Machinery
My last job taught me to hate spreadsheets.
But that’s because my last job sucked and I wasn’t interested in it.
But I’ve learned something about myself through building dozens of Excel files these last few months. And it’s that I like to build good spreadsheets. Not just spreadsheets, but good spreadsheets. Things that are robust and functional. I’m not a quant guy — I have very little interest in the math itself, or in complex computer models. What I am interested in is building a spreadsheet that is fully, 100% automated.
(Don’t worry, this whole post is not about spreadsheets.)
When I’m building a file for some research project Brent and I are doing, I’m not just looking for the output. I’m not just hurrying to the finish line and taking whatever shortcuts are necessary to get there as fast as possible. That’s what I would have done in high school. Make a messy spreadsheet, hurry up, get result, pass notes to pretty girls. But now it’s different. I’m building machinery. I’m building a fully reusable apparatus that, once completed, can take in any of a particular type of data and spit out precisely what I tell it to. Updates not required.
It’s immensely satisfying. To take a simple hypothesis (or a complex one), build out rows and columns of formulas that, once assembled, can handle new input over and over again. And that includes third- and fourth-degree cell references, data indexing, and auto-refreshing results.
Once the machinery is built, I don’t have to do anything other than decide what my inputs are. If I actually take the time to build a robust system, I save myself time and pain in the long run. Short-term sacrifice for long-term benefit.
This is how a lot of things in life work. With machinery.
Machinery that you can build, or learn, or find, yourself.
When you’re learning a new discipline, it can be counterproductive to jump straight into problem solving and specific cases or theories. It’s better to map out the basics, and then be able to assess every, or nearly every, possible scenario through the lens of the basic rules. If you were going to learn chemistry, it would be silly to try to learn about specific organic solvents and extractions before first learning about basic properties of organic versus inorganic compounds. That’s precisely what I did at my last job. And for a good long time I had no idea what I was doing. I’d learn the specific problems of the given situation, and then I’d have to work (and learn) backwards from there.
If you were going to learn about business, it would be silly to learn managerial techniques before first learning about income statements and balance sheets. You’d have to awkwardly, backwardly expand your field of vision to account for the things you never learned, rather than grasping the fundamentals and then being able to tactically narrow your focus. You’d fumble around trying to reverse-engineer good business concepts because you started halfway down the field when you were not ready for that at all.
If you don’t have the right frameworks of understanding in place, every situation feels like a case study. Case studies are great, but not when everything is a case study.
But if you do have the right frameworks in place, everything resembles something else. Everything resembles something you’ve seen before. There are good, useful frameworks that underlie just about everything.
A couple of years ago when I became interested in stocks, finance, and global markets, a friend of mine recommended an online course to me. It was free. Because, incredibly, hilariously, there is a treasure trove of free online courses now from great universities and great professors. Coursera, HarvardX… just use Google. They’re just sitting there, waiting to be taken. My friend recommended this course to me called The Economics of Money and Banking. And it was life-changingly good. Not only because the professor was excellent, but because it gave me a large section of the framework I needed for understanding money. I mean, really truly understanding money and what it’s doing in the economy.
For instance, I learned the different foundational views of economics from thinkers like Keynes and Bagehot. And I learned that shadow banking is money-market funding of capital-market lending. And I learned that literally our entire economy is made of grossly irresponsible leverage and none of it is real.
These were pivotal to my understanding of markets. I spent a couple of months studying economics and money in my free time — not really because I enjoyed it, but because I wanted the machinery. I wanted to have the baseline skills and understanding that would allow me to understand money generally, and use that to then narrow in on more specific issues. Without this framework, I would be stumbling from economic thought to economic thought trying to piece this all together in reverse. Not good.
Here are some more examples of machinery that have served me well and/or serve other people well:
Trauma does not mean “unique, life-shattering memory” or “one-of-a-kind horrific experience.” Trauma, in a basic psychological sense, means “something that you haven’t emotionally and mentally digested yet.” But everybody has a metabolism for emotions and thoughts. They just require a bit of figuring out, in a way that our physical metabolism doesn’t.
Most addicts and behavioral health patients are so lost for so long because they don’t understand how to understand what happened to them. Or what’s going on inside their heads. They don’t have the machinery for that.
But there is, in a general sense, a recipe for digesting trauma. That’s what clinical psychologists are trained for. Their job is to take the map they have of the human psyche, and their time-tested baseline assumptions for assessing someone’s emotions, and to map you as a person onto them. They see patterns over and over and over in people’s behavior. Addicts, drunks, divorcees, people who were abused as children — there are patterns everywhere. Which is why they’re often such miracle workers. They recognize in you the same things they’ve seen in others, and they have machinery they can run you through to get the desired result — which is healing.
If every case of human suffering was unique, we would be absolutely hopeless. Nobody would be able to fix us, and we wouldn’t be able to fix ourselves. Thank God there’s machinery for that.
And I also know that it can be a bit jarring for someone to say “your trauma is not unique. Your suffering is not unique. You are not unique.” But… it’s not. And you’re not. After a while most people tend to see that as a comfort, not a slap in the face. Because it means there’s a way out. Through doing what others have done.
How about in learning a language? You learn sentence structure at the same time you’re learning small amounts of new words. And you learn how to conjugate verbs at the same time as you’re learning how to talk about nouns. You’re not just learning how to speak foreign words — you’re learning how to construct the language the way a native speaker would. Language is taught starting with structure because it would be silly to teach hundreds of specific sentences, one by one. At that point it’s a memorization task. Memorization is not learning. What you want it to be is a task of building, and then operating on, new infrastructure. Robust, solid infrastructure. That’s learning.
In the book Superforecasting, Dan Gardner and Philip Tetlock talk about the habits of the best forecasters of the 21st century. You might wonder what they all have in common. And you might think the answer is high IQs, PhDs, and an incredibly strong grasp on complex math and models. But you’d only be slightly correct and you’d be missing the point.
The best forecasters are the people who know how to accumulate and decide on the best machinery for thinking about the issues at hand. They aren’t usually multidisciplinary experts, either. They just do a lot of Googling and reading. They do a lot of identifying healthy, useful assumptions. Assumptions that have applied in the past and therefore will most likely apply in the future.
The best thinking doesn’t usually come from high-resolution, detail-oriented problem solving. It usually comes from starting at low resolution and asking the most basic questions you can possibly ask. Questions that history has proven are important.
Questions like:
What is the historical likelihood of this happening? No specifics, no yeah-buts, just… how often has this happened before?
What are the most basic laws of human behavior that would apply in this situation?
What is the most basic emotional problem that child abuse victims usually suffer from?
What do central banks usually do when faced with economic fear?
The point isn’t that these are the final answers. The point is, if you don’t learn how to ask these questions, you’ll spend all of your time answering questions that are situation-specific, non-repeating, and often a drain on your resources.
If you must complicate, complicate after you cover the basics.
Knowing how to ask the right questions is 90% of problem solving. Finding answers only becomes important when you actually understand how to understand what’s in front of you. And that only comes with good machinery. And you have to build that yourself.
Drink water and then forecast the next time you’re going to drink water,
JDR
“Think before you speak. Read before you think.” - Fran Lebowitz