Diagram of the OODA loop

Organisational Strategy from First Principles: the OODA loop

It is difficult to formulate a novel, winning strategy. An efficient market will display 'no-free-lunch' behaviour to arbitrage away any excess returns if you are not sufficiently clever with how you develop your strategy. So, how do we win at a game which keeps trying to return us to equilibrium? We have to look at the underlying structures of the game: its players, rules and playing field. We have to focus on the metagame.

It is difficult to formulate a novel, winning strategy. This statement is true in a lot of domains (e.g. stock trading, chess, life; presumably it is true for domains where long-term planning is difficult, involving complexity/feedback/chaos). When talking about business strategy, the statement is true given (realistic) assumptions about competitors who are “equally savvy, watchful of your actions, […] and are equally aware of the importance of brand as you are”. In other words, an efficient market will display ‘no-free-lunch’ behaviour to arbitrage away any excess returns if you are not sufficiently clever with how you develop your strategy.

So, how do we win at a game which keeps trying to return us to equilibrium? We have to look at the underlying structures of the game: its players, rules and playing field. We have to focus on the “metagame”:

“The right question to ask in the face of a dynamic system of competition is the following: how do you become the sort of organisation or person who is able to come up with new strategies in the first place?”.

Tools for thought: the OODA loop

Frameworks help us to clarify, analyse and communicate our mental models. USAF Colonel John Boyd – “the greatest military strategist of the 20th century” – developed the OODA loop and applied the concept at the operational level of military campaigns. This is a framework which structures the general decision-making process into a cycle of 4 interlinked loops: Observe, Orient, Decide, Act.

Diagram of the OODA loop (from https://en.wikipedia.org/wiki/OODA_loop)

Distilling decision-making into a set of discrete steps gives us a tool which is helpful for structuring systems and assessing their nuances, for example when analysing an AI use case (where the framework’s feedback loops have fairly obvious interpretations):

  • Observe – data inputs:
    • Implicit guidance: what bias exists in the data?
    • Unfolding circumstances: what is the data velocity?
    • Outside information: what are the external dependencies?
    • Unfolding interaction with environment: does the system itself change the data, for example if the system is a process measuring load on its own infrastructure?
  • Orient – data preparation and model design:
    • Various considerations when integrating data into decision-making: cleansing, quality, feature engineering, ethics by design
  • Decide – model inference:
    • Application of whichever algorithm/model/decision policy – nothing really to add here
  • Action – follow-on integration:
    • Nothing really to add besides the observation that an AI model is useless for decision-making if its outputs do not go anywhere

I should mention that criticism has been levelled at the OODA loop “for lacking originality, as pseudoscience […] and lacking scholarly rigor. […] The OODA concept struggles to meet the rigorous social science standards of epistemological validity, theoretical applicability, falsifiability and robust empirical support”. But this criticism is misplaced – the OODA loop is a general framework which necessarily abstracts situational details into high-level principles that are useful for developing the “meta-strategy”.

Applying the OODA loop

At its core, the OODA loop represents a feed-forward flow of information from data source to decision point. This is a deliberate over-simplification: the feedback component of the framework is important for optimisation at the meta level, but static decision-making can still be represented without it.

  • For example, the OODA loop can be used to structure a diagnostic RCA within a DevOps context, in order to recognise best practices like collecting additional data, approaching with an open mind, and solution documentation.
  • That said, the framework can be explored in much more depth in order to drive a Continuous Delivery culture within the DevOps function (an example of a winning meta-strategy).

The OODA loop concept is also present at the heart of Lean Startup, which revolves round the Build-Measure-Learn loop of building MVPs, measuring market response, and learning from feedback. Similarly, Product Management can be represented at an operational level in an OODA structure: Observe trends in your customers and competitors, Orient around your market positioning and value proposition, make design and business Decisions based on stakeholder feedback, and Act to ship your product (continuously).

What does OODA tell us about the Product Management structure?

Imagine your are a corporate decision-maker: a CEO or a startup founder. If, as the OODA loop shows us, decision-making is an optimisation loop, then what is the most meta (i.e. fundamental) decision which forms the foundation of the organisation? Logically, we must be optimising based on inherent characteristics of the social context which existed before the organisation. In other words (i.e. attempting to bring this back from esoteric philosophising…) the first decision which an organisation must make concerns its purpose for existence, and this must be based on a set of intrinisic values held by its members.

A weak purpose is one which is only loosely-driven by social context (“make lots of money”) while a strong purpose is one which is a direct response to social context (“make positive social impact”), although some may disagree (I can see some logic behind an argument that “enormous [social] progress” comes indirectly when companies “mainly focus on making profits and not be distracted by social goals” but it is very difficult to square this against viewpoints from the UN, IPCC, scientists, system theorists…).

There is a logical set of questions to be asked subsequently (and simultaneously), which move from high-level strategic to operational choices:

  • What are our core values and aspirations?
  • What is the social context/market in which we are hoping to achieve this aspirations?
  • How do we win the metagame within this context/market?
  • What functions and capabilities are required to win the metagame?
  • What processes and resources are required for these functions to run?

If we had a team working on each of these questions, we could stick a label on them which might look something like:

  • Strategy
  • Marketing
  • Product / Engineering
  • Product Management
  • Operations, Finance, Infrastructure, HR, Legal…

Conclusion

  • Starting with a simple framework for modelling strategic decision-making, we have realised that winning in complex domains requires a focus on the metagame.
  • This process should extend to the heart of an organisation – its purpose.
  • By following a logical thread of questions out from development of the purpose, we see a set of inter-connected functions which must come together in harmony if the organisation is to win the metagame and achieve its aims.

Leave a Reply