Epistemology in the trenches

We know very, very little about anything

This is a philosophical addendum to last week’s post!

Having both an MBA and starting a business gives you a unique view into two worlds: The world of theory and the world of practice. One of these worlds is real but indescribable, the other sounds right but generally isn't.

There is a difference between empirical knowledge (knowledge from experience) and theoretical knowledge (knowledge from books and theories): It is the difference between a craftsperson and an academic. To our detriment, society values the academic over the craftsman - we care way too much about credentials and degrees. 

Craftspeople learn through apprenticeships: They see how things work in practice. They tinker (they don’t “experiment”). When you ask them to explain why something works or came out the way it did, their responses don’t make sense to theory-minded people. Listen to an architect describing a building, a painter describing a piece of art, a musician describing a song. It is all intuition, judgment, feel. This is more real than theory: most things are impossible to understand or articulate even when they are right in front of you, and even when you made them happen.

Theoretical knowledge is acceptable knowledge, which is a problem because it is generally BS. The core problem with theoretical knowledge is that most things are just not knowable. It is nearly impossible to predict anything meaningful about the future, which few people disagree with. But it is just about as impossible to know anything about the past. There’s no law of nature that says reality must “make sense” to us. Reality just is. Things that work just work. Things that don’t just don’t. Reality is complex, nonlinear, and detailed in a way we can’t process. The stories we tell, the logic we use, the data we interpret: These make us feel like we understand things, like we CAN understand things. Yet we overestimate what we know and what we’re capable of knowing because our stories “sound right”, and as a result we make terrible, terrible decisions. 

“Theory” is mostly story-telling about what we think is true, or an attempt to narrate what craftspeople know intuitively but struggle to put into words. It is explaining the incomprehensible, predicting the unpredictable. Theory makes sense until it doesn’t, and when it ceases to make sense we tend to bend reality to fit the theories that “make sense” to us. There’s no problem with stories, just the assumption that our stories are generally nonfiction.

“Logic” and “reasoning” seem uncontroversially good. But consider the accepted logic that “People only buy to increase revenues or decrease costs”, which sounds reasonable but is completely wrong. Logic and reasoning are paper-thin simplifications that rarely map to our multidimensional, complex reality. It’s really hard to know any single thing. Use abstract logic with care.

On “experimentation” and “data” - these are also overemphasized versus tinkering and judgment in modern society. In startups, we will almost never have enough data to really know anything, and in fact data is likely to blind us. Experimentation design is almost always flawed - if all of social science is unreplicable, how are we to assume that our experiments can work? Data is an excuse to wait to take action. And even when we get statistical significance, the end result of optimization seems to always, always be Kafkaesque.

Embrace that we mostly cannot know things before or after the fact, focus on honing your judgment in practice, and get things done today.