Two Ways to Grow: Learning and Unlearning

Just as the cool Internet kids say, bundling and unbuilding are the only two ways to make money in business. In the past three months, I spent a lot of time thinking about a framework for personal growth. And after listening to too much Ben Thompson during long walks, the framework of learning and unlearning keeps coming back to me.

And, in short, I think unlearning is way underrated as a key way to grow.

The Learning

This is the easy part to sell in a framework of personal growth. There is no shortage of inspirational quotes to keep us motivated to continue to learn. In particular, those of us who choose to pursue a PhD is by definition very committed to learning. In particular, in the PhD program, we have a very particular way of learning, which is best illustrated by Matt Might’s famous illustration:

A fun paradoxical effect of learning is that the more you learn, the less you feel you know. Of course, this feeling is strictly wrong. But part of what’s beautiful about learning is learning to be humble and recognize the vastness of human knowledge and most importantly the limitation of that knowledge. That’s why inevitably we learn to use very hedging language in writing, such as, “there is strong evidence that…”. Part of the learning process is to recognize that things are complicated and having the capacity to understand the nuances.

The Unlearning

What is understated, therefore, is the process of unlearning.

tweeted a couple weeks ago that I had an aha moment at work (product engineering). In research, what makes a “good” problem is completely different from what makes a “good” engineering problem. In research, if the solution is obvious, it is not a very interesting problem to solve. However, in engineering, if the solution is obvious, that is one of the best problems to pursue because it is so feasible and potentially reliable and easy to maintain. But when I came in wearing the lens of a researcher as I was trained, none of the engineering problems felt very exciting — the solution is obvious, I would shrug and complain. I had to unlearn and adapt to see things in a different light. Now I find a lot of problems super exciting because I can see where I need to go and it’s only a matter of joyful heads down implementation from here.

So the point I am trying to make is — at certain point we need to unlearn the things that once we spent so much energy to learn and hold dear to. At certain point the things that we tried so hard to learn become a thing that holds us back from seeing other possibilities. If there is no conscious effort to unlearn things, there is the danger of getting stuck in a very weird place trying to fight an impossible battle while tiring ourselves out.

Learning and Unlearning HCI

As a thought experiment, let’s see how one can learn and unlearn Human-Computer Interaction (HCI). In building a new socio-technical system such as a social network, one way to break the problem down is what we should v.s. can build:

An engineering-centric view would lead to the focus on can outweigh should:

Learning HCI means recognizing that this imbalance causes a lot of societal problems, and that these technical systems do not live in a vacuum and we should consider the human factors while not getting too excited by what we can build. Learning HCI means being more thoughtful in thinking about what we should build.

However, after a year in industry, when I read my past CHI papers, especially the design section (which in all honesty, I usually threw up my hands in the air and hope my advisor could fill them in last minute), I can’t help but find the ideas in there full of naiveté. When I call on practitioners to do certain things, I didn’t even consider that a) only a very small fraction of these practitioners have even heard of CHI and b) there are realistic constraints about the metrics, the OKRS, the performance reviews, the politics, etc. on the line that may all occupy priority before my little design recommendation.

Unlearning HCI means recognizing and owning the fact that to make a real impact, a lot of other factors need to be considered as well, e.g., business constraints.

As much as it is tempting to criticize big tech for only caring about profits, there is a super power that those of us with research background aspiring to make real-world impact needs to learn — to actually find that product-market fit for our research insights or prototypes. Smart Reply is s great example.

It is a super power to jump through all the hoops to get an idea up and running in the real-world. And to learn this super power, we need to do at least a little bit of unlearning of the research purity. Maybe a little bit of Machiavellianism is useful. But then the question to ponder next will be:

What about principles and ethics? This might be the question I will chew on in my long walks to come. Let me know if you have thoughts.

Thanks for reading!