’temporary support structures‘ to ensure that donor policies are safeguarded and implemented. Donor policies and approaches vary widely. Some donors stress the importance of flexibility and mutual adjustment, while others insist on standardisation of all tasks and strict administrative control. In order to achieve results in a short period of time, task-orientation is commonly used. Generally, donor projects have sufficient freedom to identify a suitable approach to deal with the existing problems. However, a major problem of sustainability for the partner organisations arises when the projects are terminated and funds stop to flow (Engel 1995, 112-115).

Engel (1995, 115-118) is aware that his basic configurations are ideal-types (abstract constructs of realities). He does not propose them as an account of ’what is the world‘, but rather as a complementary way to view innovation processes. The particular advantage of the concept is that they may help to identify gaps or changes in leadership patterns. Equally imaginable are multiple leadership patterns, where actors of innovation configurations exert temporary leadership in respect of specific tasks or times. Understanding innovation processes should thus be guided by the questions: whose perception is more relevant? And eventually: which perception accommodates the perspective of most actors? And not who‘s perception is right?

Engel‘s (ibid.) concept is a useful conceptual tool to characterise innovation processes. However, it is clear that there is a continuous transition between the different ideal types.


Within a given agricultural knowledge system, leadership may even vary depending on the commodity examined. The leadership patterns provides an interesting analytic frame for assessing the case of the Fijian agricultural knowledge system. In the empirical section (chapters 5 and 6), the concept will be used to analyse the main driving forces for innovation in Fiji and characterise their specific strengths and weaknesses.

The following Table summarises the main characteristics of the before mentioned configurations of social actors.

Table 2: Basic configurations

Type of configuration /Characteristic

Industry driven

Policy driven

Farmer driven

Research & development driven

Donor driven

Principal co-ordinating mechanism

standardisation of outputs / technical packages

direct supervision, stand. of work processes / technical packages

stand. of interests, norms

stand. of skills, education

mutual adjustments, stand. of technical packages, skills, work processes

Dominant leaders

market actors

agric. policy makers

farmers‘ organisations

(inter) national research


Core actors

agro-commerce / industry , entrepreneur farmers

agric. policy /extension / demo farmers

agric. policy / farmers‘ leaders

agric. Research / innovative farmers

agric. projects / demo farmers

Rural constituency

commercial farmers

program farmers

organised farmers

technological farmers

program farmers

Principal source of power/influence

market articulations, prices, quality control, resources

policies / rules and regulations, resources

political clout / resources

(improved) technology, technical expertise

financial resources, technical expertise

’Leitmotif‘ for innovation

efficiency / output quality

policy objectives

farmers‘ needs

technical advancement

intervention objectives

Accountability to

individual balance sheet

Government policies

farmers‘ interests

research community

donor policies

Source: Engel 1995, 118.

2.2 Agricultural Knowledge Systems

Research results, if not being applied by farmers become useless. Extension services alone, without new appropriate messages, are irrelevant for farmers. Innovations that do not solve actual farmers‘ problems, stand little chance of being adopted. All three statements give good reasons for suggesting a systems approach to agricultural development. The first systems approach proposed by Nagel (1980) identified research, extension and farmers as the three key elements, or as he called them, sub-systems of the overall agricultural knowledge system (AKS). He used the model to examine the knowledge flow and linkages of the three sub-systems at two Indian state universities.


The agricultural University of Wageningen has fostered research into knowledge systems since the mid eighties and a number of model variations have evolved (cf. Wijeratne 1988, Blok and Seegers 1988, Röling 1988 a, b and 1990). A definition of an agricultural knowledge and information system (AKIS) from this school is given by Röling:

“An AKIS is a set of agricultural organisations and / or persons, and the links and interactions between them, engaged in such processes as the generation, transformation, transmission, storage, retrieval, integration, diffusion and utilisation of information with the purpose of working synergically to support decision making, problem solving and innovation in a given country‘s agriculture or domain thereof (Röling 1990, 1)“.<13>

For a better understanding, some terms of the concept need further clarification: knowledge, information and system synergy.

One of the pioneers of knowledge systems thinking, Havelock (1986, 13) points at the difficulty in defining the term knowledge and consciously leaves it open. Further along, however, in the context of knowledge utilisation, he describes knowledge as something which can be transmitted or transferred (Havelock 1986, 21). In the contrary, Röling (1990, 12) defines knowledge as a property of mind which cannot be transmitted to others unless transformed or encoded. He understands knowledge processes (memory storage, transformation, etc.) as intra-personal. In the same direction Long (1992, 27) states that “knowledge is not simply something that is possessed and accumulated: it emerges out of processes of social interaction“. Arce and Long (1987, 5) proposed to define knowledge as being “...constituted by the ways in which individual members of a society or social group categorise, code, process and impute meaning to their experiences“. Engel (1995, 151) attributes four dimensions to knowledge: “Firstly, knowledge can be seen as a ’cognition‘, a human faculty to perceive or conceive; secondly, knowledge is ’practical‘, intrinsically woven into the daily practises of an individual or group; thirdly, knowledge can be perceived as a property of the ’individual‘, enabling him or her to infer from experience, observation and /or reasoning; finally, knowledge is ’socially‘ constructed, embedded as it is in social dynamics of an organisation, a community or a group“. Therefore, Engel (ibid.) suggests to talk about knowing rather than knowledge, as this expresses better the dynamic unity of learning and doing rather than the static aspects of knowledge as a statement about the world.

Röling‘s above cited understanding of knowledge as something not transferable is the reason why he includes information as a separate notion in his definition of knowledge systems (agricultural knowledge and information system). According to him, information, not knowledge is transferred within the system. But Röling is not undebated and the term information leaves ample room for different interpretation.

Havelock (1986, 14) understands information as something purer than knowledge, as something more acceptable and manipulateable in scientific terms, less freighted with


cultural baggage. For him information is close to ’message‘. With the difference being that the later stresses the attribute of transferability. Röling (1990, 12) defines information as “...a pattern imposed on data which simultaneously affects the interpretation of those data and enables them to be transmitted“. In other words information is a relative concept, as it is affected by communication between a sender and receiver. In this process information is transformed twice and thus exposed to distortions.

Following Röling‘s conception, knowledge transfer does not happen directly, but it builds up over time as an ’end-result‘ of information transfer (or better information exchange in communication processes).

Engel (1995, 62) has a similar conclusion as he suggests that knowledge transfer takes place in settings that offer “...joint learning opportunities amongst people who possess different kinds of knowledge“. These settings enable a temporary intensification of communication that in turn enables an accumulation and exchange of knowledge.

The theoretical difficulties in getting to grips with knowledge, information and communication processes, illustrate why these processes cause even more trouble in practice. For the purpose of this study the before mentioned Röling and Engel definitions of knowledge and information will be retained. These definitions are open enough to point to the various problems that may be encountered in practice. A particular problem in the study case Fiji, but the same holds true for many other development countries, is the multi-ethnic composition of the population. This obliges all actors within the knowledge system to communicate in various languages (e.g. Fijian, Hindi, Chinese). English as the official language, is a foreign language to most actors. Communicating in another language than the mother tongue entails a higher risk of unclear communication and misunderstanding.

Returning to Röling‘s (1990, 1) AKIS definition, one central idea is system synergy. The system view points to improving the management and the performance of the system, as deemed desirable by the participants of the system. By playing complementary roles and working together, the output of the system will be higher (synergy effect) than by focusing on an isolated improvement of individual parts of the system. Röling (1990, 14) contents in this respect, that “...if only the actors in an AKIS would begin to see themselves and other actors as playing complementary roles, many AKIS would ’auto-improve‘“.

Still missing, so far, is a closer definition of the term system within the concept of knowledge systems. Nagel (1980, 17) defines system as “a set of units with relationships among them“. His units are research, extension and farmers. Havelock (1986 b, 77) understands knowledge systems as being similar to concepts from the general systems theory “networks of connected entities“. Within agricultural knowledge systems thinking, different authors have connected larger or smaller number of entities to include in their system.<14> Röling 1994 and Engel 1995 created a new knowledge system view by


distinguishing hard and soft systems thinking. This new important concept will be discussed at a later stage (chapter 2.2.2). However, before doing so, it is important to take a look at what functions an AKIS has to execute and how these can be influenced and improved.

2.2.1 Knowledge system functions

In literature, quite a number of different functions are attributed to agricultural knowledge systems. Table 3 presents a list of functional steps as proposed by various authors.

Table 3: Functions of knowledge systems according to various authors

Nagel 1980, 23

Lionberger 1986, 117

Röling & Engel 1991, 125

Blum 1991, 324

Eponou 1993, 18

Need identification

Generation of innovative knowledge

Operationalization of knowledge

Dissemination of knowledge

Utilisation of knowledge

Evaluation of experiences
















Problem identification

Review scientific & indigenous knowledge

Basic Research & Development

Adaptive Research & Development

Sustainability assessment

Optimal means of Communication


Diagnose farmers' problems

Design a research program

Generate technologies

Consolidate technologies

Disseminate information and knowledge

Approve and release technologies

Multiply improved genetic material and duplicate technology packages

Deliver technologies

Evaluate technologies

At a first glance, it appears that the suggested functions differ considerably. However, a closer look reveals that many functions are similar and differences are a result of divergent terminology for basically one and the same function. For a better comparability, corresponding or similar functions are presented in the same row of the table. The functions cover the spectrum from problem or need identification to the adoption and evaluation of an innovation.


Two concepts will be used further. For the analysis of the Fiji AKIS, the concept of Eponou (1993) will be used. It is very detailed and therefore useful for the practical analysis. For the further theoretical discussion here, Nagel‘s (1980) set will be used, as it provides a very good general framework to structure knowledge systems functions. Where deemed useful, ideas of other authors and the researcher‘s own views are added to complement Nagel‘s original concept.

Need identification, opportunity analysis or potentials identification

The direction of activities within an agricultural knowledge system is determined by the actual needs of its sub-systems (or “connected entities“ - Havelock; or “actors“ Engel nomenclature) and to a certain degree by the outside surrounding (macro-) system of institutions and policy framework. Regardless of the concrete manifestations of these interests, Nagel (1980, 24) assumes that the basic determinants are the knowledge needs of farmers. Aware of deficiencies in practice he adds: “serving the needs of farmers is a postulate to which at least lip service is paid by everyone involved.

Two levels of decision making are involved in need identification. On the first level, the actual farmers‘ level, the problem of distinguishing between individual farmer‘s problems and problems that concern a larger number of farmers arises. It is a problem of prioritisation. Which of the many farmers‘ problems should be researched? On the second level, the institutional and policy level, matters may be quite removed from actual field problems. What counts here are the national policy goals, the needs of institutions and the availability of funds. However, policy formulation often leaves considerable room for interpretation. Therefore, which of the actual farmers‘ problems become investigated, also depends, to a considerable extent, on the personal preferences and prejudices of researchers and extensionists (Nagel 1980, 24).

Need identification is certainly one of the most difficult functions of an agricultural knowledge system. As Nagel (ibid.) pointed out, needs may be spelled out by farmers themselves or defined on the level of researchers or politicians.

It may be suggested that need identification could be viewed as a step-by-step process. The first step consists of a collection of existing problems. Then, potential solutions could be collected. Here, it is not crucial if the idea is actually formulated by a farmer or any other member of the system. In the contrary, it would be useful to exploit a larger pool of actors to increase the chance of finding meaningful innovative ideas. These problems and potential solutions should be analysed together with farmers. Does the new idea make sense for farmers (or a defined group of farmers)? Would it fit within the existing farming system(s)? Participatory methods could be the means to enable discussions on the search for innovative ideas. The discussions could collect, in a two-way process, ideas from farmers and act as a particular good first check for ideas originating outside the farmers‘ subsystem. This would then weed out ideas, which go against farmers‘ reality or common sense.

Considering that participatory methods can play such an important role in a knowledge system, a definition of the term participatory methods is necessary. Many different terms for participatory methods are used in developing countries, however, ’Participatory Rural Appraisal‘ (PRA) is currently the most well known term. The methods or tools described


as PRA are still evolving and their definitions can not be understood as definite. Chambers (1994, 953) suggests the following circumscription for PRA<15>:

“A family of approaches and methods to enable rural people to share, enhance, and analyse their knowledge of life and conditions, to plan and to act.“

According to this definition the focus of participatory methods is on sharing and analysing knowledge for rural people, with a view to making use of it, ’plan and act‘. PRA draws on several sources or traditions<16>. An important source is Rapid Rural Appraisal (RRA). The main difference being, that RRA is more focused on the simple gathering and analysis of information, while in PRA the aspect of sharing and making rural people participate has more weight. The understanding of participation can go as far as empowerment of the rural people.

Returning to the function of need identification in a knowledge system, Röling (1990, 24 and 1991, 10) equally calls for more participation, and advocates user control for an effective functioning of AKIS at all stages. It is agreed that user control is an important prerequisite to ensure a proper need orientation. However, a practical problem is the question, which farmers can effectively represent farmers? Here it is of crucial importance that farmers participate who best represent the farming community. Members of farmers organisations might be suitable partners, but in many countries such organisations do not even exist or represent only a small minority of farmers. Another problem is that strong user control may go against the grain of both bureaucrats and researchers in strongly hierarchically structures which prevail in many developing countries. In this respect, it is probably already a big step ahead if a reasonable farmer participation in the field can be agreed on, but to achieve user control it still has a long way to go.

Generation of innovative knowledge

Innovative inputs are the primary input in the knowledge system. Without innovations, a knowledge system would be obsolete<17>. Modern agriculture is depending increasingly on knowledge inputs. The agricultural research sector is generally given the task to generate innovations. Research can be broadly classified into fundamental or basic research and applied or adaptive research (Nagel 1980, 27). He criticises that “all too often, the theory-practise scale is perceived as a ladder with a top and a bottom, the top representing the ’pure researcher‘ involved in basic science and the bottom by the farmer who does the manual labour. A similar differentiation is to be found in within the research sub-system itself, where applied research is less highly regarded than basic research“ (Nagel 1980, 37). He further stresses that the systems perspective explicitly abandons this view. All elements of the system play a functional role in goal attainment at


the macro level. Status differences blocking inter-system linkages are undesired as they lead to malfunctioning of the system.

Considering the small size of many developing countries and their limited financial resources, it is obvious that fundamental or basic research is outside the scope of most developing countries<18>. Small research systems should be focused on adaptive research. However, returning to Rogers and Shoemaker‘s (1971, 19) definition of an innovation as “an idea or practise perceived as new“ by the user community it becomes evident that not always objectively new ideas are need. There is an abundance of information and technology available on the shelf. Ezaguirre (1996, 8) recommends that small countries, in particular, need to make better use of links to external sources of technology and information. Thus, the generation of new knowledge also includes the activity of screening existing knowledge as an important step. In many cases, however, it will be required to test external knowledge via adaptive research under local conditions. Adaptive testing has to do with knowledge generation, but the transition to the next main system function ’operationalisation‘ must be fluent.

Operationalisation of knowledge

Putting research results into useful form distinguishes the stage of operationalisation of knowledge. This includes all forms of adaptive research and field trials together with individual experiments of extension workers and farmers. Operationalisation of knowledge stands under two main aspects. In the first aspect, research results need to be processed in a such a way that they become applicable for farmers. Innovations must be transformed into sets of recommendations for diffusion by the extension agent. The second aspect refers to the locality of a recommendation. Sets of recommendations must be adapted to the agro-climatic and socio-economic conditions of specific locations (Nagel 1980, 28).

Operationalisation of knowledge represents the first real test of an innovation under farming conditions. On-farm research<19> is consequently a key activity within this adaptation step. Technologies will have to be modified until they suit farmers‘ needs adequately. If this cannot be achieved or can only be achieved partly, a technology may have to be discarded or it may only be recommended for a limited group of farmers. Operationalisation is thus also a technology screening process. In this step researchers will have to work hand in hand with farmers and extensionists<20>. This process will provide useful information for the following step of dissemination. Encountered technology and communication problems between farmers, researchers and extensionists will show where improvements are required to enable a smooth transition and transformation of knowledge.


Dissemination of knowledge

Dissemination of innovations bridges the gap between knowledge generation and large scale application at farm level. It requires a double transformation of knowledge, in a first step to extension and as a second step to farmers. Dissemination takes place in view of socially defined and accepted goals which pre-structure promotional efforts. Extension strategies need to be developed accordingly (Nagel 1980, 28-29).

After a successful operationalisation process, the technology should be appropriate for diffusion to farmers. However, depending on the type of technology, some farmer specific adaptation by the extensionist may be necessary. In this respect sympathetic understanding is required to provide the flexibility to choose the right set or farm specific modification or option of a recommendation for an individual farmer. In other words, if a technology is based on scientific principles, it is important to explain these principles to enable the user to make desired modifications. It is not sufficient if the farmer learns how to apply a new practise or technology. Only if the farmer understands why and how something works, will he be able to improve his new practise.

Utilisation of knowledge

Generating new knowledge is useless, unless it becomes applied productively in some way or other. Utilisation within an agricultural knowledge system means the integration of new knowledge into the system of agriculture and its application by farmers (Nagel 1980, 29).

The ultimate goal of a knowledge system is obviously the utilisation of that knowledge. Consequently, the rate of adoption is the key indicator of performance and efficiency within an agricultural knowledge system.

Evaluation of experiences

Nagel (1980, 30) defines evaluation as the “forming of judgements on the performance of informational inputs (= knowledge) at the user level (= user sub-system)“. Thus, the overall practicality of an innovation is finally determined at farm level. For measuring the performance of the system (quantitatively or qualitatively) a comparative goal is required. Within an agricultural knowledge system, this goal is identical to farmers‘ needs identified as part of the first system function. The evaluation function, consequently, measures how well farmers‘ needs were met. Furthermore, the evaluation of an innovation reveals its strengths and weaknesses in terms of usefulness to the farmer and thus highlights areas which require further improvement and additional research. Keeping in mind the circular nature of the whole process, the evaluation revises the state of farmers‘ needs in a more concrete manner and provides as such an input to a new innovation cycle. The further developed an agriculture is and the more farmers depend on technologies, the more important is the evaluation function, as it enables quicker responses to technology shortcomings (Nagel 1980, 30).

For studying the adoption and utilisation process of new technologies in the field, participatory methods appear as a suitable tool. They could thus be the right tool for a meaningful evaluation. By handing over the stick to the farmer, they enable the users to comment on the usefulness of the new technology in their environment. Participatory methods could in this way provide a means of more client control in practise.


However, the evaluation function could be seen in a broader context. It should not only be included as the last step, but rather as an ongoing activity. In this context the term monitoring<21> might be the better terminology. Successful dissemination shows that all system components have worked. But to timely locate inefficiencies within the system, monitoring as a kind of early warning system is necessary. In particular the knowledge transformations between the sub-systems can only be monitored as they take place. These transformations are crucial for the success of the system and thus it is important to monitor and correct system flaws as soon as they occur and not at the end when it may be too late. Röling (1990, 15) lists knowledge transformations within an agricultural knowledge system at the following points:

This long list of transformation illustrates the immanent high risk of things going wrong. A way to reduce this risk is to ensure a proper documentation of results at all steps. Röling (1990, 16) speaks in this context of the storage and retrieval function of an AKIS. Rather than a separate function, this could be seen as an ongoing continuos function required in combination with the other functions. Considering the huge amounts of information that need to be processed by an agricultural knowledge system it becomes evident that good documentation structures need to be developed. Access to findings (retrieval) is equally important. It is crucial that any member in the system can find the information he/she requires quickly. Of particular importance is a common language for all groups. To ensure that members of different sub-systems understand each other, it may be necessary that crucial documents are developed jointly (e.g. research documentation, extension materials, farmer leaflets, etc.).


Figure 1 summarises the basic knowledge system functions as a cyclic flow pattern. A cycle starts with knowledge need/ problem identification and ends with evaluation. Then a new cycle starts and evaluation becomes an additional input to need identification. The permanent ongoing activities of system monitoring, storage and retrieval of information are symbolised by the central circle. In a real knowledge system information flows and linkages may be informal or formal and there may be need for a specific information centre to facilitate information flows. Counter clock-wise information flows and shortcuts may be necessary as a means of clarification or if a technology prototype proves not to be viable and a new start is required. This is symbolised by the overlapping circles.

Figure 1: Knowledge system functions

Source: own design.

Information and knowledge may flow freely within the circle and accumulate in the centre for easy access at any time. Technology development within the knowledge system is thus, never only a straight forward process. It may take several semi-loops to fulfil one full cycle. However, as Nagel (1980, 31) stresses, a crucial condition to make the system work in the long run is that “all functions have to be performed, regardless of their sequence or the concrete organisational setting.“

2.2.2 Controversies in systems thinking

It would be outside the scope of this thesis to present an exhaustive review of systems theory. Instead two new concepts of systems thinking within the school of agricultural knowledge systems will be presented here.

Within this thinking tradition, Havelock (1986 b, 77) understands knowledge systems, similar to concepts from general systems theory, as a “networks of connected entities“. This implies the notion of ’system‘ as something concrete. Checkland and Scholes (1990, 22) go a new way by questioning this character of system: can a system be taken as something that actually exists, an ontological entity, or is it a perspective, a concept or theoretical construct we use to study real life situations?

These two ways of understanding have their roots in two fundamentally different schools of thought: positivism and constructivism. Positivism is a scientific tradition that takes the ’positive‘, the existing, the actual, the undoubtedly available as the basic principle of human perception. This philosophical school is best known under Comte‘s (1979) maxim: subordination of imagination under observation.


Constructivism, in the contrary, is a philosophical tradition that conceives perception as man made. With the words of Knorr-Cetina (1980, 227, own translation): “Instead of analysing knowledge as a representation of reality, it can be perceived as manufactured by this reality.“ In the constructivist perspective any technologies or items are seen as dependent on human perception, interpretation and communication. Its social value results only via communication. The traditional epistemiological question regarding ’what is perceived?‘ is replaced by the question ’how is something perceived ?‘ Any type of cognition or perception is thus understood as an active construction of an observer, and not as a passive picture (Schwarte 1997, 24).

Among system scientists these two basic philosophical traditions have been re-labelled as ’hard‘(positivism) and ’soft‘ (constructivism) systems thinking (cf. Checkland and Scholes 1990, 25).

Röling (1994, 387) defines hard system science as such:

“a system is a limited part of reality with a well defined border“.

Implicit in the definition are the following assumptions:

Hard system thinkers use systemic images (models) to simplify the real world. With these models they try to represent the real situation. The better the outcomes of their models coincide with actual observed events, the better hard systems thinkers consider their knowledge (Kramer and Smit 1987, 117). Hard system models are hence very suitable for physical or bio-physical environments. However, in relation to social actors and the processes of communication, problem solving or negotiation they remain weak. A main reason for this weakness is that social actors do not, as assumed by hard systems, behave according to set principles or laws.

Experiencing difficulties in working with hard systems in social contexts Checkland (1981), was the first to use this as a starting point to develop “soft systems“. In a more recent article, Röling (1995, 9) describes soft systems as “...a network of social actors, that accept a joint problem and go in dialogue through a joint learning path, in order to reach joint action.“ He further specifies that, unlike hard systems with clearly defined borders, soft systems have ’arbitrary or subjective‘ borders as the interpretation of the problems by the individual actors may influence the constitution of the system.

As a consequence of the mutual learning and negotiation process, a double sense exists: the actors influence each others realities. The motive power for the different actors to participate in the process, is the prospect that via the building of the system problem


solving possibilities emerge, which none of the individual actors could possibly realise Röling (1995, 9).

Soft systems thinkers do not perceive the world as a single system or try to develop parts of the whole into representations of the whole reality. Soft systems thinking constructs systemic images or in other words perspectives or windows of the world in order to stimulate reflection and debate. An important aspect of defining systemic images is the explicit mention of its purpose. Soft systems do not have a purpose, but are actively given one. A description of any purposeful whole must be from some declared perspective. Multiple images are thus a means to construct different images of the same situation (Checkland and Scholes 1990, 24-25). These images “are means to an end, which is to have a well-structured and coherent debate about a problematic situation in order to decide how to improve it. That debate is structured by using models based on a range of world views to question perceptions of the situation“ (Checkland and Scholes 1990, 42). Further the authors emphasis that “it is wrong to see soft systems methodology (SSM) simply as consensus-seeking. That is the occasional special case within the general case of seeking accommodations in which conflicts endemic to human affairs are still there, but are subsumed in an accommodation which different parties are prepared to ’go along with‘“ (Checkland and Scholes 1990, 29).

Engel (1995) summarises hard and soft systems thinking by comparing its main characteristics. These are depicted in Table 4.

Table 4: Characteristics of hard and soft systems

Hard systems thinking

Soft systems thinking

The world is systemic... or can be taken as if...

The world is not systemic...but sometimes it is useful to take it as if...

Images are to be systemic ...

Images are systemic when useful ...

System images are used to construct models to represent (parts) of the world ...

System images are used to construct windows to study the world ...

System images are concerned with processes, inputs and outputs ...

System images concern social actors, their activities and relationships ...

The aim of hard systems thinking is to improve one‘s knowledge of the world through improving one‘s models...

The aim of soft systems thinking is to improve human performance through debate and reflection ...

Processes are functionally articulated into a goal-seeking whole ... goals are inherent to the whole

Social actors might behave as a systemic whole if they wish to and know how to do it ... but boundaries and goals are permanently (re)negotiated

Source: Engel 1995, 30.

Both hard and soft systems thinking have specific advantages and disadvantages. Both directions of thinking may be useful in advancing innovative processes in agriculture. The key is to apply each approach in the domain were it works best. Openness to both approaches is necessary. As a rule of thumb hard systems are more suitable in situations that are better known, where variables are predictable to a reasonable degree. Soft systems are certainly more appropriate to explore new terrain, in complex situations that require iterative approaches or where decisions depend on negotiation processes. Therefore,


depending on the purpose, a prior detailed analysis of the best systems approach, case by case, is necessary. For these reasons, such a flexible approach in the use of the term system is adapted for this study.

This digression into systems thinking should be sufficient to better understand the new concepts of knowledge systems, which will be presented in the following chapter.

2.2.3 Platforms or theatres of agricultural innovation

By introducing the soft systems ideas into their conceptual framework, Röling and Engel further developed agricultural knowledge systems thinking. The influence of soft systems thinking is reflected in the new definition of a knowledge and information system (KIS):

“The articulated set of actors, networks and/or organisations expected to work or managed to work synergically to support knowledge processes which improve the correspondence between knowledge and environment, and/or the control provided through technology use in a given domain of human activity“ (Röling 1992, 48).

Comparing this definition with the older definition (see page 17) a few changes strike the immediate attention. The concept is widened to a knowledge information system (KIS), which encompasses any domain of human activity. Agriculture is no longer seen as the only area where knowledge perspectives could be useful. This opening of the approach thus creates room for including other areas such as health care, traffic and transport, rural or urban development, education or environmental concerns etc. All members in the system are considered as actors. In an agricultural context, actors could be e.g. extension agents, private consultants, commercial firms, agricultural schools, growers clubs or associations, co-operatives, Ministry departments or research stations just to name a few. All these actors manage, generate, transform, transmit, store, retrieve, integrate, diffuse and use knowledge and information. The definition summarises these activities as knowledge processes geared towards improving ’correspondence‘ between knowledge and environment and/or ’control‘ through technology use. This also needs to be understood in the context of soft systems. Here, boundaries are not fixed and depend on the perspective of the analyst, and therefore, are bound to vary with the function or purpose he has in mind for the system to perform. In a setting with several actors, each with different perspectives and understanding of what e.g. ’correspondence and control‘ may mean, a struggle and debate on perspectives and definitions is pre-programmed (Engel 1995, 37). However, this debate is seen as a positive prerequisite to develop innovations.

Looking for a metaphor to better describe such a multi-actor environment, Engel created the expression ’theatres of agricultural innovation.‘ Engel chooses theatres, as they are places where partly pre-mediated, partly improvised actions are performed. Analogue to AKIS, different actors such as directors, managers, designers, stage builders, actors and audience interact intensively to produce both structure and serendipity (Engel 1995, 8).

It is easy to agree to the example that innovation processes in agriculture often resemble very well a drama in theatre, but the term theatre also creates undesirable associations: Actors in an AKIS may not like to be compared to acting on stage. They certainly take


their work serious and may not like to be taken as play-actors or comedians. Therefore, a different term appears more neutral: platforms for innovations. The term circumscribes, just as theatre, a space were people may come together to work and discuss. The term is less freighted with associations. Röling (1995) introduced the term and suggests platforms as a means, forum or stage for different social groups to negotiate resource use. A platform as a means to improve discussion and facilitate co-operation could also be perceived as an appropriate means in an AKIS. This point will be further discussed a little later. Beforehand, it is necessary to have a closer look at the working procedures in an AKIS.

Given the fact that all actors in an AKIS have a different specialisation (extension, research, financing, policy, etc.), they all have different ideas and information to share with the other actors. But these different backgrounds may make it more difficult to communicate and, in this respect, some kind of co-ordination is necessary to facilitate the exchange of ideas for the joint learning process. Though, in many cases it will be difficult to achieve a consensus, it is necessary that at least the main actors achieve some kind of strategic consensus. The joint understanding of a common problem and the joint purpose to develop relevant innovations may help to achieve this consensus. Engel (1995, 71), based on empirical studies at a number of countries in both Central America and the Netherlands, claims that working procedures evolve automatically in the course of the working process. Depending on the purpose, the actors decide on how they want to work, what to do, how often to meet, which resources to use etc. In this process, inter-institutional relationships are build. These are a result of negotiations over objectives, tasks and resource allocations, which never remain static. In practice the content and shape is continuously readjusted. Some necessary readjustments are sealed in formal agreements, while others may be on an ad hoc basis and remain largely informal (Engel 1995, 92). This social interaction Engel calls “networking“. Thus, to innovate practices, actors experiment, gain access to a range of options and insights, engage in building and managing interactive relationships with those others whom they consider relevant to their concerns. The keys to membership are relevance and competence. This means, actors include those members in their network that are perceived as relevant: those that are able to offer something new and useful in a competent manner to the network or AKIS as a whole.

Summarising the new Engel and Röling understanding of an AKIS, it becomes evident that the view diverts away from an AKIS with a single fixed perspective of actors and institutions in a given country, to an AKIS with multiple perspectives and potentially more complex actor relationships and linkages. More emphasis is placed on the importance of social relationships between actors and new forms of learning. Developing innovation is no simple process of generating technologies and packaging an extension message with the help of an elaborate system, but a process of enabling joint learning and linkage building within a system that requires constant improvement, adaptation and fine-tuning. It thus requires that all actors engage actively in the process. This active participation is the basic prerequisite for the new direction that Engel (1995, 51) outlines as “knowing rather than knowledge, competent performance rather than use of new technologies and communicative interaction rather than communication as the transfer of messages between sender and receiver.“ This sounds good but what does it mean in practice. ’Knowing rather than knowledge‘ stresses that knowledge transfer requires


active ongoing reflection, interpretation and practise; ’competent performance rather than use of new technologies‘ draws attention to the fact that understanding and consequent competent management are a prerequisite to successful technology usage; and lastly ’communicative interaction rather than communication as the transfer of messages between sender and receiver‘ stress the importance of dialogue and feed back as opposed to a one way communication process.

2.2.4 Viewing the Fiji AKIS

The examination of the Fiji AKIS will proceed in three main steps. Each step contains a specific view, window or perspectives on the AKIS. The first step constitutes an overall view. It presents the main organisations and institutions and other social actors within the agricultural sector. Their main relations and linkages are describe briefly. This first perspective is similar to the analysis by Blum and Roux of the Swiss AKIS (cf. ROUX and BLUM 1994, Blum 1993). This perspective is the most general and in a way ’institutional‘ view. It depicts the current organisations and institutions as they appear at a first glance. The investigation touches the surface, it does not penetrate deeply into the subject-matter. At the same time the view is wide, including to some extent the outside macro system surrounding the AKIS. The purpose of this first perspective is to give an overview, characterise the wider work field and to point out the subject of the study. Its intention is to provide for a better understanding of the following closer analysis of the core segments of the AKIS.

The second step represents a perspective, which examines the three main AKIS actors,<22> farmers, research and extension, one by one as a close-up.

The view provides a detailed analysis of each actor. The investigation does not remain on the surface, but penetrates deeper into the subject-matter. For research and extension the view will include a look at policies and goals, concepts and methodologies used, priority setting procedures, availability of resources, staff qualifications and the use and flow of information. For farmers, the focus of analysis lies in their problems and constraints, production conditions and management practices. Special attention is given to the question, how farmers and their problems can be better addressed within the AKIS.

After looking at each actor individually, the interplay and the linkages between all three actors will be examined. This view investigates how the different AKIS functions are performed, which linkages exist and how information flows are organised. This analysis provides the basis for the formulation of strategies to improve the performance of the AKIS.

Viewing the Fiji AKIS only in the context of the three main actors would give an incomplete picture of reality. The private sector, donors and educational institutions all play an important role in agricultural development. However, examining all potential


actors in depth would be outside the scope of this thesis. Nevertheless, in the third and last step a wider look on the AKIS will be taken again.

In this perspective, the researcher will try to investigate how the AKIS could further evolve. Checkland and Scholes‘s (1990, 45) soft systems approach will be used to look at the knowledge system not as it is, but rather as a verbally rich picture of what could be. Focusing on the newer AKIS concepts suggested by Röling 1992 and Engel 1995, the idea of platforms as supporting structures for working groups or networks will be taken up.

Figure 2 visualises a model of how platforms could be viewed as constituted by different actors.

Figure 2: Composition of platforms

Source: own design.

Actors from different backgrounds (research, donors, policy, market, farmers, etc.) create new or join existing platforms. Members of a platform share a common goal and purpose. Platforms may overlap if some members are also members of other platforms. In the course of their work, all actors agree on joint working procedures. This includes the meeting sequence, information exchange protocols, resource sharing, joint tasks, separation of tasks and other emerging points which may appear necessary.

Figure 3 illustrates the structure of platforms and their linkages within an AKIS. Within platforms all actors are closely linked and collaborate intensively. Different platforms may be linked formally if there are interesting areas for co-operation. Besides these formal linkages, individual actors may maintain informal links to actors in other platforms. Platforms may overlap if key actors are members in two platforms. This strengthens both formal and informal linkages. The innovative direction of each platform is determined by the needs and problems expressed by the resource base (’reality‘) composed of producers (technology users), consumers and the market.


Figure 3: Platform model

Source: own design with ideas from Engel (1995, 203), Roux (1995, 19), Röling (1995, 12).

Equally formal and informal linkages exist between actors in the platform and the resource base. The platforms, thus, address a specific group (in other terminology: target group or recommendation domain) of users in the resource base. Depending on the practises and technologies developed and extended by the platforms, the user groups of the various platforms may overlap more or less.

But an AKIS would not be complete without a co-ordinating centre. This could be perceived as yet another platform with links to all other platforms. The tasks and role of the co-ordinating platform, or in other words AKIS management, are manifold: avoiding duplication of efforts, balancing financial resource allocations and co-ordinating policy and user interests. The management plays a crucial role in enhancing system coherence


and system synergy. Given the soft system character, a management of an AKIS will never execute a strong control as in a strictly hierarchical system. The actors in the different platforms are in the end responsible to their institutions or organisations. As a consequence overall consensus in all cases is unlikely. In reality this may mean that in extreme situations platforms split up in order to reflect fundamentally different approaches or actor profiles. Examples may be incompatible donor policies or private sector company goals. Co-ordination thus becomes a difficult balancing act.

The applicability of this model for Fiji will be the central theme of this last perspective. The following questions will guide the discussion: What kind of networks or platforms already exit? How does co-ordination take place? How can platforms be supported and assisted? Which leadership patterns exist? How can platforms become driving forces for innovation?

This approach of investigating the Fiji AKIS with multiple perspectives should provide for a comprehensive understanding of all actors, their strengths and weaknesses and the associated problems of co-operation. But before applying the concept, a few critical remarks on knowledge systems thinking shall be made.

2.2.5 Some criticism put forward

The main strength of the concept, the review of several actors as part of a system, could also be seen as a major weakness. The fact that several key actors and their interrelation are analysed, means in practice that each actor by himself cannot be reviewed as profoundly as if reviewed alone. For this reason specific weaknesses of individual actors may not become apparent in the AKIS approach. The AKIS concept has the advantage that it provides a better overview and highlights the relations between different actors, but at the same time some degree of detail of the situation of individual actors is lost. This is an immanent weakness of the concept.

The earlier AKIS concepts (cf. Röling 1988 and 1990) have a strong normative character. The focus lies on the questions of generating, extending and utilising of knowledge with the aim of increasing system synergy. This also depicts the limits of the concept. The behaviour of actors and their reasoning in the processes of interaction are neglected (Lühe 1996, 12). While this critique is certainly valid for the older AKIS concepts, the newer versions in particular of Engel 1995 with the inclusion of soft systems thinking towards actor behaviour have reduced this problem.

Leeuwis (1993, 285) remarks that there is a confusion between “on the one hand, the KIS perspective as a quite legitimate and popular practical tool for enhancing learning and/or developing collective agency and identity, on the other hand, the expectation ... that this perspective would provide a theory to understand and explain adequately the role of knowledge in processes of social change and intervention.“ This confusion has some roots in the fact that AKIS attempts to make a holistic picture of a given situation. Engel (1995, 33) opposes that “when we, (including Röling) emphasise ’wholeness‘ we definitely do not entertain such a unified theory aspiration. Rather we refer to the fact we probe for coherence among the events and ideas which appear relevant to our inquiry. We refer to the need for gaining more comprehensive ways of debating, not a unified


scientific theory for understanding. We see the need for doing so because of the ever growing degree of complexity involved in the debate on social and technological issues.

A second point of criticism by Leeuwis (1993, 286-287) aims at Röling‘s and Engel‘s definition of information. He argues that their information concept stress the individual-cognitive aspects of information, but fails to conceptualise the social dimensions, i.e., the normative, political and interpretative struggles that accompany and shape the production of knowledge and information in communication. He explains his argument by the example of software developments for farming: “from a KIS perspective the key word in the explanation of limited effectiveness and adoption would be lack of ’anticipation‘. That is, communication technologies can fail to foresee (or anticipate): (a) farmers‘ information needs and/or interpretative frameworks, (b) the practical procedures of problem solving that farmers and extension workers employ in their interactions, (c) the information that farmers already get from other KIS interactions (e.g. farmers‘ journals) and/or (d) the economic, material, political and cultural constraints and opportunities that characterise the KIS environment. By implication, using a participatory systems methodology and/ or more ’traditional‘ extension tools such as targeting and user research, such anticipatory problems can be prevented“. Leeuwis (ibid.) acknowledges that the KIS perspective is quite helpful in revealing that different types of anticipation problems exist and what these types are, but it fails to explain why they have come into being. Returning to his example, he concludes (op. cit., 287) that “following the KIS framework we would in the end merely have to assume that producers and developers of communication software are quite ignorant, and that, by using soft systems methodologies etc., we can facilitate the illumination, joint learning and consensus that is needed in order to prevent and cure such ’mistakes‘. Here we seem to have found a relic of the optimist ’enlightenment‘ thinking that characterised the early days of extension and extension science.“

Leeuwis (ibid.) highlights a weak point in the soft systems methodology as part of an AKIS concept. The perspective helps to identify problems, but it does not necessary reveal the reasons for problems. When Leeuwis mentions the lack of consideration of ’social dimensions of actors‘, this critique points to one of the main assumptions of the AKIS perspective. Namely, that all actors have a common interest and work towards a joint goal (because everybody realises that the goal can only be reached jointly). If this consensus of a joint goal is lacking, the system will fail. The joint learning cycle will not even be started. In other words, if important actors do not want to cooperate, for whatever reasons, learning processes may easily be sabotaged.

Leeuwis makes clear that SSM is no blue print for problem solving that will work in all circumstances. However, in criticising soft systems methodologies as ’optimist enlightenment thinking‘, Leeuwis goes too far. It is certainly valid to comment that the approach is no blue print that works automatically in all cases. However, this is also not stated. AKIS and SMS are pragmatic concepts. Their validity could be confirmed in several case studies. However, in a complex environment with many actors, innovation development and diffusions remain a very difficult task and success can never be guaranteed. But even if soft systems methodologies may not always lead to the identification of solutions, they certainly enhance debate and struggle towards finding solutions. Therefore, overall the AKIS concept appears as a very good theoretical framework to assess the conditions in Fiji with.



Constructivist and positivist viewpoints are compared in chapter 2.2.2 .


’Transfer of technology‘ (TOT) is a synonym used in much of the development literature (cf. Chambers et al. 1989; Röling 1990) to describe these ’linear‘ development steps based on the diffusion model.


cf. The training and visit system by Benor and Baxtor 1977 in the original form and as a revised version by the same authors Benor and Baxtor 1984. A recent critical review if the approach at the example of TV system in Benin is found in Lühe 1996.


The term technology itself requires further definition. The term technology is often used as synonym with technical item, technique or (less often) practice. According to the German encyclopaedia (Brockhaus-Enzyklopädie 1993, 672) the meaning of technology in the 19th century referred to the science of the development of technics. Today the term includes the totality of knowledge, ability and possibilities of a technical sector. Problems may arise as many people use and understand the term only in a more narrow sense as technique or some technical item. Two examples shall be used to illustrate the problem: If a person buys a radio, this may at first appear as all he/she needs to do in order to use it. However, the technology radio implies more. To use the technology radio it is equally important to have electricity to run it and furthermore the existence of radio stations and their programmes are necessary to listen to the radio. The later requirements are often taken for granted. If a farmer wants to use the technology fertiliser it may at first sight appear as simple as distributing the material on his field. But technology implies more. To use the technology fertiliser, fertiliser factories and distribution centres are required outside the farm. The farmer furthermore needs means of transport to the farm, a fertiliser castor (optional) and a crop variety that responds to fertiliser. In all cases knowledge to apply and use the technology are required. These examples illustrate that ‘transfer of technology‘ may be a difficult job. In agriculture, some of these problems can be avoided if the search for innovations is focused on simple practices, instead of technologies. Blum (1991, 323) points at a number of practices that may have enormous impact: changes in crop rotation, timing of sowing, renunciation of harmful activities such as burning or overgrazing.


The model is also included in a training manual for staff of plant protection services in Fiji and other South Pacific nations cf. SPC-GTZ (1993, 350).


Gremmen understands professional activities just as farmers activities as practises.


It is vital not to confuse an AKIS with a management information system. The former comprises the entire knowledge flows in agriculture, while the later generally collects and monitors output oriented indicators for better control of single institutions or companies.


Blum (1994) and Blum and Roux (1994) include a large number of agricultural institutions within their agricultural knowledge system of Switzerland. They also highlight the fact that other knowledge systems with relations to agricultural knowledge systems (e.g. forestry, ecology) exist. Röling (1990, 20) also stresses that AKIS have to be seen as part of larger systems: policy, environment, structural and market conditions, international agricultural community etc. Concepts of agricultural knowledge systems were subject of a recent OECD Conference held in Paris. Carlson (1995) outlines a number of systems concepts presented at the conference.


As a broader term for PRA approaches Pretty et al. (1995) suggests participatory learning and action (PLA).


Chambers (1994, 954) identified five streams that have cross-fertilised each other, intermingled and exchanged elements to various degrees, stand out as sources and parallels to PRA: activist participatory research (APR), agro-ecosystem analysis (AEA), applied anthropology, field research on farming systems (FSR) and rapid rural appraisal (RRA).


Without innovations a knowledge system would only contribute to maintain knowledge. This function, however, could certainly be fulfilled by farmers alone without external support. A special case of knowledge maintenance are gene banks that conserve genetic resources that might otherwise get lost.


Eyzaguirre (1996) describes how research could be organised in small developing countries.


Werner (1993) presents a comprehensive guide to on-farm research with a strong participation of farmers.


This also implies a broadened role for all actors. Researchers will need more social and communicative capabilities and extensionist will require more analytical capabilities.


Krimmel et al. (1990, 7) define monitoring as an ongoing activity during the implementation of a given project. Monitoring compares activities against a plan and suggests corrective action in case of deviations from the plan. In contrast, evaluation is a periodic activity , focusing on major changes or replanning of projects.


The researcher will use the term actor in the wider sense as proposed by Engel 1995. It therefore stands for all institutions or organisations or persons sharing this work field. In the Nagel 1980 nomenclature, it would correspond to the term sub-system.

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiML DTD Version 2.0
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Thu Mar 29 16:23:51 2001