The Decision Maker: Annie Duke

How do you analyse the decision-making process?

When you think about what route to take, you are predicting that ‘route A’ will get you there faster than ‘route B’ – but we understand it is not a guarantee. We are trying to think about what the future might hold, but we have to imagine several futures. It is a three-step process. The first is comparing options, then trying to figure out which is the most likely to get me where I want to go. The third is to mitigate the risks, because sometimes things won’t go my way.

What are some of the biggest mistakes we make when it comes to risk?

One of the biggest mistakes is saying we should have seen things coming. Sometimes there is sufficient information to warrant you seeing something coming – in the sense that you should have recognised it was a possibility. Nothing is ever guaranteed. Once the world starts observing a particular outcome, hindsight makes it seem like the only thing that could have occurred, and therefore we should have seen it coming. Both looking ahead and backwards, we should not think about the world as if there is only one possible outcome. We should try to imagine as many possibilities as we can, examine them all, and then try to take lessons.

How can we make better decisions in situations of uncertainty?

There are two influences on the way our decisions turn out. The first is luck. If I think about a decision where 98 per cent of the time I will have a good outcome but two per cent of the time something bad will happen, once I make the decision the outcome is a matter of luck. Luck is very interesting from a decision-making standpoint: you need to really embrace it.

The other influence is imperfect information. All sorts of imperfect information influence our decisions – some of it inaccurate and some of it missing. When we make decisions by modelling the future, the more predictive our models can be, the better our decisions. In contrast, when there isn’t much information, our model predictions will be very wide.

One can assume models will tighten as we get more information, but bad things can happen in the meantime. It is common to think that given enough time we can obtain certainty over our decisions. But we can’t. You need to start thinking about decisions as a trade-off between time and certainty, and become comfortable with being uncertain at the moment you make a decision.

We also delay painful choices. This is one of the big findings of Daniel Kahneman and Amos Tversky’s work on prospect theory. Essentially, if you owe me 100 dollars and I ask if you want to flip a coin, double or nothing, not only will you say yes but you will pay me for the opportunity. On the flipside, if I owe you 100 dollars and offer to flip a coin, you will refuse and ask for your 100 dollars then and there. There is a clear asymmetry.

We can think about it in the same way in terms of our choices around lockdowns during COVID-19. They were painful and nobody seemed to adopt ‘the earlier, the better’ approach and these behavioural quirks played a key part.

How important is diverse thinking in decision making?

You want to have different perspectives because our beliefs are flimsy. If you think about it, there is a universe of things you don’t know. What you want to do is take a ‘random walk’ through that universe in a way that maximises your ability to collide with corrective information.

The problem is the way we walk through that universe is not actually random. We see mainly what agrees with our pre-existing views and find all sorts of ways to dismiss any information that happens to disagree. This is also true for teams. Teams form an identity by reaching consensus quickly because it feels good. What often happens is that we hire for a diverse group but then ask everybody to think the same.

The other reason why diversity is important is that we know data does not equal truth. Two people can look at the same set of facts and come up with different conclusions, because people have different models. Your best view of the world is to take all those models and allow them to collide, because none of them are going to be wrong or right. They are all going to be better or worse in different ways.

This is important in strengthening the beliefs that inform our decisions. Creating a cultural shift in team identity towards tolerance of diverse perspectives is also crucial to improving things. Another way is to use a system solution. Let’s say I want the team’s opinion on GDP growth in Q2. Instead of asking the question in a meeting, I email each member of my team and ask them to email me back privately. There can’t be any groupthink because they cannot possibly know if they are agreeing with me, or with one another.

I now have the whole range, which I then put into a document with no names, so you don’t have the high-status influence, and I send it to the group. Now everybody will understand the opinions of the group, and you will get that spread of knowledge. What is nice about that system is that it creates the cultural shift as well, because people start to see the value in seeking a diversity of opinion.

How should we think about risk?

If you don’t embrace uncertainty, you cannot possibly think appropriately about risk. This ties in to two points. First, by not accessing a diverse range of opinions, it is likely that you will miss risks: nobody wants to be the negative person on the team, so you need to create comfort for people who may have a more pessimistic view.

Second, we will always underestimate the worst-case scenarios, partly because we imagine we have more control over the outcome than we do. One way to address that is to get your whole team to imagine we fail and the probability of it happening. That helps you to see the downside could be worse, and more probable, than you think.

People tend to avoid that, but it allows you to do three things. One, you may change your decision in order to de-risk. Two, you might look whether hedges are available. And three, sometimes hedges aren’t available, and you wouldn’t change your decision, but you can have an action plan ready. Otherwise, when the downside happens you waste time and generally decisions are poor. Having the ability to plan is incredibly valuable.