Recently, I wrote about how we can use Rubik’s cube to glean important principles of strategic decision making. When I posted the article on LinkedIn, my sister added a comment about integrity. She noted that taking the cube apart and putting it back together again might give us a quick but un-respected win. Brute-force disassembly would bring insight into the structural mechanics, but would ultimately not be satisfying. I remember taking my Rubik’s cube apart and putting it back together again, sometimes to lubricate it, and sometimes in frustration when I could not solve it.
The comment also reminded me of a seminal teaching in Cognitive Science: David Marr’s Levels of Analysis. Thinking about Marr and Rubik’s cube, highlighted a few more lessons about Strategic Decision Making to me.
A Foundation in Cognitive Science
I first came across David Marr’s theory during a Fulbright year I had the privilege to spend at UCLA. I have very fond memories of taking Keith Holyoak’s “Thinking” class. My year at UCLA would shape the rest of my life, in many different ways, and every time I return to Westwood, I feel at home. But I digress…
David Marr was interested in developing computational models of vision. He posited that every (computational) problem presents on three levels of analysis: The Computational level, the Algorithmic level, and the Implementation level. Crucially, we need all three levels to provide a satisfying, effective solution. Marr himself famously stated, “trying to understand perception by studying only neurons is like trying to understand bird flight by studying only feathers: It just cannot be done.”
Marr’s approach has guided my thinking ever since, and I believe it can also help understand what is needed to deliver satisfying solutions for Strategic Decision Making problems.
Marr’s Three Levels
The Computational Level: The What And Why Of Strategic Decision Making
At this foundational level, we define the problem and its purpose. Marr illustrated his theory via an analogy: That a of a cash register. A computational level description of a cash register is that its purpose is to add numbers and keep a financial record. But at this level, we are not yet delving into the how – we merely specify the goal. But doing so is crucial to guide the design that follows.
In strategic decision making, on this level we deal with understanding our objectives and the context in which they exist. What are we trying to achieve? Why is this goal important? Defining the “what” and “why” ensures our decisions are purpose-driven and aligned with our overarching mission.
The Algorithmic Level: The How Of Strategic Decision Making
This level focuses on the processes and rules needed to achieve our goals. For this reason, it is also referred to as the Representation level. Here, we develop strategies, models, and plans to tackle the problem. In the case of the cash register, there are multiple ways we could “do” addition: We can use the Arabic numeral system to represent number, and use rules such as adding the least significant digits first and carry the result if it exceeds 9. Or we could use binary representation of number and use binary arithmetic. In the case of strategic decision making, this translates to devising methods and frameworks that guide our actions. The “how” provides the roadmap, detailing the steps and approaches necessary to move from our current state to our desired outcome.
The Implementation Level: The Physical Realization Of Our Strategic Decision Making
Finally, this level concerns the physical realisation of the processes and representations we specified at the algorithmic level. For a cash register, this is the physical machine that operates using the defined algorithms to achieve its computational goals. This could be an abacus, geared wheels with numerals on them, or the digital circuitry that performs the binary computations. For strategic decision making, this concerns the execution of our strategies. It involves the tangible actions we take to implement our plans in light of real-world complexities. Here is where our decisions come to life and become actionable and grounded in reality.
The Importance of Synergy
No single level can stand alone. Just as understanding bird flight requires more than examining feathers, and understanding a cash register requires more than knowing it adds numbers, effective decision making demands the integration of all three levels. The computational level gives us the “what” and “why,” the algorithmic level provides the “how,” and the implementation level brings our plans into action. Together, they form a holistic approach that is robust and comprehensive.
It is crucial that we begin with the What and Why before jumping to specific Decision Frameworks, like weighing options, or identifying uncertainties. Without the What and Why, the How is pretty useless. We might have a perfectly clever roadmap, but if it is not informed by an overarching vision, it will prove pretty useless.
Once we have the What and Why worked out, we can develop a decision framework to represent and tackle our problem. Our options range from the very simple pros and cons list to complex frameworks like multi attribute utility theory.
Our implementation will heavily depend on how we have chosen to represent the decision. But it is just as important as the other two levels, for without it, we would not really achieve anything. We have to turn our thoughts into action.
