Can computers be social?

Of main concern in agent based computing is the conception that software agents can attain socially responsible behavior. This idea has its origin in the need for agents to interact with one another in a cooperating manner. Such interplay between several agents can be seen as a combinatorial play where the rules are fixed and the actors are supposed to closely analyze the play in order to behave rational. This kind of rationality has successfully being mathematically described. When the social behavior is extended beyond rational behavior, mere mathematical analysis falls short. For such behavior language is decisive for transferring concepts and language is a holistic entity that cannot be analyzed and defined mathematically. Accordingly, computers cannot be furnished with a language in the sense that meaning can be conveyed and consequently they lack all the necessary properties to be made social. The attempts to postulate mental properties to computer programs are a misconception that is blamed the lack of true understanding of language and especially the relation between formal system and its semantics.