In recent months there has been some great material on predicting the moved of players and decision-makers. The McKinsey Quarterly ran a great article in January “How companies can understand competitors’ moves” that detailed the results of a recent survey of corporate strategists (subscription required). My comments on the article can be found at the Ning Competitive Intelligence social network here. In this month’s Harvard Business Review there is an article Predicting Your Competitor’s Reaction by Kevin Koyne and John Horn (two co-authors also of the previously mentioned McKinsey Quarterly article). This attention to predicting competitor moves and reactions for strategic insights is heartening to this competitive intelligence professional. Another great example of a framework for predicting courses of action through a relatively simply mathematical analysis of the decision markers’ attributes. This framework is described by Bruce Bueno de Mesquita of New York University and the Hoover Institution. He applies his framework to the potential for Iran to develop a nuclear weapon.
While he doesn’t reference the specific study, Bueno de Mesquita claims that a CIA study has shown his prediction model to be 90% accurate even when the experts that provided the data inputs into the model got it wrong. This claim took me several attempts to properly parse– it’s not to say that the data input into the model can be wrong and still return the proper results, because the law of universal conservation of garbage in/garbage out must be maintained. Rather, the claim is that experts that provide data input into Bueno de Mesquita game theory-based model can make mistakes using their expertise-based prediction methods, presumably with the very same data that they input into the game theory model. Certainly we’ve seen some recent literature on the power of quantitative analytical models over intuitive or expert opinion-driven prediction models– one example being Ian Ayres’ compelling SuperCrunchers. That book has several examples of breaking down expert-driven data requirements to get to some small number of variables that are highly predictive in a large majority of instances. A compelling example is a diagnostic framework for cardiac-related emergency room visits that can predict with high certainty the likelihood of cardiac arrest using a small handful of seeming low-relevance variables.
Bueno de Mesquita provides excellent detail of his framework and the important attributes of decision makers that are incorporated into his model:
- An analysis of the individuals that have a stake in the decision.
- What is their stated preference. This is an important distinction because Bueno de Mesquita actively dismisses the notion that we would need to know what individuals want in their heart of hearts. Their stated preferences are strategic choices that reflect their full analysis of the decisions that need to be made and the circumstances. Intuitively this makes sense because in the case of a corporate decision interested parties’ statements will tend to reflect their assessment of their own political standing and interests inside the organization.
- How salient or important is the issue for them? In other words, how willing is the decision maker to drop everything else or put other issues or decisions on hold for this one issue. This seems in me to be related in part to the decision maker’s willingness to take risk related to the issue at hand.
- Probably the most important question is how much influence each decision maker has on the situation. This attribute is important to distinguish between influencers and ultimate decision makers.
One of the more interesting elements of Bueno de Mesquita’s framework is the importance of where each party falls on the Outcome – Credit framework. I have to admit this is always a difficult one for me to grasp, and I think this is related to the second attribute mentioned above: that stated opinion incorporates not only a person’s preference for a specific approach or outcome (more or less) and also an assessment of the organizational politics. As an ENTJ (off the charts on the N and the T) I tend to be very outcome-oriented. I suspect the Thinking – Feeling attribute of the Myers-Briggs has some relationship here, and I would be interested in knowing what others think of this and other methods analysts might be able to use to determine where influencers and decision-makers’ fall on this spectrum. Could the individuals corporate function play a role as well, and how might a person move along the Outcome-Credit spectrum over their career?
With these inputs, Bueno de Mesquita argues, we can apply a game theory analysis to determine the following:
- What choices are available for the decision that needs to be made?
- What risks or chances are the decision makers and influencers willing to take?
- What values do they hold?
- What are the decision makers’ beliefs about other people?
Bueno de Mesquita’s framework presents some very interesting possibilities for scenario analysis– perhaps a more reelable alternative to war games and Delphi surveys. His assertion about stated preferences has real implications for the government intelligence practice, because it places higher value over the statements available more generally from open sources (OSINT) and de-emphasizes the tradition priority for human intelligence (HUMINT) or signals intelligence (SIGINT).
