InterJournal Complex Systems, 272
Status: Submitted
Manuscript Number: [272]
Submission Date: 990211
Revised On: 10301
Reply to Chris Phoenix's Review of MS 226
Author(s): H. Van Dyke ( Parunak

Subject(s): CX.06, CX.68

Category: Brief Article

Abstract:

Mr. Phoenix takes issue with three aspects of our work: its realism (or lack thereof), the reducibility of some of our model's behaviors to the modulo function, and the rules of thumb we suggest for modeling systems. A more detailed response to these criticisms is available at http://www.erim.org/~vparunak/226Reply.pdf . Briefly: The model is simplified to facilitate experimentation, but it is not unrealistic. The agent behaviors were obtained from a manufacturing manager in a Fortune 100 company. The putatively unrealistic features that Mr. Phoenix raises (no shipment of partial batches, repeated ordering over capacity) are in fact not uncommon in the real world. I told him of our industrial validation when we spoke at the conference, and it is unfortunate that his review does not reflect this conversation. The Modulo function is presented as an analytical abstraction of the model, the most abstract form of model considered. Data in the paper shows that this abstraction does not duplicate certain behaviors of the agent-based model (e.g., transients and phase-locking). It is presented as a simplification whose deviations from reality are justified by the clarity of understanding it offers. Such a simplification may be uninteresting from an academic point of view, but represent a considerable cognitive advance for manufacturing managers to whom we have showed the model, and make possible new insights in managing manufacturing systems. The reviewer's challenge to the paper's recommendations rest first on his false assumptions that the model does not match reality, and second on his skepticism "whether any interesting real-life system can be reduced to equations or analytics," and his suggestion that "the ability to reduce a system in this manner, in fact, may serve as a warning that the model's behavior is considerably less interesting than the real-life system under study." This wholesale rejection of equation-based and analytical models runs counter to well-established scientific practice in many fields. Any model represents a trade-off between fidelity and understandability, and simplified models with reduced fidelity are used every day for engineering applications where they are "good enough." The paper encourages researchers to make these trade-offs explicit, rather than adopt a dogmatic stance like that of the reviewer. In sum, Mr. Phoenix's criticisms of the paper reflect a memory lapse concerning what I told him at ICCS, inaccurate assumptions about the nature of manufacturing, a highly academic view of what constitutes "interesting" behavior, and a predisposition to reject abstract models (which is to say, any models at all). His criticisms do not even engage, much less challenge, the paper's substantive claims about the relative capabilities of different modeling formalisms.

Retrieve Manuscript
Retrieve Previous Revision's Abstract
Submit referee report/comment


Public Comments: