Algorithms are everywhere. Usually, us humans teach them what to do and they learn what we prescribe. In this blog series I explore what we can learn from algorithms. I am certain that some of their structures and underlying principles are highly applicable to our daily lives. This is the first part of a weekly series that explores how we can improve our processes through learning from algorithms.
You might have come across the letters IO (sometimes styled as I/O) in the tech scene or as an URL ending of a particular cool startup. These two letters stand for “Input and Output”. Input and output are fundamental to every algorithm. The algorithm takes what it is handed, the input, and then performs some predefined actions. At the end, the algorithm returns an output.
This holds true for algorithms simple and complex alike. Let’s look at the straightforward “algorithm” of addition. It takes two numbers as an input and returns one number, namely their sum, as an output. A picture classification algorithm uses pictures as an input and outputs a word (“cat”, “dog”). For other algorithms the input can be even more complicated. A credit score algorithm, for example, takes your personal information, your credit history, and data from the council as an input and returns a single number, your credit score.
Each algorithm has a specific data type it can handle as an input. For example, the summing algorithm wouldn’t know what to do with your credit history. Likewise, each algorithm produces a specific type of output. A picture classification algorithm will always return a word; it will not spontaneously write a pop song about the object in the picture (unless you train it to). When many algorithms work together it is essential that the output of one algorithm can be used as an input for the next one and so on.
This very clear definition of input and output can be applied in our daily lives. When we are explicit about what we expect as an input and what the output looks like it helps us avoid friction. Processes run more smoothly if each step receives the input it needs and produces the correct output for the next step.
Let’s look at an example from a business setting, where two colleagues, Alice and Bob, are responsible for a social media post. Usually, Alice prepares the raw text and Bob formats it and sources a suitable image. Bobs expects the text to be fact-checked and free of typos so he can focus on the visuals. If Alice sends a half-baked draft, this means additional, unexpected work for Bob. If Bob suddenly wants Alice to find an appealing graphic, she has to read up on the internal guidelines for the picture and gain access to the relevant data bases. This takes time and resources. When they deviate from their usual process and its inputs/outputs, unnecessary work piles up and slows down the whole system.
Input and output are also relevant in a private setting. Think of the “algorithm” of doing the dishes. The input is clear: a kitchen with dirty dishes. So you load up the dishwasher, put it on, and dispose of the dishwasher tablet wrapper. But what about that wooden chopping board on the counter that cannot go into the dishwasher…do you clean this by hand? This depends on your desired output. If you define the output as “dishwasher is on”, then you don’t bother with the chopping board. If you expect “ all dishes clean” as the output, then you give the chopping board a quick rinse. Both options are perfectly fine by themselves, but when someone has a different expectation of the output than you, friction and conflict can occur. If you ask your partner to “do the dishes” and the chopping board stays dirty, you may be disappointed even if they are under the impression that action has been completed successfully.
It is helpful to explicitly talk about the input and output of actions we take in our daily lives. This is certainly not needed for everything we do, but when conflicts arise you may find it helpful to think about the input and output. Agree on what is acceptable and what is not – be that with yourself or with others. And this is how we learn from algorithms.
This article has also been published on LinkedIn.
Leave a Reply