تجزیه و تحلیل تجربی از رفتارهای تیم خود سازمانده

This paper is focused on the study of self-organizing team’s behaviors which are dependent on the interaction rules and the decision factors of team members. The self-organizing team’s behavior means that team members work unconditionally with one of the three work attitudes (diligence, average, and shirking). A small-world network is suggested as the basic relationships of team members. Different from the traditional models, Reciprocators encourage their friends if they work diligently and punish them if they shirk work. It is supposed that team member’s decision of choosing work attitude depends on four decision factors, humanity, herd instinct, rationality, and follower tendency. Firstly, all of the four decision factors’ weights are supposed as 0.25. Multiple experiments were conducted to analyze the behavior of a team by a multi-agent experiment system. It is found that, in order to increase the fraction of diligent team members, different strategies should be used under different Reciprocators’ fractions. Increasing Reciprocators’ fraction is beneficial to the increase of diligent members; however, the increase rate will slow down after an inflexion (here it means the inflexion of Reciprocators’ fraction). After the previous experiments study, extended experiments were developed to work on the influence of the four factors’ different weights. A self-adaptive algorithm is suggested to achieve the four decision factors’ weights. The results of self-adaptive algorithm have different influences on the team’s behaviors under different fractions of Reciprocators. Finally, influences of members’ different relationships are studied by other experiments. It is also proved that the fraction of diligent members is not dependent on the structure of team members’ relationships. The results demonstrate that the self-organizing team’s behavior can be significantly influenced by its scenario while managing a self-organizing team.

The essence of self-organization is that a system spontaneously arranges its components or elements in a purposeful (non-random) manner, under appropriate conditions but without being guided or managed by an outside source. Self-organization is usually associated with more complex, non-linear phenomena, rather than with the relatively simple processes of structure maintenance of diffusion. The self-organizing behavior of social animals suggests that self-organization should be expected in human society. Usually the growth of social networks is fueled by social contexts and ideology of all participants in the network. Due to the social networks’ complexity and non-linearity, this work is focused on the study of self-organizing team’s behaviors by a multi-agent system. A multi-agent system (MAS) is a system composed of multiple interacting intelligent agents. Multi-agent systems can be used to solve problems which are difficult for an individual agent or monolithic system. Social structures are appropriate to be modeled by multi-agent systems. Bowles and Gintis (2004) offered a model of cooperation and punishment that is called strong reciprocity, where Reciprocators only punish shirkers. In our work, a team consists of three types of members, Reciprocators, Cooperators and Selfish. It is supposed that Reciprocators encourage the diligent team members besides punishing shirkers. Additionally, it is assumed that each team member has one of the three categories of work attitudes, diligence, average, and shirking, and a small-world network (Watts & Strogatz, 1998) is suggested as the relationship network of all members. Under the assumptions described above, this paper works on the self-organizing behaviors of a 100-team members’ social network. In this model, the typical self-organizing behavior is that multiple members work diligently with high cost without being guided or managed by others. This paper also assumes that member’s decision is dependent on four factors, humanity, herd instinct, rationality, and follower tendency. All the four decision factors’ weights are supposed as 0.25 at first. By a multi-agent experiment system, the relation between the Reciprocators’ fraction and the team’s behaviors was found by these experiments. Increasing Reciprocators’ fraction is beneficial to the increase of diligent members; however, the increase rate will slow down after an inflexion (here it means the inflexion of Reciprocators’ fraction). Extended experiments were developed to work on the influence of the four factors’ different weights. A self-adaptive algorithm is suggested to achieve the four decision factors’ weights. The results of self-adaptive algorithm have different influences on the team’s behaviors under different Reciprocators fraction. Finally, influences of members’ different relationships are studied in this work too. The rest of the paper is organized as follows. The related literature is reviewed in Section 2 and then the social model is developed in Section 3. A series of experiments are conducted in Section 4. Four propositions under different experiments are generated and a detailed result analysis is presented in Section 4. Finally, the conclusions are summarized and future work is suggested in Section 5.

This paper proposed a model of self-organizing team with small-world relationships. In this model, Reciprocators encourage his friends if they work diligently and punish them if they shirk work. Many experiments were made to analyze the fraction of diligent work members’ in this model. On standard running, it is found that different populations (Reciprocators, Cooperators, Selfish) have different presentations under different Reciprocators’ fractions. Increasing the fraction of Reciprocators is beneficial to improve the ratio of diligent members; however, the diligent members’ increase rate will slow down after an inflexion (it means the inflexion of Reciprocators’ fraction). It is assumed that team members’ decision is dependent on the four factors. A self-adaptive algorithm was suggested to achieve these factors’ weights. It is found that these self-adaptive members’ behaviors changed differently under the conditions of different Reciprocators’ fractions. Through these experiments, team managers should use different management strategies if they want to manage self-organizing teams’ behavior reasonably. This paper also proved that self-organizing team’s behaviors are not dependent on the team members’ relationship networks. This work can be extended in many areas. The evolution rules, which may support high levels of self-organization, can be captured for this work. Different evolution rules will show a significance for the behavior of self-organizing teams. The extended experiments should be developed to work on the evolution of self-organizing teams under different rules. Some empirical analysis of the evolution of self-organizing teams may be suggested too. Additionally, it will be beneficial to utilize these rules to manage self-organizing teams reasonably.