Saturday, August 18, 2012

Three strategies for strategic actions

In a recent issue of Strategies+Business, authors Tim Laseter and Saras Sarasvathy tell us about three different approaches to strategic actions. After reading through it, I understand their three points, but I think their first one is academic filler... but you judge for yourself:

Approach 1: Planning and Positioning. Characterized by knowledge of the probabilities and impacts of risk events. Decisions to maximize strategic value are made on the basis of expected value. Typical tool is the classic decision tree

Approach 2: Organizational Learning: Probabilities (and therefore probability distributions) are unknown, so risks are now "uncertainties". Events, conditions, and perhaps impacts can be estimated, but not their frequency. This has been termed Knightsian Uncertainty, named for its conceptualizer, Frank Knight who wrote it all down in in book: "Risk, Uncertainty, and Profit".

One really important idea here is that the first step in assessing uncertainty is to estimate whether the downside is affordable. If not, then that's to be taken care of first. Then, choices can be made about the strategic upside of opportunity. Expected value usually doesn't enter the picture since there's no knowledge of probability.

Approach 2 has buried in it our natural avoidance of ambiguity. When the stakes are small, we may choose strategies that are vague but probably doable. But as stakes change, so do attitudes.. bigger stakes, more conservative, eschewing ambiguity for more clarity; in effect, a violation of utility. An understanding of this is generally credited to Daniel Ellsberg, and called the Ellsberg Paradox.

Approach 3: Constructive Transformation. In this strategic formulation, you more or less take what live gives you and then make something out of it. In other words, you don't lead with vision; rather, you lead with process (and tools, and talented people). So, this is something like what Lou Gerstner said when he took over IBM. In effect, when asked about his vision for IBM, he said he didn't need one. What he needed to do was to get the machinery of IBM working again. This, of course, he did, and he did it quite well.

Constructive Transformation is something akin to abductive reasoning: that is, you hypothesize what might be possible that could link many disparate events, objects, or conditions.

So, what do you think? For my money, approach 2 or 3 are close to the way it really works. Approach 1 is theoretically correct, but generally impractical because the information simply isn't available to drive the decision tree.