Introduction

Brian Christian and Tom Griffiths’ 2016 book, Algorithms to Live By: The Computer Science of Human Decisions, explores how algorithms created for computers can help solve common decision-making problems in daily human life. They use accessible language to explain complex mathematical concepts and provide actionable insights based on these algorithms. The book spans various areas of computer science, demonstrating that these principles can extend effectively into personal decision-making with concrete real-world examples.

1. Optimal Stopping

One of the first algorithms discussed in the book is the Optimal Stopping problem—when to stop looking for a better option and commit to a current choice. This is exemplified through the “Secretary Problem,” where an employer wants to choose the best secretary from a pool of applicants.

Concrete Example: The authors suggest interviewing and rejecting the first 37% of the candidates. After this, you should hire the next best candidate who is better than all previous ones. This percentage, derived from the formula 1/e (where e is the base of natural logarithms), maximizes the probability of selecting the best candidate.

Specific Action: When apartment hunting or hiring someone, review and reject the initial 37% of options. Then, commit to the next option that better meets your criteria.

2. Explore/Exploit Trade-off

The Explore/Exploit trade-off addresses the dilemma between exploring new options and exploiting existing ones. They discuss the multi-armed bandit problem, where each arm of a slot machine has a different, uncertain payout. Balancing exploration (trying new things) and exploitation (sticking with what works) is key.

Concrete Example: In contexts like choosing restaurants, Christian and Griffiths recommend the balance of exploration and exploitation based on the duration of stay. If you’re new to a city, explore more; if time is limited, exploit known good options.

Specific Action: Apply the “Upper Confidence Bound” strategy, which means choosing options that have the best potential outcome along with calculating confident intervals around these estimates. For example, try new places early in a move to a new city, then stick with your favorites as you settle in.

3. Sorting Problems

Sorting efficiently is a common computer science problem with real-world implications. For example, the sorting algorithm can explain better ways to organize books, files, or cooking supplies.

Concrete Example: Using techniques like “Merge Sort” or “Quick Sort” for organizing books. Create small categorized sections, then merge them slowly into larger, organized collections.

Specific Action: When organizing a collection, start by grouping items into smaller categories, then combine these groups in an orderly fashion to create a well-organized whole.

4. Caching

Caching involves storing frequently accessed information for quick retrieval, which can be translated into optimizing everyday decisions.

Concrete Example: The Least Recently Used (LRU) algorithm decides which information to cache by keeping recently used information and discarding the least recently accessed items. A practical application is organizing a fridge so frequently used items are in front.

Specific Action: Regularly re-evaluate the placement of commonly used items in your workspace, kitchen, or digital files to ensure that the most frequently accessed are the easiest to reach.

5. Scheduling

The authors offer insights into various scheduling problems. The “Shortest Processing Time First” (SPTF) rule suggests handling the quickest tasks first.

Concrete Example: In managing tasks at work or home, knocking out small tasks first can create momentum and clear the way for larger tasks.

Specific Action: List daily tasks and prioritize them by duration. Complete the shortest tasks early to free up more time and mental energy for larger projects.

6. Bayes’ Rule

Bayes’ Rule helps refine probabilities as new evidence appears, guiding better predictions and decisions.

Concrete Example: Healthcare applications often use Bayes’ Rule, for instance, to update the probability of a disease given new symptoms. This principle can help make informed decisions under uncertainty.

Specific Action: Use Bayes’ Rule in daily decision-making by updating your beliefs as you gain new information. For example, if multiple weather forecasts predict rain, adjust your plans considering the latest updates.

7. Overfitting

Overfitting occurs when a model too closely follows a particular set of data, losing its general applicability. It’s comparable to making decisions strictly based on past experiences without adaptability.

Concrete Example: In sports, if a coach only uses game plans that worked previously without adapting to new teams’ strategies, it reflects overfitting.

Specific Action: Balance between learning from past experiences and adapting to new situations. Maintain flexibility in personal and professional strategies to ensure they are adaptable to future changes.

8. Game Theory and Nash Equilibrium

The book discusses Game Theory, especially Nash Equilibrium, where individuals make decisions based on optimizing their outcomes given others’ choices.

Concrete Example: In negotiations, like salary discussions, understanding that each party aims to optimize their gain can help in strategizing approaches that acknowledge mutual benefit.

Specific Action: Use game theory to predict and influence outcomes by considering each participant’s strategies. In negotiations, aim for win-win situations where each party’s decision benefits the overall outcome.

9. Randomness and Monte Carlo Methods

Monte Carlo methods use randomness to solve problems that might be deterministic but are complex to compute directly. This approach can be applied to various decision-making scenarios.

Concrete Example: Financial portfolio management uses randomness to simulate different investment outcomes, providing a range of possible future scenarios.

Specific Action: When faced with complex decisions, simulate different scenarios using random inputs to gain a broader perspective on potential outcomes. For instance, use random sampling to estimate the potential effectiveness of marketing strategies.

10. Networking and the Power of Weak Ties

Drawing on sociological research, the authors discuss the impact of weak ties versus strong ties in a network, indicating that weak ties often bring more novel information and opportunities.

Concrete Example: Job searches often benefit from weak ties as they connect people to different circles and new opportunities.

Specific Action: Foster and maintain a wide network beyond close friends and family. This can be achieved by attending industry events or engaging in online communities where brief interactions can lead to significant opportunities.

11. Overcoming Bias

Recognizing and overcoming cognitive biases is a critical part of making better decisions. The authors explore how algorithms can help identify and counteract these biases.

Concrete Example: The Confirmation Bias can be addressed by intentionally seeking out information that challenges your preconceptions.

Specific Action: Implement decision-making practices that force consideration of contrary viewpoints. When making a decision, actively research opposing opinions before finalizing your viewpoint.

12. Multi-Touch Attribution in Marketing

The book extends algorithms to areas like marketing, where multi-touch attribution models help understand consumer behavior across touchpoints before a purchase.

Concrete Example: Companies often use attributed algorithms to determine which marketing channels contribute most to sales conversions.

Specific Action: Analyze different touchpoints in your marketing strategy to determine which ones have the highest conversion rates. Adjust strategies to focus more on these touchpoints for efficient resource allocation.

Conclusion

Algorithms to Live By furnishes readers with a suite of computational strategies to streamline and improve various aspects of daily life, grounded in rigorous computer science principles. These methods, from optimal stopping to embracing randomness, provide actionable steps individuals can take to enhance decision-making efficiency. By adopting these computational techniques, one can approach choices systematically, improving both personal and professional outcomes.