# Decision Making Under Uncertainty

Many of you know that one of my favorite books is Superforecasting by Philip E. Tetlock and Dan Gardner. The book dives into the science of forecasting and gives frameworks for better forecasting. An important notion is that superforecasters are not off-the-charts brilliant or better forecasters out of the gate. Instead, superforecasters put deliberate effort into becoming better forecasters.

My fascination dates back to my days as a nerdy math major in college. I had decided to concentrate on pure math but then took an illuminating class: Decision Making Under Uncertainty. So many of life’s important decisions involve uncertainty, and one has to choose the best course of action based on imperfect information with unknown outcomes. It is just such a fascinating problem. After that class, I changed my concentration to Applied Math, specializing in Decision and Control Theory.

The mathematical and statistical concepts were challenging, but the hard part was translating the theory into real-life actions. For a while, I was learning many theoretical and technical frameworks that I thought would never be useful. Finally, while discouraged, I visited my thesis advisor, Professor Myron Fiering, to find a real-life problem to apply to my education. As with many great educators, he did not solve my problem but offered an observation and asked a Socratic question. “Have you noticed that successful people in almost any field are good expected value forecasters and risk calculators? How do you think they do it?”

Through that provocation, I found similar answers, but I was less precise than what Tetlock and Gardner wrote in their book. Superforecasters tend to have a growth mindset, willing to put forth a framework, incorporate new data, seek disconfirming information, and iterate or completely change their frameworks to improve forecast accuracy. Moreover, they break large and unwieldy problems into smaller and more defined components that are easier to solve, think of the world probabilistically, challenge dogmas, and are willing to update their priors in response to new information.

To make the best decisions, we express our forecasts in investment memos or spreadsheets for many business problems, whether an investment in a company, a business unit, a project, or a product. In these memos and spreadsheets across different industries and use cases, there are similar components: the probability of success, the size of the prize, and the total investment cost, with each element potentially having a range of outcomes. Of course, how you put it together is both art & science, but superforecasters strive to turn as much of the art into science.

Not all successful people express their frameworks in a spreadsheet. For example, Alex Honnold spoke at our offsite a few years ago and was asked whether he feared falling and dying. He is not without fear but still free climbs. Some of us may question Alex’s decision to free climb, but he is also an expected value forecaster and risk calculator, given his answers.

Alex mentioned that free climbing has high consequences because he will likely die if he falls. His job is, through deliberate practice, to reduce the probability of slipping to almost zero, which means knowing his path, having backup plans, and knowing the whole mountain like the back of his hand (pun intended) before he attempts a free climb. In more challenging sections, use more caution and move more conservatively. In more accessible and familiar areas, be willing to move more quickly and aggressively to improve his time.

Roelof closed our first Base Camp with a line that I still remember: “at the end of our life, we are a sum total of all our decisions.” So here is to improving and making better and better decisions every single day.