Skip to main content

Outlier’s Path

Updating Your Priors

About a week ago, Bogomil arranged for some of us to meet with the leadership of Amazon Web Services (AWS). After the usual pleasantries, both sides exchanged their views of technology mega-trends. The conversation would only be complete these days by discussing artificial intelligence and machine learning (AI/ML). An interesting comparison that AWS CEO Adam Selipsky made was to compare AI/ML today to the Internet in 1999. Both will be bigger and more pervasive than what we can imagine. Foundational AI/ML technology is changing so fast. As a result, today’s conversation topics are vastly different from those we would have had three months ago and will have three months from now. In these times of accelerating change, we need to ensure we are constantly updating our priors.

Updating priors is one of my favorite simple yet powerful concepts from Bayesian statistics. You start with your prior knowledge, which can be nonexistent. You incorporate new information to update your prior knowledge when you receive new information. When you receive more information, you further update your prior knowledge. For example, take a GPS navigation system, which tracks your location, direction of travel, speed, traffic patterns, etc., to tell you the best route to drive to your destination and predict your arrival time. When traffic changes, it recalculates. When you make a wrong turn, it recalculates. Since the navigation system is powered by computation, it can constantly update its prior predictions.

Humans are much worse at automatically updating our priors. Of course, as we learn more, our priors should change. Unfortunately, most of us are lazy and stubborn and prefer to stick to our prior beliefs and habits rather than properly weighing prior and new information. Even when we do, we tend to do it too slowly. As Roelof mentioned in his WEF Summary, Larry Summers said, “one of the most common mistakes in investing comes from people doing today what they wished they had done yesterday.” Feel free to replace “investing” with whatever challenges you face today.

In arenas of little change, holding more firmly to your priors is appropriate. Such stubbornness may even be called conviction. However, it is essential to be flexible and update your priors more frequently in accelerating change, or you risk being left behind.

Since we began with AWS, I will leave you with a quote from Jeff Bezos about being both stubborn and flexible in the AWS reception area, “if you’re not stubborn, you’ll give up on experiments too soon. And if you’re not flexible, you’ll pound your head against the wall and you won’t see a different solution to a problem you’re trying to solve.”