Associations of multi-agent systems and role-playing games (RPG) have shown their relevance to tackle complex and dynamic social systems sharing common resources. Now, some are used in participatory processes as group decision support tools to promote information exchange of between stakeholders. In a RPG, stakeholders are placed in a virtual world where roles are allotted and rules are defined. In this approach, a question arises: do they adhere to the rules given by the game or do they use parts of their own reality? This article focuses on the link between play and reality in negotiation processes. The research was conducted in irrigated systems of the Senegal River valley. A methodology is proposed to test how reality is brought into the game. Qualitative interviews about negotiation processes in reality and in the game allowed us to analyze interactions and behaviors of participants.Results showed in the case presented that (1) stakeholders have accepted the schematic representation of their reality, (2) the social background of players interferes with roles playing in the game, (3) the game reveals to observers social relationships between players.

Keywords:

Role-playing game, Social reality, Senegal, Irrigated system, Conversational analysis

Introduction

1.1

Experiments which associate multi-agents systems (MAS) and role-playing games (RPG), are increasingly numerous. They can be used together to tackle complex and dynamic social systems sharing common resources (Bousquet et al., 2002). When villagers of the same social community play a game, what are they doing in reality? Are they just following the rules or are they using their reality to play the game? Though these questions are of crucial importance when using RPG to improve the local management of resources, there is no methodology for answering them. In this article our main purpose is to propose a methodology to determine whether or not there is a relationship between the role-playing of actors and their social reality. The relationships between play and reality can be seen in two ways: on the one hand, how reality is brought into the game; on the other hand, how a game may influence real life. This research focuses on the first aspect.

1.2

A role-playing game was designed to open the black box of a multi-agent model and explain its contents to farmers of the irrigated systems of the Senegal River valley. As the research progresses, the game will be used as a group decision support tool with farmers. It is assumed that whether they accept the rules of the game as they stand or whether they invent new ones. The question is: to what extent does social reality influence players' interactions for resolving problems encountered in the game?

1.3

In order to shed light on this issue, we broke it down into three questions:

• For players, what is the reality represented in the game?
• If the players accept this representation of reality, will they use their real behaviors, norms and relationships in the game?
• In that case, will the game reveal these elements to observers?

1.4

To tackle these questions three hypotheses must be tested.

Hypothesis 1: The game is accepted as a schematic representation of reality

1.5

The game was designed as a schematic representation of the real management of Senegalese irrigated systems. But though it was built using a companion modeling approach (Bousquet et al., 1999; Bousquet et al., 2002) the game is partly an expression of the researchers' point of view. So, the first question was to validate whether farmers accepted this point of view or not. Barreteau et al. had already presented the RPG to them and the interviewed stakeholders had partly validated the proposed representation (Barreteau et al., 2001). The characters presented in the game are caricatures of real roles, as farmers have already recognized. The proposed rules are not exhaustive. So what do they accept exactly: the rules, the types of roles or both? This research tries to take model validation one stage further. The aim is to analyze behaviors of farmers in situations of action where they are obliged to interact. Negotiation processes in the game and in reality represent such situations of action. This analysis should be helpful in taking us beyond the first validation step already completed. Because people in action are dynamic, the analysis of their interactions should encompass all elements (about norms, behaviors, rules really followed in negotiations,...) not given in the first validation. And these elements are essential for understanding how reality is represented in the game.

Hypothesis 2: The social background of players interferes with role playing in the game

1.6

For Huizinga (1951) and Caillois (1967), the game is detached from real life. It is assumed that players abandon their social background before playing. Our second hypothesis is opposed to this assumption.

1.7

The roles in the game are circumscribed by formal rules. The characters played are not real because their actions are limited by the game. We assume that the actions and behaviors of players are under control. In this research, players are members of the same village. They share values, norms, habits and history that determine their behaviors and relationships in real life. All these elements structure their actions. And this may interfere with their behaviors caricatured in the game. How do players use their social background to act in the game? May the real social or economic hierarchy be reproduced or changed in the virtual environment of the game?

Hypothesis 3: The game reveals social relationships between players

1.8

If the second hypothesis is accepted, may the RPG be useful as a tool to investigate social reality? The challenge is to identify the elements of players' social background used in the game. Is the game helpful to reveal these elements to observers?

1.9

This paper aims to present a methodology for analyzing interactions between players involved in a RPG and to test the three hypotheses presented. A specific RPG called Njoobaari ilnoowo serves as the support for this research. It was built as a schematic representation of the real world. The methodology is based on complementary analyses of virtual and real negotiation processes. Farmers are embedded in the same real social system. In particular, they are faced with the challenge of finding loans which are essential for crop production on their plots. Together, they face the seasonal problem of sharing insufficient rural loans. Relations between the RPG and reality are analyzed here for this specific collective process. The methodology consists in characterizing situations where people are interacting to solve common problems.

1.10

This paper begins with a presentation of the context of the research : negotiation processes for collective loan finding in an irrigated scheme and a description of the model underlying the RPG. The second part describes the methodology used to characterize negotiations in the game and in reality: qualitative analysis of game sessions, ethnographic analysis of a specific collective loan finding process. The last part addresses each one of the three hypotheses, through examples provided by results of these analyses, in order to test the relevance of the proposed methodology.

Context: the Njoobaari ilnoowo role-playing game and collective loan finding processes

2.1

This section presents the real and virtual contexts on which the methodology was tested. It describes collective loan finding processes as the context of real negotiation. We then present the RPG used to support the research.

The real context: negotiation processes for collective loan finding in an irrigated scheme

2.2

It is important to characterize real negotiation processes in order to obtain information about farmers' behaviors in reality. So, in the real world, we decided to focus on negotiations dealing with rural loan management in irrigated systems. There are two reasons for this. Firstly, access to loans is necessary for crop production in the irrigated systems in the Senegal river valley. Secondly, because available loans are not sufficient to cover all needs, the management of these collective loans is a subject of seasonal negotiation. The next two sections show that loans are central to the management of these irrigated schemes and are the topic of recurrent disputes and negotiation processes. For greater clarity, we present the example of Wuro-Madiu village, located in Podor department, in the middle of the Senegal River valley. Wuro Madiu was chosen because the companion modeling approach to build the MAS and the RPG was partly developed there.

Brief history of irrigated agriculture in the Senegal river valley

2.3

Irrigation in the Senegal river valley increased with the independence of Senegal in 1960. This political choice was made to support the economic development of the country by generating income from agricultural production. Irrigation, mainly financed by international donors, was based on modern hydraulic infrastructures and on the training of farmers. Everything was mechanized. Primary, secondary and tertiary schemes were built for even greater control over water supply. After the droughts of the 1970s, the Senegalese government built more tertiary irrigated schemes. Valley development evolved differently from one region to another, depending on population density. In the sparsely populated delta, big irrigated perimeters (BIP) were constructed, while in the densely populated middle and the upper valley, smaller community irrigated perimeters, intended for small groups of farmers, were built in association with the BIPs (Boivin et al., 1995; Diemer and van der Laan, 1987; Crousse et al., 1991). Between 1976 and 1996, irrigated areas increased tenfold, particularly with the development of private irrigated schemes (Liagre, 1997)

2.4

Nevertheless, the results did not meet the expectations of the funding agencies and the Senegalese government. Production is very expensive because of over-mechanization. Farmers thus became dependent on crop loans to produce in irrigated schemes. Yields stagnated and infrastructures rapidly deteriorated. The cost of their rehabilitation was high. Faced with these relative failures associated with economic difficulties, the Senegalese government was forced to withdraw from agricultural activities. The functions of its rural management entity (SAED^[2]) were transferred to others players: hydraulic management was transferred to farmer organizations, financial management to the national bank of crop loans (CNCAS), and tenure management to rural communities. The remaining activities were placed in the hands of private enterprises (Crousse et al., 1991).

Loans are the topic of recurrent negotiations

2.5

Because of political choices, farmers are now financially dependent on outside loans. The cost of mechanized production is so high (about 330,000 Fcfa /ha of rice) that farmers need loans to produce in irrigated schemes. As a general rule, farmers' organizations (EIG: economic interest groupings), with the help of the SAED, express their needs to the CNCAS (stage 1 to 5). The bank only gives loans for rice and tomato production. (figure 1). Loan allocation is dependent on the repayment of loans contracted by the EIG during the previous season. Thanks to the vouchers given by the bank, the president of the EIG buys inputs and distributes them to his members (6 to 10). At the end of the season, the EIG has to collect rice to pay back the loans to the bank (11 to 12). The bank does not recognize individuals. So, in a EIG, all members are collectively responsible for the loans obtained. They are dependent upon each other.

Figure 1. Financial organization of rice seasons in the Wuro Madiu EIG (1997 to 2000)

2.6

Moreover, with the high costs of production in irrigated systems, some farmers have developed strategic behaviors with regard to loan repayment. They now tend to allocate loans to produce different crops in their different plots. This risk-spreading strategy is partly responsible for low yields in rice production. With low yields, all the loans obtained are not paid back to the bank. But for the CNCAS, the EIG is still responsible for loan repayment. When one of the members does not pay, all the others have to repay his debt if they want to receive new loans. Otherwise, all members of the EIG are penalized.

Figure 2. Self-financing of rice seasons in the Wuro Madiu EIG since 2001

2.7

This is what happened to the Wuro-Madiu EIG in 2001. Because its members were not able to pay back their debt of 6,000,000 Fcfa, the CNCAS refused to grant all the loans needed by the Wuro-Madiu EIG (figure 2). The bank gave them loans according to their reimbursement rate. How did the farmers of Wuro-Madiu manage loans obtained to produce rice in their irrigated scheme?

2.8

The villagers devised two ways to allocate the funds obtained. In order to maintain social cohesion, some preferred to find a way to distribute CNCAS vouchers to all farmers. Farmers would be able to produce but only in a part of their plot, depending partly on their reimbursement rate. On the contrary, others (called the 100% group) had made sacrifices to repay all their loans. These people wanted to keep the credit to cultivate their entire plot. Several formal or informal negotiation meetings were held to solve the problem. The figure 3 shows how some farmers called in a facilitator (AB) to help them (stage 1 to 2). He met men of influence in the village (the head of the village, the imam: religious leader) and succeeded in convincing them to support his point of view (3 to 6). They authorized him to form a small group to think about a solution to the problem. After several group discussions (7 to 8), the first proposition was presented to all the farmers. The majority voted for the proposition (9).

Figure 3. Spatial dynamics of the negotiation process in Wuro-Madiu

2.9

Loans thus lie at the center of negotiation processes between members of EIGs (farmers), villagers and others members of the social community living outside the village (migrants, NGO^[3] representatives,...). Exchanges are all the more important because, with the high cost of production, a minimum number of farmers producing in the irrigated schemes is necessary. With each new season, the issue of loans re-emerges and villagers must conduct negotiations to solve the problem. Access to loans and their management are central to the irrigated schemes. They underpin certain social interactions between stakeholders.

The virtual context: The Njoobaari ilnoowo RPG

2.10

Mucchielli (1983) defines the role-playing game as a construction of a problematic situation in which people are acting given parts. He assumes that a RPG is formed by three elements. The game comprises a system of specific rules. It also describes the world where the session takes place. The animator, also called game-master, organizes the session. He knows all the rules. The players are the people taking part in the game. Each player makes his own role by following the rules (Mucchielli, 1983). For us, there are also observers of the game. They do not participate in the game but they are present and they note all data that is important for analysis of the session.

2.11

The Njoobaari ilnoowo RPG was built using a companion modeling approach developed with farmers of irrigated schemes in the Senegal River valley (Barreteau et al., 2001). This RPG is extracted from the Shadoc MAS (Barreteau and Bousquet, 2000). It is formally a MAS whose agents are not computer entities but human beings in a given environment interacting among themselves according to a set of rules. Inasmuch it complies with current definitions of MAS, such as the one given in (Ferber, 1999).

2.12

Initially SHADOC was designed to gain a clearer understanding of co-ordination patterns among farmers in irrigated systems and their consequences. The aim was to use simulation as a means to understand the sensitivity of a virtual irrigated system to various sets of rules. It was designed on the basis of field interviews, field observations and discussion of the results of earlier versions with key local stakeholders. The next step was to explain the game's content to farmers in the villages. We thus "translated" the MAS into a role playing game named "Njoobaari Ilnoowo" by its first players. This "translation" involved simplification to make the game practically possible to play (Barreteau et al., 2001).

Introduction to Njoobaari ilnoowo

2.13

The game is played with 10 to 15 players. Players take the roles of farmers in an irrigated scheme in the Senegal River valley. Rice is the only crop cultivated in this scheme.

2.14

Space is divided into two areas. One represents the irrigated scheme and the other, villages. Because in reality the villages and irrigated scheme may be several kilometers apart, the two areas are hidden one from one another. In the first area, players are divided into two villages several kilometers apart. Due to a long-standing dispute, there is no communication between villages in this zone. The second area represents the irrigated scheme where players are organized into two farmers' organizations (EIGs: economic interest groupings). People from both villages can be in the same EIG. The irrigated scheme is drawn on a blackboard. When players are in this zone, they can communicate. Each EIG is in charge of managing a watercourse. Each player has only one plot. One pumping station supplies water to every watercourse. An abacus is used to allocate water among the open plots. The allocation depends on the number of plots along a watercourse opened at the same time and their relative position. Everybody knows this abacus. A water evaporation factor is also applied.

2.15

Players follow the rules given in the game. At the beginning, farmers draw three cards at random. These cards written in Pulaar, Wolof ^[1] and French define their basic behavior during the session. Three parameters characterize each player:

• a social status, (one of four), that determines the possibilities of exchanging labor or credit (figure 4);
• a plot production goal, (one of three), with an increasing level of intensification (figure 4);
• a loan repayment rule, (one of four) (figure 4)

Figure 4. Njoobaari cards

.

How the RPG is played

2.16

The game covers several successive cropping seasons. During one season, three stages take place: search for credit, irrigation/production activities, appraisal of the season (figure 5). In the first season, it is assumed that loans are granted by the bank to cover all purchases of inputs.

Figure 5. The Njoobari ilnoowo game

Stage 1: Search for loan

2.17

To cultivate rice in the irrigated scheme, farmers need loans. In this first stage, we assume that they all obtain the loans they need. They receive different loans according to their goal card and will be able to reimburse the loan during the appraisal stage (stage 3). So at the beginning of the game, this first stage is not played.

Stage 2: Irrigation and production

2.18

Before starting the game, each player receives the cards that define his role. They are distributed randomly in the villages and in the farmers' organizations. They choose (i) a chief for each farmers' organization, (ii) a pumping station manager and (iii) a person in charge of loans. In the scheme area, the chief and his group define the rule for water allocation along their watercourse. So these players have two roles: a collective one and an individual one. The game then begins.

2.19

In the village area, the players belonging to the first farmers' organization draw an opportunity card at random (figure 4). Depending on the combination of the card drawn and their goal card, they may or may not be able to go to the scheme area. In the scheme area, a board is drawn with the position of each player's plot. Each plot is initially water-deficient because the soil is totally dry. This deficiency is represented by a negative water level. When all the pores of the soil are filled, the water level reaches "0" which represents the soil saturation. Farmers decide to open or close their pipe by following the collective rule of water allocation in their watercourse to irrigate their rice. To sow one of the two varieties of rice (long or short), they must pay for inputs according to their goal card. They then leave this place to return to the village area and it is then the turn of the members of the second organization who are allowed to go to the scheme. At each turn, the water level and the state of cultivation of each plot are computed according to an abacus (figure 6). Everybody knows this abacus. This is repeated about eight times in a season until the rice is ripe enough to be harvested.

Figure 6. The abacus

At the end of the season, the station manager stops pumping.

Stage 3: Appraisal of the season

2.20

When the rice is harvested, farmers sell their products and receive the equivalent in virtual money in the form of paddy bags (figure 1). This equivalent depends on their goal card and the number of stress events suffered by the crop. They draw a repayment card at random (one of four) (figure 4). Each card describes a different behavior: reimbursement of collective creditors (bank and pumping station), individual farmers, both or none. Managers of the bank and of the pumping station sit and wait to be reimbursed by farmers. The players in charge of the bank and the pumping station select a loan rule from among a choice of three. Each one can choose a different rule. Then, if allowed, the players reimburse their loans.

Stage 1(2): Search for new loans

2.21

At this stage, farmers can change their repayment and goal cards. If necessary, farmers can get into debt with individual or collective creditors. At the end of this stage, farmers are ready for a new season.

2.22

An average session comprises two cropping seasons and is played in about a half-day. The number of players is between 10 to 15 persons. With less than 10 participants, the game does not reveal the complexity of water management because the farmers do not need to interact to work on their plots. With more than 15, it is difficult for the animator to manage the game and observers cannot follow the exchanges between players.

2.23

Now that both virtual and real contexts have been described, the next section presents the methodology developed to analyze the involvement of reality in the game.

The methodology: complementary analyses of negotiation processes in the RPG and in social reality

3.1

Our methodology must be helpful for testing our three hypotheses. It must be capable of recording all relevant information in the game and in social reality in order to piece together the different stages in the negotiation processes. The methodology was developed with a view to characterizing negotiation processes in the game and in social reality. It involves:

• setting up a system to record information obtained from the role-playing game (individual or collective behavior of players), analyzing this information, and then asking specific questions to players for a clearer understanding of their behavior and actions during the game.
• identifying stages, stakes and stakeholders in real negotiation processes.

3.2

The aim of this methodology is to identify behaviors that belong to the rules of the game sensu stricto and those which do not. Therefore, the methodology is based on a synergy between real and virtual contexts, the one highlighting the other and vice-versa.

Recording material to collect information

In the role-playing game

3.3

A sheet was created in pulaar (vernacular language) to record all physical operations on the irrigated scheme. A facilitator helps the players to fill in this sheet. All cards (opportunity, repayment, goal, status) are noted by each player on a personal sheet. On each turn, they indicate in advance the actions they intend to perform. After going to the scheme, they indicate the actions they have actually performed. The sheet for players is highly simplified. Symbols are used to allow illiterate players to record information too (figure 7). All formal actions are thus recorded.

3.4

Video recording is used because many interactions occur between more than two players. They are moving in space, expressing their actions and reactions with their body. A great deal of information emerges in the game and it is difficult for one observer to record everything. Though we could have increased the number of observers, the amount of information conveyed and the observer's limits made it preferable to use a video recording system. The problem is the large quantity of information. This major leads to design methods for a priori classifying information that will be suitable for the analysis. The camcorder is used to record negotiations in the irrigated scheme area. The camcorder remains there until the end of the season. During the reimbursement and loan finding phases, the camera is located in the village area.

Figure 7. Players' sheets

In the real negotiation process

3.5

For real negotiations, the video camera was not used. Information was recorded by means of sociological interviews. The aim of these interviews is to construct, a posteriori, the history of the credit negotiation process in the previous season. Information given by different sources is cross-checked. Stages, stakes, stakeholders and individual farmers are identified. The farmers are interviewed about their role in the process, their actions and their opinion of the solution found.

Qualitative analysis of negotiation processes

3.6

Negotiation processes in the game and in reality are analyzed from two angles: the players' interactions^[4] and the topics covered in the discussion.

3.7

First, we focus on the topics of discussion in the negotiation. We consider that during the negotiation process as a whole, people are not involved to the same extent in all the themes of a discussion. Consequently, we note data concerning: topics and stakes of discussion, the players who take part in the discussion, each participant's arguments, his success or failure in convincing other members, his tactics, the members who share or oppose his point of view.

3.8

Secondly, after the negotiation meetings or at the end of the game, we conduct individual and collective debriefings with participants to analyze interesting interactions recorded by the observers. During debriefings, the relationships between participants are identified. Sociological interviews (life history of players, focus-groups) are performed to complete the video-recorded data and shed light on our analysis of players' behavior.

3.9

With this set of methods to analyze negotiations, the interactions between participants and the themes developed in the discussions are investigated in order to understand negotiation processes in the game and in social reality. The synergy between these analyses is necessary to determine the extent to which reality influences the players' behavior. But how are these analyses combined in real and virtual negotiations?

Results

4.1

To test whether the methodology is useful for determining the influence of reality in the game, we applied it in the real context of the search for cropping season loans in Wuro-Madiu and in the virtual context created with the Njoobaari ilnoowo RPG presented in the first chapter. This last part of the article examines this point in relation to our three initial hypotheses.

For the players, to what extent does the game represent their reality?

4.2

In this section, we examine the extent to which the rules proposed in the game correspond to what the players expressed as representative of their reality during the debriefing interviews.

4.3

Firstly, they recognized that not all farmers behave in the same way. Some go to the irrigated area more often than the others. These farmers use more inputs to obtain better yields. Others have other activities and prefer not to go to their plot if an opportunity occurs. So they recognized both the "goal cards" and their links with the "opportunity cards". But in the real world, these goals are not so obvious and not expressed so clearly as by the cards.

4.4

Secondly, even if players behave in various ways with regard to loan repayment in reality, these behaviors are governed primarily by events (water deficiency or surplus, pest attack,...) occurring in a cropping season rather than by personal choice. The owner of a plot attacked by pests may be authorized by the EIG members to not reimburse his loans. So "repayment cards" exist but they are not rules as represented in the game. Moreover, they recognize that repayment behaviors are not the same if the creditor is an individual living in the village, or a business located in the city. So they only partially accepted the representation of the "repayment rules cards".

4.5

Thirdly, the statuses are accepted as representative of the social hierarchy in the village. The mock anger displayed by the real tooroodo when he draws a "maccudo card" and the way the other players (even the real maccudo) make fun of him show how this hierarchy is accepted. Nevertheless, they express a difficulty with the highest level (laamdo), which they do not consider as a status. For us, it is more closely linked to certain functions in the village (imam, head of the village,...). However, the rules associated with these statuses for organizing production in the irrigated area are not accepted. Nowadays, though status may still be important in the village, it does not affect relationships in production activities. The Tooroodo may help their friends to cultivate their plots even if they are maccudo.

4.6

Fourthly, the abacus used to compute water level on each turn does not correspond to reality. This abacus (figure 3) always favors upstream plots. But in reality, a solution to this physical rule has been found and downstream plots may be favored by partly plugging the irrigation pipes with clay to reduce their diameter.

4.7

Lastly, they accepted the RPG as an accurate representation of their reality even though cultural practices are poorly defined.

4.8

Analysis of the gap between the game and reality is very important because it reveals that the game is close enough to reality to make farmers express unease about certain rules.

How does reality interfere in the game session?

4.9

During the debriefing, players pointed out several aspects of the game session that corresponded to situations faced in their real irrigated system.

4.10

Firstly, it reflected their past submission to technician advice. For example, they admitted that they were a little lost at the beginning of the game because they were not used to the rules. But to begin with, they accepted the rules proposed. They decided not to ask questions but to find out how the rules worked as the game progressed. To take this point one step further, we suggest that ignorance of the rules of the game is also a metaphor of the farmers' lack of knowledge about irrigated management when the first irrigated schemes were built in the 70's. But, little by little, they learned how to manage water in these artificial areas.

4.11

Secondly, there is no rule in the game about the players' choice of plot. One player owns only one plot. They decided to draw their plot location at random, and all groups except one acted in the same way. By doing so, they represented the way plots were officially distributed by donors or development agencies. But it is also method habitually used in their small irrigated areas and it is considered the fairest. Reality was accurately reproduced in this case.

4.12

Thirdly, the application of water distribution rules also reflects the way reality interferes with the game. At the beginning of the session, each group decided how to allocate water to the plots along their watercourse. But the rule was not systematically applied. In fact, we detected the emergence of a second rule during the game: the priority was to protect sowed plots from stress to avoid a decrease in yields. The water distribution rule, considered less important, was adapted to the state of the crop. We asked players in the debriefing why the distribution rule was not strictly applied. They confirmed that in the real scheme the second rule is more important than the first and they presented different cases in which the distribution rule is temporarily modified.

4.13

Fourthly, certain deviant behaviors were observed during the sessions, such as theft of water. With the help of the game, they recognized that theft occurs in the real world and they explained to us that several types of theft exist. The worst is when it causes damage to others crops. So, even if it is officially blameworthy (and there are rules to punish theft) some types of theft may be understandable (for example, when the crop might suffer from water deficiency).

4.14

Last but not least, the most visible influence of reality in the game appeared when a conflict occurred in the second group during the fourth turn of the second season. The conversational analysis of the conflict was very helpful (i) to analyze the stakes and the interests of the players involved in the conflict and (ii) to piece together the elements of the conflict that arose during the first steps of the session. The disagreement concerned a conflicting interpretation of the water distribution rule by two players. One of them (participant 15) wanted to strictly apply the rule decided at the beginning of the session by all the members of his group. This strict application was favorable for his plot but unfavorable for the others. The head of the EIG (participant 13) was one of the opponents. Both reality and the rules of the game were used by P13 to justify his position. Because he was the chief, he decided to change the rule with no more explanations. It is interesting to see how reality was used by P15 and P13 as an argument to justify their point of view (P13: 75, 80 / P15: 61,95,97).

Figure 8. The irrigated scheme drawn on the blackboard

Figure 9. Dialogues extracted from the second turn

4.15

At the fourth turn, the game had to be stopped. The players involved in the conflict decided to ask others players for help. A committee was set up in which the facilitator (participant 4) was called and played a major role in resolving the conflict. In reality, he is the same facilitator who intervened in the problem of rural loan allocation (cf 1.2.2). He pinpointed crucial elements to find the source of the conflict. The method used to solve disputes in the real world was very helpful to solve the problem in the game and legitimized the solution obtained. Here people were called in to solve the conflict because of their role in the real world.

4.16

Numerous elements provide confirmation our hypothesis. All participants do not act the same way in the game. For some, the game was regarded as a simulation of reality whereas for others it was more like a stage play, a scene detached from real life. Therefore, some tried to follow the real technical recommendations on crop growing while others simply did the minimum to water, sow and harvest their crop. One player seemed to use the game as a social space to show others players that he is a good farmer in reality.

Does the game reveal social relationships between players?

4.17

The players belong to the same social system. They know each other and share values and norms. Usually in ethnographic studies, the researcher lives with the population he is studying for a long period of time. He has to be immersed in the society to grasp all the information required to analyze social relationships. Participatory observations and interviews are used to fulfill the goals of ethnographic research.

4.18

The RPG was built after a series of interviews to understand how people coordinate their actions in an irrigated area to manage water and crop production. As we have already shown, our methodology is partly based on sociological interviews to understand social relationships, behaviors and norms in the village and in the game. The game is a concentration of time and space. In the previous section we showed that people interact not only by following the rules of the game but also by using their experience of reality. Here we want to show that this sort of game may reveal social relationships to the researchers and can be used as a new tool to investigate social relationships.

4.19

In the conflict presented in the previous section, the conversational analysis also shows that P15 assumes different roles as the conflict progresses: complainant, victim of an injustice, and opponent to the establishment. We recorded his life history after the game session. And it reveals the same character. If he think he is in the right, he will not hesitate to express his point of view even if he is in a minority. To illustrate this reflection, it is interesting to see his position in a conflict that occurred in the village in 1990. Two NGOs wanted to work with villagers. One offered a large sum of money to finance some activities in the village. But it demanded exclusive rights. No other NGO was allowed to work in the village if the villagers accepted its conditions. This NGO offered solutions to short-term problems but did not propose any training program. The majority of villagers accepted these conditions. A minority were opposed to this exclusive agreement. Why should they refuse help coming from another NGO? The conflict divided families. His brothers and fathers were opposed to him. The village authorities, politicians and public administration intervened to solve the conflict. Some of the minority gave in to the majority. But a handful of farmers held their positions and P15 was one of them. Now, the first NGO has disappeared and almost nothing except the scars of the conflict remain.

4.20

With this example, the game has revealed the personalities of certain players and sociological interviews have confirmed them. It means that the game can be used as a tool to investigate social reality, though it cannot be used alone. Others sociological interviews (more conventional) are necessary to grasp whether hypotheses about the behaviors of certain players are relevant or not.

Perspectives

5.1

To what extent does social reality affect players' interactions during a role-playing game? It is very important to answer this question if we wish to understand what is really happening in a participatory process where RPG is used as a group decision support tool. To identify the link between play and reality in our study, we have examined three hypotheses in this paper:

• Hypothesis 1: The game is accepted as a schematic representation of reality
• Hypothesis 2: The players' social background interferes with role playing in the game
• Hypothesis 3: The game reveals social relationships between players

5.2

We implemented a methodology to determine what is really represented for the farmers in the Njoobaari ilnoowo RPG. Our methodology is based on an analysis of the interactions of people involved in negotiation processes.

5.3

The methodology presented in this paper is founded on different synergies to tackle our first hypothesis. Firstly, the materials - sheets, video and sociological interviews - used to record data in real negotiation processes or in the RPG are complementary. The synergy of the three is necessary to record information. Secondly, qualitative methods - thematic analysis, collective or individual debriefings - are useful to analyze some aspects of discussions in small groups. The synergy of these methods is a main feature of the methodology. Lastly, there is also a synergy between the two contexts. The analysis of the RPG gives information about negotiation processes during the game. The analysis of real negotiation processes has given data to improve our understanding of the common social background and the way it affects players' behaviors. Therefore, the data recorded in one situation are helpful to understand what is going on in the second.

5.4

Qualitative data were obtained by means of sociological interviews in real negotiation processes about rural loans. Analysis of the management of rural loans in the previous cropping season has shown stakeholders and their power to legitimize their action, the succession of formal and informal meetings to disseminate ideas, the importance of a final formal meeting to ratify the solution proposed by all villagers. In the game, people in conflict called for a formal meeting. They asked for the help of the real facilitator who showed his ability to solve the problem. Debriefings were also helpful to analyze how players accept the rules and the roles. In general, they were able to identify the links with their real rules and roles. They accept the roles proposed but some rules were too restrictive to allow them to play their roles.

5.5

The game seems to be partly accepted as a representation of the reality of an irrigated scheme. Though the rules are highly simplified, as the farmers play the game, we witness such an explosion of possible combinations of rules and roles, information given, and interactions between players, observers and farmers that the complexity of the real world reappears.

5.6

Over the longer term, a methodology will be developed to analyze the influence of these tools on local negotiation processes.

Acknowledgements

Many thanks to the CDE program of the French Environment department and the PCSI program for partly funding this work and to all members of ComMod group, from CIRAD, INRA, CEMAGREF and IRD for the on-going active discussions, which sustain all these reflections, and to the anonymous reviewers for their helpful suggestions.

Notes

¹Two Senegalese languages.

²National society for the management and the exploitation of Senegal river valley and delta lands.

³Non governmental organization

⁴We tried to use the Bale's method (Bales, 1950) as a way to quantify players' interactions, but the method is cumbersome and the results are poor. It simply confirms information obtained by means of qualitative analysis.

References

BALES, R.F. (1950), Interaction Process Analysis. Chicago, University of Chicago Press.

BARRETEAU O. and BOUSQUET F. (2000), SHADOC: a multi-agent model to tackle viability of irrigated systems, in Annals of operation research, vol.94, pp.139-162.

BARRETEAU O., BOUSQUET F. and ATTONATY J.-M. (2001), Role-playing games for opening the black box of multi-agent systems: method and lessons of its application to Senegal River Valley irrigated systems, in Journal of Artificial Societies and Social Simulation vol. 4, no. 2, <https://www.jasss.org/4/ 2/5.html>.

BOIVIN P., DIA I., LERICOLLAIS A., POUSSIN J.-C., SANTOIR C. et SECK S.M. (1995), Nianga, laboratoire de l'agriculture irriguée en moyenne vallée du Sénégal, ORSTOM éditions.

BOUSQUET F., BARRETEAU O., LE PAGE C., MULLON C., WEBER J. (1999), An environmental modelling approach : the use of multi-agent simulations, in Advances in environmental and ecological modelling, BLASCO F., WEILL A. (Eds), 219 p.

BOUSQUET, F., BARRETEAU, O., D'AQUINO, P., ETIENNE, M., BOISSAU, S., AUBERT, S., LE PAGE, C., BABIN, D. and CASTELLA, J.-C. (2002), Multi-agent systems and role games : collective learning processes for ecosystem management. In Complexity and Ecosystem Management: The Theory and Practice of Multi-agent Approaches. M. Janssen, Edward Elgar Publishers.

CAILLOIS R., (1967) Les jeux et les hommes: le masque et le vertige, Gallimard, Paris.

CROUSSE B., MATHIEU P. et SECK S.M. (1991). La vallée du Fleuve Sénégal: évaluation et perspectives d'une décennie d'aménagements. Karthala, France.

DIEMER G. et VAN DER LAAN E. Ch. W. (1987), L'irrigation au Sahel: la crise des périmètres irrigués et la voie Haalpulaar, Karthala/CTA, Paris.

FERBER, J. (1999). Multi-agent Systems: An Introduction to Distributed Artificial Intelligence, Addison Wesley Longman, England.

HUIZINGA J. (1951), Homo ludens: essai sur la fonction sociale du jeu, Gallimard, Paris.

LIAGRE L., (1997) Les effets de la dévaluation du franc CFA et des politiques de libéralisation sur la filière riz irrigué de la région du fleuve au Sénégal, OSIRIZ, Montpellier.

MUCCHIELLI A. (1983), Les jeux de rôles, coll. Que sais-je ?, PUF, Paris.

Return to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, [2003]