The information age that we live in is powered by Algorithms. For most of us, the very word “algorithm” is intimidating and brings to mind the hidden, complex math that lies underneath.
Behavioural economists have pointed out the cognitive biases that make many of our decisions “less than rational” or “what-a-computer-would-never-do”. So can we use any of these smart algorithms for our everyday problems?
This interesting question has been addressed by Brian Christian., and Tom Griffiths “Algorithms to Live By: The Computer Science of Human Decisions”. The authors make a deadly combination of a Computer Scientist and a Cognitive Science Expert.
Here are a few algorithms mentioned in their book, which I find most easily and profitably applied in our lives
1. Algorithm for deciding when to stop looking for the best “xyz”
We all have faced this decision situation : How many times to circle the block before pulling into a parking space? How long to hold out for a better offer on that house or car? These set of problems are similar to the classic “Secretary Problem”, where you can only screen secretaries once. If you reject a candidate, you can’t go back and hire him/her later (since they might have accepted another job). The question is: How deep into the pool of applicants do you go to maximize your chance of finding the best one?
Deciding when to stop your hunt for the ideal parking space, apartment, or ideal dating partner or spouse is difficult for us humans. If you stop your selection with the first one that you saw, you leave the best ones undiscovered. When you stop too late, you were holding out for a better choice that may not have existed. And the search in this process is exhausting you down. With more information, you are better placed to make the right choice, but this will call for evaluating all the options first before you make the decision. This option is of evaluating all before deciding is not available here. So, when should you take the leap and look no further? What is the Optimal Stopping Point? An algorithm exists, which answers this question- 37 %.
The Thirty-seven percent rule
The mathematical answer to the optimal stopping point is 37% (to be precise: 1/e). It is also called the Look-Then Leap Rule. You set 37% of time or number of options for “looking”—that is, exploring your options, gathering data—in which you categorically don’t choose, no matter how impressive. After that point, you enter the “leap” phase, prepared to instantly commit to any option that outshines the best options you saw in the “look” phase.
So if you plan to decide on a best apartment in a month’s time, spend 37 % of your time (eleven days) exploring options non-committally, and after that point, get ready to immediately commit to the very first place you find that beats whatever you’ve seen already.
2. Algorithm for deciding should we try new things or stick with our favourite ones?
Every day we are constantly faced with this dilemma of choosing between what you know to be good (‘exploitation’) and choosing something you aren’t sure about and possibly worth trying out (‘exploration’). For instance, trying out a new restaurant in the town against sticking with the one(s) that have satisfied your taste buds in the past.
Our understanding of life and instincts make us believe that life is a balance between novelty and tradition, between the latest and the greatest, between taking risks and savouring what we know and love. Not exploring is no way to live and never exploiting can be every bit as bad. So the question unanswered remains: where lies the balance?This is the classic Explore/Exploit Trade-off problem.
Computer scientists have come up with various algorithms to solve this problem, but the simplest approach to finding a balance to explore/exploit trade-off is “Seizing the Interval”.
Seizing the Interval
Exploration is gathering information, and exploitation is using the information you have to get a known good result. So explore when you will have time to use the resulting knowledge, exploit when you’re ready to cash in. The interval makes the strategy.
When balancing favourite experiences and new ones, nothing matters as much as the interval over which we plan to enjoy them. The value of exploration lies in trying new things but this value of finding a new favourite through exploring only goes down with time, as the remaining opportunities to experience (or exploit) it, declines with time.
Being sensitive to how much time you have left is exactly what the algorithm of the explore/exploit dilemma suggests.
Say you are relocating and leaving a city, it makes more sense to go back to all your old favourites, rather than trying out new stuff… Even if you may end up finding a slightly better place, why take the risk when you are anyway leaving the city? Discovering a delightful café on your last night in a town, doesn’t give you the opportunity to return. You are better off trying a new restaurant when you move into a city than when you’re leaving it.
So explore when you will have time to use the resulting knowledge, exploit when you’re ready to cash in. The interval makes the strategy. This means that over time we’re usually better off sticking with experiences we know than experimenting with what we don’t.
We see this algorithm being played out all around us without really understanding what is happening. The not so young (the old) keep going to their old favourites, show clearly defined preferences, while the young keep experimenting. Now we should understand, that the old are in the “exploit” phase- they have done their share of experimentation (“explore”). The dynamism comes from the explore phase. So what happens when everyone reaches the state of “exploit” at the same time?
Tyler Cowen in his recent book, “The Complacent Class: The Self-Defeating Quest for the American Dream“, laments the lack of dynamism in the USA. One reason is of course that US is aging.
3. Algorithm to decide if the effort in arranging stuff makes sense
I remember classifying my emails in Outlook/Gmail in nicely organized folders with the names of projects, customers, prospects that I handled. I don’t do this anymore and never realized that I was applying this algorithm- the sort-search trade-off.
Weigh the amount of time you spend organizing against the amount of time it takes to use the search to retrieve. The search capability with time has become extremely efficient. The sorting process will take you more time than the time you’d spend searching for whatever you need, when you need it.
The algorithm in this case, called the Sort-Search Trade-off, points to leaving your emails in blissful disarray. The search-sort trade-off suggests that it’s often more efficient to leave a mess.
Christian and Griffiths explain why you shouldn’t bother trying to organize and manage your emails. Organizing your desk/workspace is also inefficient. Most likely it’s already in optimal order, because our desk piles tend to be stacked reverse chronologically. The sort-search trade-off algorithm in this case tells us that the last thing you touched is likely the next thing you’ll need, so if you want to save time looking for stuff, leave it that way.
We come across problems every day that forces us to make difficult choices. Our inherent biases hardwired in our brains, makes our decisions less than optimal. What if we find solutions to everyday problems from algorithms/computer science? Thinking algorithmically about the world, understanding the fundamental structures of the problems we face can help us see how good we actually are, and better understand the errors that we make.
Optimal stopping tells us when to look and when to leap. The Explore-Exploit trade-off tells us how to find the balance between trying new things and enjoying our favourites. The Sort-Search trade-off tells us when our effort in arranging stuff makes sense.
It is important that we use and understand these and similar algorithms so that we are better prepared for the age of the machines.