PAXsims

Conflict simulation, peacebuilding, and development

Daily Archives: 19/10/2009

David Earnest on “unknown knowns” and MMO-based similation

I recently posted some thoughts (“The Internet is for COIN?“) on an article by David Earnest on “Growing a Virtual Insurgency: Using Massively Parallel Gaming to Simulate Insurgent Behavior” in the most recent issue of The Journal of Defense Modeling and Simulations (October 2009). David was kind enough to reply to my thoughts with some additional thoughts of his own, which I’ve moved from the comments section so that they are less likely to get missed:

* * *

Thanks for the review of my article, Rex. I share your sense that an MMO simulation could help the aid and development community. Indeed, the genesis of my article is the finding of complex systems researchers that massively parallel systems–whether simulated or real-world–exhibit surprising abilities to adapt and solve problems. While this adaptability is characteristic of a range of physical and ecological systems, my interest is in human organizations (I’m a political scientist by training).

One thing that interests me is that human organizations, while often very large, are not always “massively parallel” either in the gaming sense (e.g. the architecture of the social network) or even in terms of organization theory. Because human organizations often are bureaucratic and hierarchical–and political scientists typically focus on the most bureaucratic and hierarchical organization there is, the nation-state–they are not very good at tapping into the latent expertise and knowledge of individuals. This is a feature of many fields of human endeavors, not simply COIN operations. While I have no personal experience with aid and development organizations, I wouldn’t be surprised if these organizations also have lots of latent experience and knowledge that needs to be drawn out. How to do so is an interesting question; gaming is but one possible approach.

This interest in latent knowledge highlights a paradox that intrigues me: Organizations exhibit higher-level learning even when no one individual learns. That is, organizations “know” things that even individuals in the organization do know realize they know. A few years back, a colleague at a conference suggested an addendum to Donald Rumsfeld’s famous quote about “known” versus “unknown unknowns”. To Rumsfeld’s list we should add “unknown knowns”–that is, organizations have expertise and knowledge somewhere in their ranks, but we don’t really recognize it or know how to tap into that expertise.

Your point about the inaccuracies in Wikipedia is an important one. Massively parallel human organizations may be prone to all sorts of pathologies, from groupthink to cycling majorities and other problems of collective action . But I think (or perhaps hope) that the Wikipedia example illustrates how massively parallel systems cope with bad information. Over time, mechanisms of positive and negative feedback, plus selection pressures, weed out the “bad” information while promoting the “good” information. Thus, while at any given moment in time a massively parallel system may be “inefficient” (in the sense of a signals to noise ratio), over the long run efficiency and the quality of information improves. This can only happen, however, if organizations (or gamers) design appropriate mechanisms of reward, censure, and feedback (or as I say in the article, conservation, selection and innovation). In ecological systems, these mechanism arise endogenously. The challenge of human systems is to endogenize them without either eliminating the massively parallel adaptive architecture or creating too much noise.

On a related note, Mitchel Waldrop wrote an article in 1996 about how Dee Hock, the founder and CEO of Visa (the credit card organization), solved these organization problems. It’s a fascinating read: see the magazine Fast Company, October 1996.

David C. Earnest
Associate Professor of Political Science and International Studies
Old Dominion University

%d bloggers like this: