Automating Intuition is Annoying but it’s a Good Idea

Why is a handmade chair so much more beautiful than a mass-produced one from Walmart? Why are people better than computers at so many kinds of decision-making? Same reason.

When we do anything — make a chair, or make a choice — we use intuition to some extent. Some of us are great at justifying our intuition with reasoning, so it seems like we make rational decisions. But usually, we don’t. Usually we’re really just making highly believable rationalizations of our at least partly intuitive decisions.


We’re capable of handling complex, multivariate calibrations intuitively. So the work we do when we work on something personally, intuitively, can be far more sophisticated than what we make when we try to reproduce that process analytically. The handmade chair is made through thousands of intuitive little actions, but the Walmart chair is designed once in a computer program, and built over and over, simply, on a factory floor. Our intuition outpaces our analysis pretty consistently — perhaps not in its correctness, but in its consideration of many factors and patterns. The handmade chair may be less consistent in its straight edges, but it gains the benefit — the intuitively considered complexity, the beauty — of each two-second swipe of the sandpaper its creator made without really “thinking”, until she felt the chair was finished.

Intuition is incredibly powerful. But there’s much to be gained from reconstructing, and changing, processes analytically. We have to do a lot of careful discovery and reverse engineering to even begin to mimic the results of intuition. It’s hard. Understanding how the chair is made is a totally different, challenging skill than actually making it. We need rich and diverse scenario testing, and comparison to intuition, to get good analytical reproductions that aren’t reductive, and start to match intuition’s results. The Walmart chair reduces the process down to a simpler one, and the result shows it.

But in that process of reconstructing intuition, we discover things we didn’t know about ourselves, and our processes, and what they produce — whether that’s a thing or a decision. Once we’ve learned those nuances — we can create change in our outcomes.

It’s very hard to change something you do intuitively. We understand so little about how we as humans make decisions and where our behavior comes from. (If attempting to understand that interests you, I recommend Behave.) Behavior modification is insanely hard, because even well-studied behavior is usually too complex for us to change easily. We bend over backwards creating strategies to trick ourselves into behavior modification — to train our intuition. (We read lots of books, create habit loops, time our every action, hold and evaluate everything we own.) It’s a huge and bizarre part of our culture. It exists because changing our behavior — our intuitive processes — is so essential, so often, to get the results we want. But we’re terrible at it.

Why is behavior change so essential, so often? Because there’s nothing perfect about intuition. Intuitive processes can be problematic because they’re too taxing to execute (when the handmade chair takes 500 hours to make), or they’re broken and create bad results (if the chair breaks when you sit), or both (as is the case with most hiring and staffing, the area I focus on, but that’s another story). When we try to change them, fighting our intuition’s inner workings, we struggle.

But once we’ve done the reconstructing work — the discovery and reverse engineering — we can externalize our most problematic processes. We can take them out of the dim lighting of our internal intuition, and put them under the spotlight, right in front of our eyes. Then changing them becomes much, much easier.

When we externalize processes, pulling them out of intuition and putting them into a form we can clearly see, we give ourselves a chance to rigorously validate their component parts, and to test improvements to them. We can get a lens on each piece of the process, rather than just the whole. That’s especially handy when we pretty clearly understand what’s wrong with them — like, let’s say, unconscious bias in job interviews — but also extremely handy when we don’t. When we don’t know why we’re having problems, picking through a process lets us search for the reason, and move toward a fix.

When the tool we use to do something — like make a decision, or a chair — is our intuition, then fixing that tool is really hard. When the tool we use is externalized — out of our intuition and into another format like an algorithm, let’s say, or an established furniture-making process — we can get the outcome right more of the time, and understand why.


What makes a handmade thing so beautiful is the intuition — the gorgeous complexity of the intuitive, personal attention it got in the way that it was made. That’s sometimes what makes social interactions beautiful too — they’re so much more complicated than the transactional way we often think of them analytically. But when the outcomes from an intuitive process — whether a piece of furniture or, let’s say, a hiring decision — aren’t good enough, the beauty we lose from intuition is well-outweighed by the value we gain from externalizing.

And that’s my pitch for automation, friends.