Thinking in Bets by Annie Duke


I was first introduced to Annie Duke's concepts in a recent podcast with Stewart Butterfield interviewing Ben Horowitz and Marc Andreessen on the 10th anniversary of A16Z. In the interview Stewart asks Marc about any mistakes he believed he had made building A16Z. Marc starts off by referring to Annie Duke's work and the concept of resulting, which is the tendency to equate the quality of a decision with the quality of its outcome, which is a risky and incorrect thing to do with any decision that has a probabilistic outcome. And he shares how A16Z has significantly improved the quality of their decision making over the years by focusing on the very process by which they make their investment decisions in the first place.

As someone who deeply values being an infinite learner, I saw an opportunity to improve the quality of my learning cycles from decision making so decided to take Marc's recommendation and read Thinking in Bets by Annie Duke. Annie brings such a fascinating perspective to decision making based on her 20 years as a professional poker player. As someone who has had to make thousands of split-second decisions whose quality determined whether she immediately won or lost tens of thousands of dollars at a time, she's had the opportunity to refine her decision-making approach. And ultimately to apply it to the world of business in her more recent consulting work.

Throughout the work, Annie introduces us to a variety of cognitive biases that all humans suffer from that are incredibly detrimental to making high-quality decisions. I've already introduced you to the first bias, resulting, which is equating the outcome of a decision with the quality of the decision. This is dangerous because there are clearly circumstances where you have a good outcome, but it's mostly luck and not based on the quality of your decision. In that case, you haven't learned anything that's helpful for future decisions. At the same time, it's also quite possible to have a bad outcome but your decision making was actually spot on. But it's just so easy to lose this nuance and judge the decision solely on its outcome. Closely related to this bias is hindsight bias, which is the tendency after an outcome is known to treat the outcome as inevitable. In this case, we often forget that outcomes are probabilistic in nature and there was a chance the outcome could have in-fact gone another way.

Self-serving bias is also a fascinating one: we tend to equate negative outcomes resulting from our own decisions as due to luck, while we equate positive outcomes as due to our superior skill. Why do we do this? To bolster and protect our own egos of course. But here is an even scarier one: motivated reasoning. We often think that we are good at acquiring new information and adjusting our beliefs based on that new information. The reality is actually far from it. Motivated reasoning tells us that our existing beliefs actually affect the way that we process new information when we hear it. We are far more likely to alter our interpretation of that new information so as to ensure they fit our existing beliefs. Look no further than the current political environment and how each side becomes more entrenched through their respective interpretations of various news events. All of this goes to show the compounding detrimental effects of these cognitive biases on our likelihood of making high-quality decisions.

So how does one go about making high-quality decisions despite these challenges? Annie introduces us to the idea of treating decisions as bets as a mechanism to find learning opportunities in uncertain environments. It's easy to see why a poker game is clearly a bet. It's also often easy for folks to understand how investing in the stock market is a bet. But it turns out any decision which has an uncertain outcome is a bet, regardless of whether it's financial in nature. When we make a decision, we are betting whatever we value (happiness, success, satisfaction, money, time, reputation, etc) on one of a set of possible and uncertain futures.


Annie goes on to introduce us to the ideal learning loop. Having experiences and becoming an expert are two different things and in Annie's view, the only way to become an expert is by fine-tuning your learning loop. Annie encourages us to think about every decision we make as an explicit bet we are making based on our beliefs and then to post-mortem those decisions after the outcome is known to determine what part was luck and what part should be captured as new skills. By doing this explicitly with every decision, we maximize our learning and increase the chances of making high-quality decisions. It's critical to internalize that our beliefs are ultimately what drive what bets we make. So that way you realize that part of the skill in decision making comes from learning to be a better belief calibrator, using experience and information to more objectively update our beliefs to more accurately represent the world. To fight self-serving bias and motivated reasoning, we need to substitute the routine of truth-seeking into our learning loop instead of the more typical instinct to seek credit and avoid blame. Annie suggests decision groups as one critical tactic for ensuring you adhere to this process. Decision groups can help make the beliefs that are driving your decision more explicit to all members of the group. They can also help when post-morteming a decision, as long as you encourage diverse opinions to be shared and avoid groupthink.

Annie's work helped me appreciate all the cognitive biases that can easily result in poor decision making and her framework of treating decisions as bets and leveraging an ideal learning loop routine for every decision provided a path to reducing those biases and fine-tuning my decision-making process. If you are interested in getting serious about the quality of your decision-making, I would encourage you to also check out Thinking in Bets by Annie Duke.
Enjoyed this essay?
Get my monthly essays on product management & entrepreneurship delivered to your inbox.