Information Technology Reference
In-Depth Information
Therefore the behavioral focus of our research is on how producers choose from
alternative new product designs where the alternatives might differ in performance
along dimensions related to consumer utility. We are especially interested in
settings where the performance dimensions emerge endogenously rather than
being fixed or introduced exogenously. For example, this would arise when new
uses are found for a product, a use case is dramatically altered, or when new classes
of users enter the market place.
To study this class of behavior and phenomena, it is necessary to observe the
value systems and behaviors of both producers and consumers during the post-
design phase, and preferably over several generations of product design to record
changes in design trajectories. It is also useful to explore various experimental
treatments to understand how alternative conditions of cognition, value systems,
and social interactions affect outcomes. The phenomena of interest arise through
the dynamic social interactions between agents, and between agents and artifacts,
far from equilibrium. Therefore, it is important that the research method allows for
the study of the endogenous processes for learning (direct and indirect), social
interactions and network formation, and be capable of rich emergent phenomena.
For these reasons, we have chosen to build a computational laboratory with
multiple agent types, artifacts that agents produce and consume, and social inter-
actions between agents. One of the main benefits of this research method is that
many different experimental treatments can be explored and the detailed internal
state of all agents is fully available for examination.
There is also a broader spectrum of interactions and forces at work in the post-
design phase, including influences of government policies (e.g. intellectual property
institutions), social interactions between producers (e.g. communities of practice),
the value-shaping influence of third-part agents such as market analysts and 'gate-
keepers', and the upstream influence of funding organizations and research
institutions. Though these are beyond the scope of our current research, the
architecture of our computational laboratory is extensible to include these other
types of agents, artifacts, and interaction types.
One of the goals of this paper is to demonstrate the viability of Agent-based
Modeling (ABM) to study innovation ecosystems and their social aspects. We have
implemented a multi-agent system (Ferber 1999 ; Weiss 2000 ; Wooldridge 2008 ) .
It is designed to be a computational laboratory (Casti 1999 ) to support a wide
variety of experimental settings and tests. Using multi-agent systems to simulate
social systems is part of an emerging interdisciplinary field called Computational
Social Science (Epstein 2012 ; Epstein and Axtell 1996 ; Gilbert and Conte 1995 ;
Miller and Page 2009 ). Briefly, agents and their micro-level behaviors are formal-
ized using relatively simple rules and limited/plausible capabilities for reasoning
and behavior. An open environment is provided for agent interaction, often in the
form of a grid or network. Through interaction with other agents and the environ-
ment, each agent alters its internal state, learns and adapts. The general research
strategy is to study emergent phenomena that are not simple aggregations of the
micro-behaviors (Gilbert 2002 ; Goldstein 1999 ; Holland 2000 ) .
Search WWH ::




Custom Search