Predictability.

Right now I’m reading Michael S. Malone’s The Big Score, and one thing that I love about it is how much it believes that key individuals drive and create industries. It’s an infectious belief, and a necessary one to write a concise, coherent narrative story about the origins of Silicon Valley. It’s something I’ve thought about a lot as well in my career, and also while writing my upcoming book on operating as an engineering executive–how much do good executives really matter?

My ego’s too frail to sustain a proclamation like, “Executives don’t matter!” Further, that doesn’t reflect my lived experience: I think executive quality matters a great deal. That said, I do think that there are non-obvious ways that seemingly mediocre executives outperform the sum of their capabilities, and seemingly brilliant executives underperform their talents. One of those ways is the extent that they create a predictable environment.

Uber gives a clear example and counterexample of predictability:

• At Uber, our CTO strongly preferred letting teams work through disagreements themselves. The exception was existential issues (“our database will run out of space in six months”) or CEO decrees (“we will build a datacenter in China in the next six months”), where one of the CTO’s trusted advisors would select a top-down plan.

  Many folks disagreed with both the mostly bottom-up approach (“it’s just politics”) or the trusted advisors (“why does he listen to those folks?”). Many disagreed with the specific decisions. However, it was predictable how decisions would get made, and that made it easy for teams to plan around. Teams knew how to make forward progress, even if they often disagreed.

• Later, Uber hired an engineering executive beneath the CTO who started rapidly changing a number of technology decisions without much input. He actively avoided seeking input because he was convinced the existing team’s context was irrelevant due to their relative lack of experience. Decision making became unpredictable, both in terms of who was expected to make which kinds of decisions, and which decisions would be reached.

  Many folks believed the specific decisions being made were better than previous choices, but decision making became extremely unpredictable. Teams got stuck, unsure how to make forward progress, even those teams that agreed with the vast majority of decisions.

Although I never thought about predictability directly, much of my onboarding approach as an executive is around increasing predictability while I come to understand the business, team and technology enough to make more context-specific decisions. In the first several months, it’s difficult to decide whether to shut down a business unit, but you can absolutely increase predictability by leading with policy rather than exceptions and explicitly documenting the engineering strategy that the organization already follows.

That said, the moral of the story here is that predictability is valuable, not that it’s a universal cure. A mediocre but predictable executive will likely outperform an extraordinary but unpredictable executive, but both are unlikely to be successful in the long-run.