Algorithms to Live By Summary

How do you know when to stop searching for the perfect partner, apartment, or job? Algorithms can help you decide.

1. Algorithms Are Everywhere – Even in Your Daily Life

Algorithms aren’t just for computers; they’re a part of how we think and make decisions. An algorithm is simply a set of steps to solve a problem, and humans use them intuitively all the time. For example, when you follow a recipe, you’re using an algorithm to create a meal. Similarly, when you weigh the pros and cons of a decision, you’re applying a mental algorithm to reach a conclusion.

The book explains that while human algorithms are often imprecise and subjective, they serve the same purpose as computer algorithms: to find solutions. For instance, when apartment hunting, you might have a mental checklist of criteria like rent, location, and size. Once an apartment meets your standards, you stop searching and sign the lease. This is a human version of an algorithm in action.

The beauty of algorithms is that they can be borrowed from computers to improve our decision-making. By understanding how computers solve problems, we can apply similar methods to our own lives, making better choices in less time.

Examples

• Following a recipe to bake a cake is an algorithm in action.
• Using a checklist to decide whether to accept a job offer mirrors a computer’s decision-making process.
• Apartment hunting with specific criteria is a human version of an algorithm.

2. The 37% Rule: Knowing When to Stop Searching

One of the most useful algorithms for decision-making is the optimal stopping rule, which helps you decide when to stop searching and commit to a choice. The rule states that if you’re evaluating a set number of options, you should spend the first 37% of your time or options gathering information without making a decision. After that, choose the first option that meets your standards.

This rule works because it balances exploration and commitment. For example, if you’re looking for an apartment and plan to visit 100 options, you should spend the first 37 visits learning what’s available. After that, you commit to the first apartment that meets your criteria. While it doesn’t guarantee the absolute best choice, it gives you the highest probability of success.

The 37% rule applies to many areas of life, from dating to job hunting. It’s a practical way to avoid endless searching while still making a well-informed decision.

Examples

• Apartment hunting: Look at 37% of options before committing to one.
• Dating: Spend the first 37% of your dating life learning what you want in a partner.
• Job searching: Interview for 37% of roles before accepting an offer.

3. When to Explore and When to Exploit

Life often presents a dilemma: Should you stick with what you know or try something new? This is known as the explore-exploit tradeoff. Algorithms like the Upper Confidence Bound can help you decide. This method suggests starting with exploration to gather information, then gradually shifting to exploitation, where you stick with the best option based on what you’ve learned.

For example, in a casino, you might try several slot machines to see which one pays out the most. Once you’ve identified the best machine, you focus your efforts there. This approach also applies to dating, where you might meet several people before committing to the one who aligns best with your values and goals.

The explore-exploit tradeoff is also used in fields like medicine, where doctors test multiple treatments before settling on the most effective one. It’s a powerful way to balance risk and reward in decision-making.

Examples

• Slot machines: Test several before sticking with the one that pays out most.
• Dating: Meet different people before committing to a relationship.
• Medicine: Adaptive clinical trials test multiple treatments to find the best one.

4. Sorting Isn’t Always Necessary

Not everything in life needs to be perfectly organized. Sometimes, a little chaos is more efficient. For example, if your desk is messy but you know where everything is, there’s no need to spend hours tidying up. Algorithms like Least Recently Used (LRU) show that the most important items naturally rise to the top, making them easy to find.

However, if you do need to organize, algorithms can help. The bubble sort method organizes items one pair at a time, while the insertion sort method involves placing items in their correct order as you go. For large collections, the merge sort method divides items into smaller groups, sorts them, and then combines them.

These sorting methods can save time and effort, whether you’re organizing your bookshelf or managing digital files. But remember, sometimes it’s okay to embrace a little disorder.

Examples

• A messy desk often has the most-used items on top, making them easy to find.
• Bubble sort: Organize books one pair at a time until they’re in order.
• Merge sort: Divide a large collection into smaller groups, sort them, and merge.

5. Scheduling Your Time with Algorithms

Managing your time can feel overwhelming, but algorithms can help. The Earliest Due Date algorithm suggests starting with the task that has the closest deadline. If you’re running out of time, Moore’s Algorithm advises skipping the longest task to maximize overall productivity.

However, scheduling itself can become a time sink. The book recommends focusing on one task at a time and avoiding distractions like emails. Multitasking burdens your working memory and slows you down. By concentrating on one thing, you can accomplish more in less time.

While no algorithm can solve every scheduling problem, these methods provide a framework for tackling your to-do list efficiently.

Examples

• Earliest Due Date: Start with the task that’s due soonest.
• Moore’s Algorithm: Skip the longest task to get more done overall.
• Focus on one task at a time to avoid the mental cost of multitasking.

6. Predicting the Future with Bayes’s Rule

Bayes’s Rule is a mathematical method for predicting the likelihood of future events based on past data. It’s a way to update your beliefs as new information becomes available. For example, if you buy three lottery tickets and all are winners, you might reasonably assume that most tickets in circulation are winners.

This method is also used in fields like medicine and finance. Doctors use it to predict the effectiveness of treatments, while investors use it to assess the likelihood of market trends. By understanding probability and distribution patterns, you can make better predictions in your own life.

Bayes’s Rule shows that the more information you gather, the more accurate your predictions become. It’s a reminder to stay curious and keep learning.

Examples

• Lottery tickets: Use past results to estimate the proportion of winners.
• Medicine: Predict treatment effectiveness based on patient outcomes.
• Finance: Assess market trends using historical data.

7. Handling Data Overload

In today’s world, we’re bombarded with information. Algorithms like Exponential Backoff can help manage data overload. For example, if a server is overloaded, this method suggests waiting a few minutes before trying again, doubling the wait time each time you fail.

Another useful algorithm is Additive Increase, Multiplicative Decrease (AIMD), which helps determine the maximum amount of data a network can handle. It starts small and gradually increases until the system reaches its limit.

These methods can also be applied to personal productivity. If you’re overwhelmed, start small and gradually increase your workload as you find your rhythm.

Examples

• Exponential Backoff: Wait longer between attempts to access an overloaded server.
• AIMD: Gradually increase data flow until the system reaches its limit.
• Personal productivity: Start with small tasks and build momentum.

8. Guiding Decisions with Game Theory

Game theory explores how people make decisions in strategic situations. The prisoner’s dilemma is a classic example, where two people must decide whether to cooperate or betray each other. While betrayal often seems like the best choice, it leads to worse outcomes for both parties.

Mechanism design is another branch of game theory that forces people to behave in desirable ways. For example, if employees aren’t taking vacation time, making vacations mandatory ensures they get the rest they need.

These concepts show that understanding human behavior can help you design better systems and make smarter choices.

Examples

• Prisoner’s dilemma: Betrayal leads to worse outcomes for both parties.
• Mechanism design: Mandatory vacations ensure employees take time off.
• Strategic decision-making: Use game theory to predict others’ actions.

9. The Limits of Algorithms

While algorithms are powerful, they have limits. Overfitting occurs when a model becomes too complex and loses its ability to adapt to new data. For example, if you create a model to predict obesity and overemphasize location, it may fail when applied to different regions.

The book advises aiming for “good enough” solutions rather than perfection. For example, the traveling salesman problem, which seeks the shortest route between multiple points, becomes unsolvable at large scales. Allowing for slight inefficiencies can save time and effort.

Understanding these limits helps you use algorithms wisely without expecting them to solve every problem.

Examples

• Overfitting: A model that overemphasizes location fails in new regions.
• Traveling salesman problem: Allowing inefficiencies makes the problem manageable.
• Aim for “good enough” solutions to save time and effort.

Takeaways

1. Use the 37% rule to decide when to stop searching and commit to a choice.
2. Focus on one task at a time to maximize productivity and avoid distractions.
3. Embrace “good enough” solutions instead of striving for perfection in complex problems.