Skate to where the puck is going

01 Jan 2021

Walter Gretzky told his son to skate to where the puck is going, not where it has been. It’s advice about the future, which is the one thing data cannot tell you about. Every number you have describes the past. Optimise against it and you get a better version of what already exists — a data-driven approach produces increments, never the thing that makes the increment irrelevant. Strategy is a bet. So be thesis-driven: have a thesis for how the world will change and how you win if it does, then measure the results of your actions along the way to find out whether the thesis holds. Data has a role; it comes after the imagination, not before.

The thesis comes from constraints. Identify the constraints your industry lives under, then ask which of them will change. Bezos framed the mirror image: what’s not going to change in the next ten years — customers will always want lower prices, faster delivery, more selection. What stays unchanged is people. What changes is the cost of everything around them. Most industries have one binding constraint at a time, and when it lifts the whole cost structure reshuffles. The incumbents, organised around the old constraint, are the slowest to move.

The clearest history of this is computing. For decades memory was the binding constraint, and programmers spent their cleverness squeezing programs into it. Moore’s law removed that constraint, and the companies that figured it out first — Apple, Microsoft — built for the hardware that was coming rather than the hardware they had. Make the product great; maybe it doesn’t run today, it will run tomorrow. Games did the same, running a tiny bit ahead of the curve so they’d be viable soon. Games are non-fungible and hardware is fungible, so the game that arrives just as the hardware catches up owns the moment. Too far ahead and you’re wrong for long enough to die; the skill is being ahead by exactly the time it takes to build. Paul Graham made the same call for software in 2001 with The Other Road Ahead: the browser would win, releases would become continuous, nobody would think about the operating system. He was right, and early.

So look for the big wave in whatever is getting cheaper. Right now that’s energy. The question to ask is: what non-commodity, customer-facing thing just uses too much power to be viable today? What would work far better with energy at a fraction of the cost? Heating and cooling, recycling — anything that’s energy-bound rather than idea-bound. Joel Spolsky’s Strategy Letter V says smart companies commoditise their complements. The other side of that coin is to work on things that are complementary to what’s decreasing in price: if renewable energy heads towards zero cost, demand for its complements goes up. Figure out what those are before it’s obvious.

The hard part isn’t the analysis. It’s holding a view that isn’t the consensus. People overestimate how many others share their views, so a non-consensus thesis feels wrong even when it’s right, and consensus feels like evidence when it’s just company. Whatever you focus on grows: if you focus on risks, you find more risks; if you focus on opportunities, you find more opportunities. Both are real. Choosing which to look for is the strategy.

← Theses