If you take a pregnancy test, most people would think there are only two outcomes: pregnant or not pregnant. When I was around 18, I took a statistics class that left a deep impression on me. Any tests that gave you a binary outcome (either yes or no) actually have four outcomes. Using pregnancy test as an example:
the test says you’re pregnant and indeed you are (true positive)
the test says you’re not and indeed you are not (true negative)
the test says you’re pregnant but actually you’re not (false positive)
the test says you’re not pregnant but actually you are (false negative)
If you take a pregnancy test, most people think there are only two outcomes. Actually, the pregnancy test has four outcomes.
I forgot the rest of those statistics stuff, but I never forgot the insight that a binary test has four outcomes of which two are wrong. My focus in this post is not about statistics. It’s about a pattern I keep noticing. A pattern I call “Two Mistakes”.
It’s a general theory that you can use “Two Mistakes” to perform a kind of pre-mortem analysis about how something may go wrong. Almost always, I can find two diametrically opposite ways things can go wrong in any situation.
Abstraction Levels - What Can Go Wrong?
Here’s an example. If you have read a ton of business books like me, you would notice a pattern of these books. You would easily find two different books arguing with seemingly diametrically opposite approaches on the same topic and they both backed up their separate stances with anecdotal evidence, and sometimes data. Just as easily you can find situations (but the authors will almost never include these) where these approaches were used, yet they failed. Why?
You don’t have to squint very hard and see that most business books, self-help books are essentially books that contain a variety of abstractions. These books essentially boil down to:
giving you a list of abstractions then
show you evidence that these abstractions can work.
If you’re taught to only think in terms of Aristotelian logic (i.e. the law of Excluded Middle), this situation of contradictory abstractions work (and not working sometimes) can induce a severe case of cognitive dissonance. One resolution (but not my only resolution) is to apply the kind of “Two Mistakes” approach to reconcile seemingly opposite approaches that work.
Explaining Away Different Approaches with Two Mistakes
Let’s go back to choosing the right abstraction levels and applying “Two Mistakes” thinking to it. How can we pick the wrong abstraction levels?
I first came up with two dimensions: precision and abstract-vs-concrete where abstractions can be wrong. And in each dimension, I apply “Two Mistakes” thinking. I think about how things can go wrong in either direction.
Thinking about how things can go wrong in either direction is analogous to the false positive and false negative in the first example about pregnancy tests.
Technically, you can add more dimensions to analyze abstraction levels with. And each dimension you can think of, you can keep applying the “Two Mistakes” pattern. If the dimensions are not mutually exclusive, how you can generate wrong abstractions can get complex at an exponential pace. But I want to keep things simple. So, I stopped at just the two: precision and abstract-vs-concrete dimensions.
On hindsight, that was the first time in this newsletter I apply “Two Mistakes” pattern to my thinking. I will continue to liberally use “Two Mistakes” going forward as an experiment to see how useful this is. So be warned not all pieces I wrote using this pattern will be useful or even correct.