Fat tails and boring edges

In light of feedback received during my ongoing search for my first product management role, I re-read Taylor Pearson’s article, How to Get Lucky: Focus on the Fat Tails. The following section stuck:

Almost everyone systematically under-allocates resources to the fat tails. We tend to spend most of our time and energy thinking about the middle, because we see the world through a bell curve lens, and most of the area in a bell curve is in the middle. But in reality, we live in an 80/20 world, where the top 1% of fat tails account for 50% of the results. The people who understand this seem to know how to attract luck, but really they’ve just adopted the 80/20 curve model.”

In terms of job/work hunting, that means de-emphasising traditional methods (cover letters and CV submits) in favour of fat-tailed methods (asking people questions, attending meet-ups regularly, joining virtual conferences, working in public). Visually:

I’ve considered the “focus on fat tails” approach in another domain, too: stack entry.


For a long while technology has been an interest. But it’s never been more than that. It’s been like an invitation to dinner that I’ve never taken the host up on. Until now. Recently, I decided I wanted to build a simulation model that would allow me to play with the mechanics of trust de- and re-generation. One problem: I don’t know how to code. The agents in hash.ai’s agent-based models are programmed in either Python or Javascript; I decided to start by learning Python. The above wasn’t as whimsical a decision as it sounds.

The internet is an integrated software-hardware stack. For example, the Internet protocol suite is modelled with four layers: link layer, internet layer, transport layer and application layer. Visually:

I’ve spoke with people about learning to code before and my preference has always been to enter the stack at one of two points: the bleeding edge or the boring edge.


I recently listened to Dawn Song’s appearance on the Lex Fridman podcast. Dawn Song does all sorts of interesting things and one of those things involves neural program synthesisbuilding a program that can solve a problem by writing its own program. That is bleeding edge.

In the same episode Lex Fridman also mentioned that he was looking for an expert in Fortran. Fortran is a third-generation programming language but one that is older and increasingly hard to find “experts” in. Below third-gen languages there are second-gen languages (assembly languages) and first-gen languages (machine-level languages). That is boring edge.

In contrast to the bleeding and boring edges of the stack is what Taylor Pearson called the body of a bell curve: the lumpy middle. Entering the stack at the lumpy middle is, on most accounts, fine. It’s okay. Reasonable. The most sensible. After all, the lumpy middle is lumpy for a reason; the skills and knowledge contained there are both readily available and relatively valuable. But I think entering at the bleeding or boring edge is a better proposition, both in the short-term and the long-term.

Short-term: I suspect learning about the bleeding or boring edge is just more exciting. Long-term, I suspect it turns out to be much more valuable. Until expertise in the boring edge is needed and until the functionality of the bleeding edge is realised, both are undervalued, but when those times inevitably come a premium can be had. In other words, the fat tail of the technological stack is the bleeding edge and the boring edge.

Another reason not to begin in the lumpy middle: generally, I suspect it is easier to move from one of the B-edges to the lumpy middle (or even straight to the other side) than it is to do the reverse. I don’t have data (even anecdotal) to back this up; it’s just a hunch (feel free to inform me otherwise).

As to choosing between the two edges in the context of stack entry: I think it depends on one’s preference for abstraction. Bleeding edge competencies tend to revolve around system-level problems and solutions–the realm of high abstraction. Think Elon Musk’s dallying with rockets and brain chips or attempts to solve super wicked problems like climate change or wealth inequality. Boring edge competencies tend to revolve around component-level problems and solutions–the realm of minimal abstraction. Think circuit board design.

Personally–as I think my objective to toy with simulation models and the joy I find in writing stories demonstrates–I lean towards abstraction.