[Note: assumes familiarity with Cynefin]
Ever since the explore-exploit tension came to my attention in O'Reilly and Tushman's 2004 HBR article on the "ambidextrous organization", any discussion of it has been of interest...primarily because of its resonance with Cynefin ("Exploit" is an Ordered domain tactic, and "Explore" a Complex domain tactic).
The most recent discussion I've seen is an HBS working paper ("Accountability and Control as Catalysts for Strategic Exploration and Exploitation: Field Study Results", Simmons).
As Simmons observes, there's no clear agreement on how organizations should balance the tension between Exploit and Explore. At one extreme, both can be integrated at the lowest organizational level. At the other extreme, Exploration can be assigned to a dedicated group at the enterprise level. Regardless, managing the tradeoffs between the two seems to be a Complex (probing/exploratory) job.
In this working paper, Simmons uses data from 102 field studies to study two related organizational variables, span of control and span of authority, and how they can be varied to explore one aspect of the tension between Exploitation and Exploration.
My initial reaction was that a traditional approach to control systems and delegation of authority would probably look more like analyzing Complicated domains (e.g., balanced scorecards, Six Sigma, etc) than probing Complex ones. Although Simmons acknowledges the need for the former, his earlier research focused on interactive control systems that are designed for the latter.
In this paper, he investigates violations of the "controllability principle", which states that a manager's span of control and span of accountability should be the same. He comes to the conclusion that innovation (Exploration) may actually require that the span of accountability be significantly larger than the span of control.
Span of control (C) and span of accountability (A) differ depending on job type (a CEO has a wide range of both; a front-line supervisor has a narrow range of both), and can be combined in various ways:
- C = A As discussed above, this is considered to be the desired norm at all levels for Exploitation activities. My reaction was that it maps to Cynefin's Simple (for lower organizational levels) and Complicated (for upper organizational levels) domains.
- C > A At all levels, this is a recipe for inefficiency...the inefficiency gap may be Chaotic (i.e., agents may be relatively unconstrained in those areas where they exercise control without accountability)
- A > C This creates what Simmons calls the "Entrepreneurial Gap", where managers must explore areas outside of their control to close the gap between accountability and control. This seems to map to Cynefin's Complex domain. This is consistent with the oft-made observation that constraining resources seems to spark innovation. Influence, not control, must be used to span the gap between C and A. The use of persuasion, not coercion, requires the manager to use attractors and boundaries that can exert influence to achieve a goal (rather than the blunt imposition of Ordered controls). This forces the organization to orient itself toward key uncontrollable entities related to accountability (customers, partners, etc). Where A exceeds C across the enterprise in a coherent way, a coherent external orientation may also emerge across the enterprise, providing a framework for coherent Exploration across organizational stovepipes...an activity that is not, by definition, subject to detailed planning and control.
Sometimes, A is increased or C is decreased (or both) deliberately to encourage Exploration...but an entrepreneurial gap can also be unexpectedly imposed by shifts in external factors (e.g., technological shifts, financial crisis, etc). In these situations, adequate organizational resources, including such "soft" resources as identity and experience in exploring, can determine whether the organization flounders or thrives. And, the opposite can occur...external changes cause A to decrease, C to increase (or both), resulting in increased sloppiness & inefficiencies (e.g., the American auto industry after WWII).
See the paper for more details....the primary concept I took away was that C & A are organizational variables that can be changed to better match the organization's activities (explore/exploit) to its context (simple, complicated, complex).
