The Roofshot Manifesto

A worthy read from Luiz Andre Barroso, Google VP of Engineering

Google’s “moonshot factory” is inspiring and ambitious, but there’s a less talked-about route to many of Google’s great achievements -- the consistent, short-term, incremental “roofshots” that make our products better year after year.

Don’t get me wrong. I want flying drones that can bring me fresh produce. I’m excited about contact lenses that measure blood sugar. And I look forward to the day that self-driving cars are on the road everywhere. These initiatives are examples of some visionary programs being pursued by Google and Alphabet teams, collectively referred to as moonshots -- disruptive, 10X leaps in technology.

But there has been a growing perception that moonshots are the primary model for radical innovation at Google, and chiefly responsible for our greatest product and technical achievements. What I have seen during my 15 years at Google does not match that perception. I contend that the bulk of our successes have been the result of the methodical, relentless, and persistent pursuit of 1.3-2X opportunities -- what I have come to call "roofshots."

When I was a software engineer in Search, I witnessed how our product continuously got faster and more amazing. It happened through thousands of scoring code changes, new click data, faster networking, clever use of in-memory indices (later through Flash memory), and so on. Each of them a roofshot. The combined result, though, was awe-inspiring.

Our datacenter innovations followed a similar storyline. We started with a pile of machines in third-party facilities and decided we could do better. We kept gaining electrical efficiencies, changing how airflow was provisioned, improving maintainability and automation of repairs processes, and learned how to rethink datacenter design as a computer design problem. Roofshot after roofshot led to some of the most efficient and reliable datacenters in the world.

Seen from afar, these kinds of achievements could be mistaken as moonshots. They were, in fact, sequences of roofshots. A sequence of roofshots is a compelling innovation model that can produce both quick returns and sustained, transformative results. A 1.3X roofshot per quarter is a clever way to achieve the equivalent of a 10X moonshot in less than 3 years, with the added bonus of giving you a 30% improvement in the first 12 weeks!

Although true moonshot leaps are rarer, particularly in mature engineering areas, we must learn how to identify and seize them. So how do we find those rare opportunities? I subscribe to the artist Chuck Close’s position that “inspiration is for amateurs; the rest of us just show up and get to work.” Moonshots tend to reveal themselves to people who chase roofshots with passion.

I’ve had a chance to contribute to some of Google’s successes over the years and had a front row seat to many other wonderful achievements by my colleagues. If there is a winning formula, it is this: go out there and dream big; then show up to work the next morning and relentlessly and incrementally achieve them