Second Principles

Knowing how to do something from first principles is a valuable skill. However, always starting from first principles is a good way to relearn many mistakes.

I am the kind of person who likes to really understand things. I love a good spreadsheet or back of the envelope model. I want to know how all the pieces fit together and convince myself that it all makes sense. Sometimes this comes back to bite me when I've tricked myself into thinking my brand new model is a complete view of the world.

In these cases, the model often ends up being incomplete. Some of my assumptions were wrong, and factors I didn't consider came into play. All the rules of thumb and lessons learned by other people have to be rediscovered. This can be a long and frustrating process. Why would I keep doing it then? Well, sometimes the model does, in fact, reveal something new and interesting. But I digress.

I am less worried if I'm working on things for my own benefit. If it's a project relating to work, there's not often time to reinvent things. Beyond that, you're often far better placed, assuming you have the cash, hiring someone who has experience in an area than to start from zero.

Overall, if you're running a business, keep the number of things you do from first principles to a minimum. Building from first principles may uncover some competitive advantage, but it comes at a cost, and there's no guarantee that you'll find anything.

These days, what I now do instead is start from 'second principles'. I assume that whatever is best practice got that way for a reason. I might still try to find out why it became the best practice, but my starting point is whatever is normal.

From there, I still feel free to model, shift, and try new things, but I start out from whatever local maximum the world has reached thus far. This has the side effect that benchmarks are usually readily available.

Take a sales or marketing process, for example. If you're doing things 'by the book,' there are plenty of industry benchmarks around. If you make up your own process from scratch, you're on your own for the most part.

Not many problems have a single, straightforward guide to best practice, so starting from 'whatever everyone else is doing' often still requires some experimentation. Looking across the spectrum of what people consider best can help filter out the common, critical components.