Before I started working at Google, I reached out to one of my friends, Gabe, for advice. I always looked up to Gabe as a mentor, though he probably never realized that. He was a couple of years older than me, and by that time he’d already been at Google for an entire year. I was about to start my first real job after college and I wanted to learn from Gabe how to be successful at Google.
Gabe told me that working at Google “felt like being a small cog in a very large machine."
Today, July 6th, marks my fifth year at Google, and I have been thinking about Gabe’s words a lot lately, especially because I think he was right. There's an inside joke that you get promoted once for launching something, then promoted again for deprecating the same thing.
I’ve often wished, especially earlier in my career, that my work was more impactful. I wanted to build something that would have a tangible impact on Google, something long-lasting, something people would remember.
I first learned about the world’s largest ball of paint from John Green’s essay of the same name. In 1977, Mike Carmichael and his three-year-old son decided to paint a baseball. And they just kept painting it. For over forty years, people have been adding layers of paint to that ball. Today, it has over 29,000 layers of paint and weighs 11,000 pounds.
In his essay, John Green draws an analogy between life and the ball of paint:
You carefully choose your colors, and then you add your layer as best you can. In time, it gets painted over. The ball gets painted again and again until there is no visible remnant of your paint. And eventually, maybe nobody knows about it except for you.
But that doesn’t mean your layer of paint is irrelevant or a failure. You have permanently, if slightly, changed the larger sphere. You’ve made it more beautiful, and more interesting. The world’s largest ball of paint looks nothing like the baseball it used to be, and you’re part of the reason.
To me, Google is a lot like the world’s largest ball of paint. You own your small part of the codebase, and you add your layer as best you can. A bug fix here, a line of code there, a feature that few people will use. You know it will be painted over, and yet you paint it anyway.
The code I wrote in my first year? A lot of it is now dead, its purpose fulfilled and then forgotten.
But some of my work has evolved beyond what I originally imagined, extended and maintained by engineers I’ve never even met. My first team, the YouTube Ads Machine Learning team, still uses the load testing tools I built many years ago. They’ve used these tools to test thousands of models and they’re an essential part of how YouTube Ads makes cost tradeoffs today. Then there are the inference optimizations I implemented, that still execute billions of times per day, invisible and unremarkable, just part of the machine.
The world's largest ball of paint will never be finished. People will keep adding layers, and those layers will keep disappearing beneath new ones. But the ball keeps growing, keeps changing, keeps becoming something new.
I think I'm okay with that.