Sunday, July 10, 2016

Thinking about Governance and Managing Constraints

I follow Dave Snowden's writings on a regular basis.  The following tweet by Dave grabbed my attention, as it is in line with my evolving ideas on complexity in bureaucratic systems.
“We need to stop talking about governance and start talking about constraint management”
I was about to reply with a quick question on governance, but didn't. In pausing I gave the matter some further consideration. In the shower I asked myself, what other practices do we need to consider?   The answer I found is posted here

Basically what we have a 2X2 matrix of Managed Constraints, Algorithms, Instinct and Governance. These correspond with the Cynefin domain's ideas of practice:

  • Managing Constraints is emergent, and thus locked to the Complex domain. 
  • Algorithms are Complicated, but note that there is a penchant for local optimization. As a result, we miss the black swans. 
  • Governance is overly constrained and thus fixed in the Obvious domain. 
  • Instinct is inherently Chaotic as the combination of individuals and circumstances is path dependent (see slide 4).
Further consideration of this matrix reveals the characteristics of the system (slide 6):

  • In the complex domain we design rituals to modulate decision processes.  See the Wikipedia entry to gain a sense of  how rituals are enacted into practice.
  • Treating governance as programmable as suggested by Daniel Kkahneman's 'Strategy firms can think smarter', is just the latest idea in a long string of algorithmic management approaches dating back to Taylorism, McGregor's Theory X, and others. 
  • In the obvious domain we design rules to modulate behaviors. It is easy, nay obvious, to prohibit bad behavior, and most rules are written accordingly.  In government every incident of fraud, waste and abuse seems to result in another rule, the consequence of which is bureaucracy by straight-jacket - an employee's hands are tied behind their back and the public is left wondering why service is so poor.
  • In both the Algorithmic and Governance quadrants the activities we choose - local optimization and a penchant for stability - serve to hide key aspects of social systems. Intermediation is amplified in these domains.
  • Engagement with the subject in both Managing Constraints and Instinct are by definition means of disintermediation. They differ in their approach. Dave recommends probes and safe-to-fail practices along with other ideas; please read his blog.
  • Instinct is far from perfect, witness the many types of cognitive bias that have been cataloged. And after all, the future is unknown. There is a particular class of these unknowns that are important here in the Chaotic domain.  This is the subset of unknown-unknowns we recognize as unintended consequences. This result of these consequences is damaged relationships and a loss of potential relational energy.
No decision is consequence free. However, how we approach decisions matters. There is a significance difference between the probing methods of Dave's safe-to-fail, and the brute force (fail-safe?) or totally naive practices that people partake.

No comments:

Post a Comment