# Understanding Uncertainty: 7 Decision-Making Tools for Predictions

Written on

Making predictions is a fundamental aspect of enhancing our comprehension of the world and improving our decision-making skills. This article aims to introduce several tools that can help refine our predictions, applicable to various scenarios—ranging from financial forecasts to personal decisions.

To clarify, we will define two crucial concepts:
- **Prediction Problems:** These involve estimating the likelihood of specific events occurring. The goal is to achieve the most accurate probability assessment possible.
- **Decision Problems:** Here, the objective is to choose the best option from a defined set of alternatives, typically the one that maximizes expected value.

A notable point is that every decision problem can often be reframed as one or more prediction problems. For instance, the question "Should I choose X or Y?" can be restated as "What is the expected value of choosing X?" and "What is the expected value of choosing Y?"

Making predictions combines both art and science. Although the aim is to generate quantitative outcomes, it fundamentally relies on qualitative insights and creative problem framing. The methods discussed here draw from Philip E. Tetlock's influential work, *Superforecasting: The Art and Science of Prediction*, which is a valuable resource for anyone seeking to enhance their predictive abilities.

It's essential to acknowledge that predicting the future is inherently challenging. While we cannot forecast events with certainty, we can estimate uncertainties as probabilities. The tools provided will assist in enhancing the accuracy of your predictions, offering valuable starting points rather than definitive answers.

Let’s delve into the first and foremost tool: Bayes’ Rule.

**Bayes’ Rule — Zoom Out Before Jumping In**Bayes’ Rule is a cornerstone of probability theory. For those unfamiliar, P(A|B) represents the probability of event A given that B is true, while P(B|A) denotes the probability of event B given A.

A classic example involves assessing whether Alice, a smoker, has cancer. If the general cancer rate is 1%, the smoking rate is 10%, and 40% of cancer patients are smokers, Bayes’ Rule helps us determine the probability that Alice has cancer, given her smoking status.

Most often, when making intuitive predictions, we neglect the importance of prior probabilities. For example, knowing that 40% of cancer patients smoke may lead us to assume smokers have a high cancer risk, overlooking the low overall incidence of cancer.

Forgetting to consider the prior probability—often referred to as “zooming out”—is a common pitfall in making accurate predictions. The key takeaway is: always start by estimating the prior probability before making any predictions.

Another illustrative case comes from

*Thinking, Fast and Slow*by Daniel Kahneman, which poses a question about a character named Steve, described in a way that fits the stereotype of a librarian. Most would inaccurately conclude that Steve is a librarian, neglecting the fact that there are vastly more male farmers than male librarians.**Bayesian thinking**involves starting with a prior and updating it as new information becomes available.**Laplace’s Rule — Will the Sun Rise Tomorrow?**Laplace’s Rule applies when dealing with events that have never occurred. The intuition is that if an event hasn’t happened before, its prior probability should be very low but not zero.

For example, if we want to estimate the probability of the sun rising tomorrow based on historical data, we can conclude it is exceedingly high, while the chance of it not rising is correspondingly low.

Caution is warranted when using Laplace's Rule, as repeated occurrences do not guarantee future outcomes. Let’s apply it to predict whether the stock market will crash by the end of 2021.

By examining historical patterns, we can ascertain that significant rallies have often been followed by sharp declines, suggesting a high probability of a market drop.

**Lindy’s Effect — Will the Bible Outlive Harry Potter?**The Lindy Effect suggests that the lifespan of non-perishable items (like ideas or technologies) can be predicted based on their current age. For instance, the Bible, having existed for millennia, is expected to last longer than a modern phenomenon like Harry Potter.

This concept is both intuitive and counterintuitive; intuitively, we might think that older items have a higher chance of survival, while counterintuitively, the predictions may shift over time.

**Debiasing — Play Devil’s Advocate**To gain an inside view, one must dissect the problem into smaller parts and analyze them from various perspectives. Recognizing biases and challenging one’s own assumptions can yield a more balanced viewpoint.

**Triangulation — Weigh Others’ Opinions**Triangulation involves synthesizing diverse opinions to form a more accurate understanding of a situation. It’s crucial to gather insights from multiple sources, especially experts, but remain aware of their potential biases and incentives.

**Expected Value — Make Rational Choices**The concept of expected value helps quantify the potential outcomes of decisions, allowing for better-informed choices. It’s important to consider not just the financial implications, but also job satisfaction, growth opportunities, and other personal factors.

**Brier’s Score — Calibrate Your Predictions**After making predictions, evaluating their accuracy is crucial. Brier's Score helps assess how well calibrated your predictions are, ensuring that probabilities align with actual outcomes over time.

In conclusion, mastering these tools enhances one’s predictive capabilities, enabling better decision-making in uncertain environments. For those seeking further insights, I highly recommend *Superforecasting* by Philip Tetlock and Dan Gardner, as well as the *Incerto* series by Nassim Nicholas Taleb, to delve deeper into randomness and the implications of improbable events. With practice, you can sharpen your ability to predict and make informed decisions, transforming uncertainty into opportunity.