|InterJournal Complex Systems, 203
|Manuscript Number: |
Submission Date: 980730
Comment on manuscript revision number 39927|
Reply to the referee report 193.
Category: Brief Article
As explicitly stated in the abstract of the paper, it “is intended to serve as an invitation for empirically grounded modeling of cognitive and emotional dynamics in co-acting groups.” In other words, my goal was not to present verified findings but to attract potential collaborators by describing initial yet quite elaborated dynamic conceptual framework, unusually rich and authentic data that is available, and first advances in relating the two. Given novelty and complexity of analyzing human groups as dynamic systems(for corroboration of this statement see an article by Joseph McGrath “Small Group Research, That Once and Future Field: An Interpretation of the Past With an Eye to the Future” in Group Dynamics: Theory Research, and Practice 1997, Vol. 1, No. 1, 7-27), the paper makes a substantial contribution by outlining a possible research approach. It took 2 years of observing “in situ” problem-solving groups--about 30 of them--and of searching through and compiling existing ideas in the field to come up with what the referee considers to be “random thought.” To judge my paper adequately, it is important to distinguish between scientific studies at two stages: (1) generation of hypotheses and (2) their testing. The criteria at the first stage should be driven by the question “Will it fly?” rather than “Has it flown?” To give a reliable answer to “Will it?” question, one has to have strong professional intuition based on hands-on knowledge of the field, and real curiosity about the substantive issues addressed by the study. This sounds fuzzy. Yet unless space and attention are given to emerging ideas now, in the future it will be boring to apply rigorous criteria of validity and reliability of findings to research projects that were originally killed by oversimplification while trying to look “scientific” prematurely. Since the paper was written, moving along the lines sketched in it I have been able to establish much tighter link between the theory and data and hope to present my findings during the 2nd ICCS. Also, I will be happy to rework the concluding section and to add the following passage to it. In his popular book on plectics Gell-Mann (1994) introduces the notions of crude and effective complexity. He defines the former as “the length of the shortest message that will describe a system, at a given level of coarse graining, to someone at a distance, employing language, knowledge, and understanding that both parties share (and know they share) beforehand” (p.34). In contrast, effective complexity “can be roughly characterized as the length of a concise description of the regularities of that system” (p. 50). The distinction is important for envisioning how to proceed while developing two kinds of models necessary for relating a problem-solving process to effectiveness: one - of the process itself, and another one - of the links between the events and/or variables of the process and indicators of effectiveness. A task of developing a model of the process itself is close to figuring out crude complexity of a problem-solving group. Regularities of group behavior are important, because knowing them will enable us to formulate rules making the description shorter. Yet not shorter than necessary for capturing dynamics of the problem-solving process in the sense of characterizing each new act as a function of the immediately preceding cognitive and emotional states of all participants. Because participants have to deal and are dealing with developments in group behavior they cannot predict, it becomes important to model such unpredictable--and in this sense random--events. Discovering links between the events, sequences of events, variables of the process and indicators of effectiveness is a very different task from creating “artificial life” that simulates problem-solving behavior. Effectiveness is a conceptual construct and we are interested only in modeling regularities that relate it to group processes. In this case we try to figure out the effective complexity. Although this task is different from accounting for randomness of group dynamics, it is not necessarily simpler. Newell and Simon (1972: 301) noticed long time ago that when we have large combinatorial behavior spaces--as in the case of group problem-solving behaviors--small instantaneous changes can bring about great divergence further down the behavior paths. Impact of events can accumulate or they can cause radical changes in behavior at fluctuation points. Also events can combine in many ways compensating, increasing or canceling impact of each other on effectiveness. Thus it becomes rarely possible to speak about impact of any isolated event on effectiveness.
|Retrieve Manuscript Abstract|
|Submit referee report/comment|