A Subtle Architecture of Safe Decision-Making


"We need to be more decentralized in our decision-making," she urges in an impassioned plea to her team of managers. "Indeed," they heartily agree. "The power of many minds is a source of great strength!" A burst of electronic music interrupts, as a text message buzzes in her hand. She reads for a moment, looks up, and without missing a beat proclaims, "The Executive Team and I think that we should be more centralized in our decision-making." "Hear, hear," comes the echo. "An expert opinion is a good opinion!"

- - - o o o O O O o o o - - -

In shades of George Orwell's Ministry of Truth, without a flinch, the prevailing notions of decision-making-at-the-centre or decision-making-everywhere seem to buffet us like the winds. So which is better? "Well, it depends," we cry. With true Hegelian flair, the debate ends on this synthetic note. Or does it? Perhaps a few heuristics can trim our sails just enough to take advantage of the nuance of the debate. Heuristics that navigate the two main caveats of the collective decision process. Firstly, is our decision-making seascape Safe enough? Secondly, is it Subtle enough?

- - - o o o O O O o o o - - -

I think that runaway ideas can be dangerous. A ship that has slipped her mooring is not only a danger to herself, but other craft besides. There are perhaps two categories of runaway ideas. The first kind comes in the form of a Rogue, yet charismatic, leader: a "Mule" in the terms of Isaac Asimov's psycho-history. An unforeseen and unforeseeable galvanizer of hearts and minds that threatens to either sink an organization, or aquaplane it to better futures. The second kind of runaway idea takes the form of the Herd - the characteristic group-think coalescence of many thoughts towards a single (sometimes mediocre) thought, when the libido for consensus is stronger than the rationality of a good outcome. So, we need a Safe way of managing the equilibrium between Rogue and Herd that can act as a metaphoric circuit-breaker for runaway ideas.

- - - o o o O O O o o o - - -

I think that bad ideas are almost (but not quite) as dangerous as runaway ideas. The danger arises from the uncertain naming of the goal of the decision for which the ideas are the required. What I mean is this: 1. Are we looking to make a simple yes/no, or left/right or short/medium/long decision, or are we looking for a vast pool of possible options? 2. Are we trying to find a pure mathematical aggregate of ideas, or a consensus with a deep understanding of the degree of disagreement? 3. Is the mean sufficient, or do we also care about the outliers? Single or multiple outcomes; aggregate or consensus? This is the Subtlety - it is called an Architecture.

- - - o o o O O O o o o - - -

Galleys stocked with the above provisions (or provisos), we can venture forth with the following heuristics.

  1. Heuristic One: Don't use a chain-saw when a scalpel will do. Not every decision requires the same approach or mindset. Some decisions require collaboration; others do not. Be confident to select the appropriate methods and processes in each case. Don't engage in the centralize against decentralize dogmatic crusade. It is not as simple as that - know how to use both.
  2. Heuristic Two: A bigger crowd means more complexity, and sometimes better wisdom Don't believe the con that you always have better decisions when more people are present in the room. Every additional "node" in a linked network, an exponential number of connections is made, each with its own social, trust, political, power-game, emotional and intellectual characteristics. When these connections interact we can expose ourselves Rogues and Herds unless we are careful. But if you do have many people participating in a single conversation, do ensure that you have, as James Sorowicki puts it, the 4 conditions under which Crowds can be Wise
    1. Independence
    2. Diversity
    3. Local knowledge
    4. A robust mechanism of aggregating data
  3. Heuristic Three: Name the outcome Principles of organizational design have structured organizations in ways that make sense to the way things work best (OK, assume this to be true for a moment). The question is to decide where in these organizations we need many minds and voices and where do we need only a few? Some examples to help us: 1. Is the outcome time-sensitive? Information on the past causality of a current circumstances is a good use of collaborative techniques; long term alternative futures of an organization are not. 2. Is fast adaptation required to achieve the outcome? A sales person with a discretionary discount allowance in hand is a good place to have decentralized decision making; industrial relations policy is not. 3. Is deep technical expertise required to achieve the outcome? Organization culture change is a good place to canvas many anecdotal fragments on the best course of action; heart surgery is not. 4. Will a better outcome be achieved through proprietary or open standards? My examples are weak here - refer back to Heuristic One.
  4. Heuristic Fourth: Protocol must be defined At the most appropriate level of the organization, the agreed application of these heuristics should be debated and agreed. Authority must delegated and trusted; circuit breakers should be put in place where needed; discretion must be permitted where necessary; agreement must be sought where possible; mitigation should be designed where risk is possible.

- - - o o o O O O o o o - - -

Leaders must lead.
Be Safe!
Be Subtle!