Double loop learning: Are we doing the right things? vs Are we doing things right?

The innovative urban policy making is all about organizational learning. Most organizations — whether public or private companies — are good at single loop learning — aka problem solving. However, most successful innovations in urban policy have been those that used double loop learning to frame the original problem to solve. Double loop learning is more than simple problem solving — this learning style re-evaluates and reframes existing goals or governing values in order to achieve a lasting result.

The author of organizational double loop learning is Chris Argyris who found out in 1977 that peoples’ belief in the ability of organizations to get things done showed that public confidence reached a peak in the late 1960s, and since then it had been deteriorating. At the same time, “information science technology and managerial know-how have continued to increase in sophistication”. So why organizations appear to be less effective as the technology to manage them becomes more sophisticated? His study argues that the “management theory underlying the new sophisticated technology is the same as the one that created the problem in the first place”.

Single loop learning is a fairly straightforward approach similar to a thermostat taking a measure of the room temperature — it receives information (temperature) and takes a corrective measure (temperature adjustment). What is not happening, however, is that thermostat is not asking itself — why is that it’s set to 68 degrees? Are the set goals or governing values right in the first place?

This is a much deeper data analysis that today’s data scientists should undertake when they are being asked to deal with a certain urban issue — the one which creates a much more contextualized view of the problem. However, as one of Arup’s consultants noted, this process is “time-consuming and resource-intensive, so it has largely been confined to academia and larger-scale policy interventions”.

Following upon the previous post about taking into consideration the real context of cities before implementing experimentation, there should be a high-context approach to even starting thinking about the problem - asking the right questions for which to use statistical methods and data analytics tools.

What was true in 1977 still seems to hold true today — despite the sophistication of computational tools, we might still find ourselves asking the wrong questions in the first place and being focused on doing things right instead of doing the right things.


  1. Katz, R. Let’s evaluate urban interventions properly. (
  2. Argyris, C. Double loop learning in organizations.