Decision Risk Analytics’ Graeme Keith was recently invited by the Hamburg-based oil company DEA to present on their “Meet the Expert” forum.

This is a summary of Graeme’s presentation. The slides are available here.


Releasing your inner mathematician to make better decisions under uncertainty

Despite the title, this isn’t a mathematical presentation – at least not in the sense of having numbers or equations. The word mathematics comes from the Greek mathema, which literally means “what you have learned” as distinguished from what you know by intuition. It is mathematics in this broad sense that needs to be deployed to make good decisions under uncertainty.

More often than not, when facing decisions circumscribed by uncertainty, our intuition fails. There is no shame in this. It’s not because we’re stupid or because we weren’t paying attention in class. It’s because it is inherently extremely difficult to develop intuition about uncertainty.

We develop intuition by repeatedly interacting with systems and internalizing the feedback we receive relative to our evolving expectations. Uncertain systems, by their nature, give unpredictable and diverse responses, even to the same interactions. It’s only when we’ve interacted with an uncertain system tens, hundreds, sometimes even thousands of times, that we have any chance of understanding the disposition of these responses, which makes it incredibly difficult to develop intuition on uncertain systems individually, and by extension in on uncertainty in general.

When our intuition fails, we must bring to bear what we have learned, mathematics in its broadest sense.

Through three simple examples, this presentation introduces one simple, but incredibly powerful “mathematical” (in the broad sense) approach to decision making, namely causal mapping (very closely related to influence diagrams).

The language of causality is uniquely well-suited to decision analysis. We start with our objectives – what we are trying to achieve; our decisions – the levers we have to pull that can help achieve those objectives; and our view of the foggy, uncertain factors and parameters through which those decisions must influence those objectives. Causality gives us a framework to describe how our decision choices propagate through these uncertainties to effect our desired outcomes.

The examples show how the process of mapping causes can reveal logical pitfalls (inferring from effect to cause) and the sometimes obscure nature of the information we have available. Causal mapping also allows a form of virtuous simplicity whereby complexity can be rolled up and packed away without being cut away or ignored.

There are references and a bibliography at the back of the presentation.