What can algorithms teach us?

This is the subject of “Algorithms for Living – The Exact Science of Human Decisions” by Brian Christian and Tom Griffiths, a very original and informative book.

In a world where artificial intelligence is advancing by leaps and bounds and, in the process, frightening many people, it is unusual to find an analysis of what we can learn from such developments. After all, what have the countless algorithms that computer science has given us in recent years taught us? How can we use these teachings in everyday life?

Remember that an algorithm is a series of finite steps that are used to solve a problem. Surprisingly, many of the problems we face in computer science are not very different from the everyday challenges of most humans. Therefore, the algorithms created to solve them can help us make better decisions in our lives.

This is the subject of the book.”Algorithms for Living – The exact science of human decisions“, in Brian Christian that it Tom Griffith (2017, The Arts Company). In it, the authors outline several algorithms created to solve computational problems, explain the logic of each, and draw unusual and interesting parallels to problems faced by the vast majority of people. The result was a very original and informative book, as well as a dense one.

Below, I’m including a summary along with a few of the countless insights I gained from this reading.

When does the search end?

Let’s say you are looking for an apartment to live in. The more visits you make, the greater your pool of information for making a decision, but also the greater the chance of missing out on the best available apartment. How many visits to do then?

Computer Science has an exact answer: 37% of the selected properties. That is, the optimal solution is to visit 37% of the selected properties without obligation and then choose the first apartment that exceeds (almost) all previously visited properties.

There are several problems with the same chassis, called “optimal stopping problems”, to which the 37% rule also applies. Two more examples:

• How long are you waiting for an offer to buy your car? Answer: 37% of the maximum acceptable time.

• How many interview candidates before submitting a job offer? Answer: 37% of the selected resumes.

In each case, the 37% rule strikes the right, mathematically optimized, balance between impulsiveness and overthinking.

Try a new one or continue with what’s your favorite already?

Exploring something new, in general, has value, because in doing so there is always a chance of discovering something better than what is currently seen as the best option. But the value of exploring the new depends on how much time you have to take advantage of any discoveries. If you’re short on time, it’s best to stick to your favorite picks for now.

Would you try a new restaurant or go to your favorite restaurant if you knew that after the meal you would be moving to a new city?

One way to think about this kind of problem is to think about how much we deduct into the future. With discount rates so high, it’s best not to explore and stick with your current top picks.

Algorithms created to determine the degree of exploration versus preservation of current choices consider the useful time of exploration outcomes (young people should explore more than older people), as well as the usefulness of exploration (a rapidly changing world requires more exploration), among other factors. The book discusses many of these algorithms.

Keep some mess

One of the classic problems in computer science is determining the time you spend ranking versus searches (think of the Google search engine).

Is it worth putting books on the shelf in alphabetical order? Probably not: the time you spend searching for a book with your eyes will generally be less than the time you spend arranging books with your hands.

In this spirit, Computer Science Algorithms shows that procrastination and letting a degree of chaos can be more than an easy choice — it can be the perfect choice.

But what if you have a lot of bookshelves and you frequently search for certain books? In this case, the benefits of somewhat organizing books may be worth it. But the perfect arrangement will not be the perfect choice.

A sea of ​​advice

Christian and Griffiths’ book contains many other practical suggestions, based on computer science algorithms, for improving decisions in a variety of situations. It’s impossible to do them all justice, but here’s a short list of a few more:

• Physical Space Management: In the absence of space (a typical problem with computers), (1) get rid of what has not been used recently – this is the optimal algorithm, and (2) store things near where they are most likely to be used

• Time management: If a new task comes up, divide its importance (measured by your criteria) by the time it should take you to complete it. If the score is higher than the task you are currently doing, go to the new task.

• Under uncertainty, think less and simplify: When uncertainty is very large and data is limited, prediction models should be as simple as possible (such as equal probabilities for each outcome) in order to avoid overfitting with the available data

The authors also address issues related to randomness, networking, strategy games, and computational efforts in human interactions, among other topics.

In short, the book shows that we have a lot to learn from computer science and that the algorithms it has developed can greatly improve decision-making in different situations. More than just threatening us, these advances allow us to better deal with the important and complex problems that many of us face.

Really worth reading!