global optimals vs local optimals

TL;DR

• Local optimal and global optimal are two different things

• Local optimal is within your control, global optimal is outside of your control

• Understand the global, concern yourself with the local

I was walking to the tube this week when I thought about a project I’ve been working on recently that has been experiencing a number of limiting factors. I’ve been frustrated at times, seeing so clearly how things could have gone and comparing them to how they have ended up going.

This thinking led me to articulate for the first time the idea of a local optimal and a global optimal, and I’d like to share these with you.

I’d been obsessing over the global optimal. The global optimal is the best possible outcome, it’s what would happen if everything had gone perfectly. Global optimals don’t concern themselves with practicalities. They’ve never heard of Murphy’s Law.

We should all understand global optimals. They’re a crucial element in benchmarking for success. In business I see this most frequently when early stage startups make revenue forecasts using the Total Addressable Market (TAM) for their solution. The TAM is their global optimal, from which they make realistic estimates about what they forecast they could achieve. The global optimal has authorship, insofar as it is a constant in the equation around which everything else is baked. Global optimals are important to hold in mind, but it’s dangerous to get too friendly with them.

Local optimals are what we can realistically expect given the constraints we’re aware of (and the overhead for those we’re not!). Local optimals are the figure startups get to after they multiple the TAM for their solution by whatever multiplier they’ve reasoned their way to from research and analysis. They’re what happens if we can do all it is that we can control.

We should be spending a lot more time on local optimals than global optimals. As I’ve said, understanding global optimals is good because they have authorship in defining the local optimal. But beyond that, their value is limited. For me, this idea conjures stoic notions that you may have picked up on in the last paragraph.

Global optimals are what would happen if everything we can’t control goes our way alongside everything we can control. They’re baked with uncertainty. Local optimals, on the other hand, are what would happen if we take care of everything we can control and don’t factor in too much of what we can’t.

What I had done was to allow myself to stray into thinking too much about the global optimal. Dangerous. We must understand the global optimal, but concern ourselves with the local optimal. 

We do all we can, and we let the cards fall as they may.