Dealing with variability

As a result of some back and forth with members of AEA’s Systems TIG, I have been thinking about the idea that over and above specific activities in any one organization, what’s also important is its capacity to deal with variability. And presumably, because this is important in organizations, it is important in evaluations as well.

At 10:11 PM -0400 1/11/10, Jason Burkhardt wrote:
Good evening,

This is just a thought based on my experience with families and family systems. The model we used had the idea that interventions with families did not necessarily focus on the specific behaviors of one individual, but the system as a whole. For example, while it would be impossible to predict the day to day actions and decisions of an unpredictable parent in order to design contingencies for every possible action, we looked at the propensity of the parent towards unpredictability as the main issue, as well as the reaction by others close to that parent when they were unpredictable.

For evaluation, then, rather than having the ability to forecast any one positive or negative action taken by the organization, one could look at whether the organization is capable of making more positive decisions, whether the organizational structures can support “healthy actions”, or whether the political and social contexts that surround the program have the ability to support positive actions and decision making. I am not sure that this is necessarily breaking ground, but hopefully this perspective is helpful.

Jason Burkhardt

Several thoughts flow from that.

One is that this is pretty much what Stafford Beer talked about in his notion of “requisite variety” for a viable system. If one sub-system has insufficient variety to handle another sub-system’s variety then the entire system is unviable. Information management and decision-making protocols are classic examples of this.

Also it reminds me strongly of one of Deming’s argument about quality control that has been largely ignored although the recent book by John Seddon “Systems Thinking in the Public Sector : the failure of
the reform regime and a manifesto for a better way” has attempted to revive it). This is the simple fact that the main factor which undermines “quality” in most management processes is the outliers.
Outliers disrupt processes, and thus good quality control is all about how these unexpected and/or unusual events are handled. This is at the basis of the criticisms of “lean processes” such as Six Sigma, where the focus is on sustaining the quality of the majority processes and not building capacity to handle disturbances. How “disturbances” are handled is also at the core of approaches based on Activity Systems (which is why it has been especially successful in addressing health and safety issues in areas such as the transport sector, nuclear power and the oil industry)