Our Sister Mei: Silicon Valley's Beauty and Its Model Worker
Marissa — known to all as “Sister Mei” — was once a darling of the media, and she has been out of the public eye for a while now. The last time people remembered her was when Yahoo, that one-time titan and internet forefather that once carried a market cap in the hundreds of billions, handed itself over to Verizon for $5 billion.
Yet five years ago, if you asked who the hottest person in Silicon Valley was, and Sister Mei dared to claim second place, even Zuckerberg wouldn’t quite dare to claim first. The first 13 years of her career (1999–2012) were magical and glorious: starting as a fresh Stanford graduate, she grew into Google’s most important VP, the one in charge of Google Search, the right arm of the three founders — had she never left Google, there probably wouldn’t be much of a role left for Sundar Pichai (“Firewood,” as Chinese-speaking Googlers pun his name). Then came the next five years (2012–2017) as CEO of Yahoo, when her luck seemed to drain away entirely: she toiled and she struggled, and drove a company that had only recently cashed a nearly $50 billion check from Uncle Gates and could still draw breath straight into a $5 billion fire sale.
Humans are base creatures: we cling to the most recent thing and forget everything that came before. Some people spend the first half of their lives doing good, then ruin it all with a single misstep — losing their integrity in old age, as we say. Others scurry and scheme their whole lives, doing evil without pause, yet catch the wave and rise to riches and renown, becoming Buddhas on the spot without even changing their vestments, praised on all sides as if the earlier evil had never happened. Sister Mei finds herself in just this awkward position: until her next big thing arrives, what people will always remember is her great defeat at Yahoo, and the label Fortune pinned on her in 2016 — “one of world’s most disappointing leaders.” Who still remembers that this same Fortune lavished every honor it had on her in 2014: #6 on the year’s “40 under 40,” #16 among the year’s “world’s most powerful businesswomen” — to say nothing of her many earlier glories at Google.
Originally, Sister Mei and I — an ant and a Bodhisattva — were unlikely ever to intersect. But by a twist of fate she became an advisor to Tubi TV (on the startup advisor system, I may write a separate piece someday), and so I had the good fortune, together with CEO Farhad and CTO Marios, to sit and talk with her in our office for an hour.
Sister Mei is poised and beautiful — the years have left a few marks on her face, yet I still find it hard to believe that this elegant woman before me is the iron-willed force who, for more than a decade without a break, worked over 100 hours a week. It defies science; if I worked 100 hours a week, I would surely age so badly that toddlers would call me Grandpa. She is full of energy, warm and outgoing, speaks at a breakneck pace, and her thoughts leap; lose focus for a second and you have fallen behind. Yet her voice is strikingly hoarse — exactly like a teacher who has wrecked her throat after years of lecturing from a podium.
We talked with her about Tubi TV, about Google, about a handful of methodologies. Bringing a laptop to such an occasion didn’t feel right, so I could only keep repeating in my head the maxims that lit up my thinking; the instant the meeting ended I rushed out, flipped open my MBP, and quickly wrote down the snapshots still surviving in my mind. Hence this article.
What follows is material that can be published publicly and that I think is worth a read. Please forgive the memory of a nearly-forty, state-certified middle-aged man — if times, places, names, or exact quotes are off, defer to the other sources.
There Are Many Roads — Step Back and the Sky Opens Up
Sister Mei talked about an early Google “failure”: Google Answers. At the time Google Search had only just gotten off the ground and was already in a competitive stance with Yahoo. Yahoo had a natural content advantage, especially the heavily used Yahoo Answers. Google worried that one day Yahoo might suddenly forbid its crawlers from indexing Yahoo Answers pages, giving Yahoo an unfair competitive advantage (the Taobao–Baidu war and the Taobao–WeChat war are still vivid in memory). So Google Answers, a defensive product, was greenlit. Yet competing on content against a Yahoo Answers that had years of accumulated material and very sticky users was no easy thing; Google Answers got off to a bad start — even with generous bounties, it managed only about 100 questions a day. Then, in one discussion, someone had a sudden inspiration: our goal is to capture and organize as much information as possible, and building our own content is not our strength — so why not give people an incentive to work with us, to keep creating original content for Google to index, thereby boosting their exposure and earning their own share of ad revenue? Thus Google AdSense was born. Google AdSense liberated a great many content producers — at first newspaper and magazine editors, who discovered that their spare-time output could earn them a tidy income, sometimes more than their day job. Word spread, more and more sites adopted Google AdSense, and more and more individuals built their own blogs, working hard to write good content, raise their rankings, and earn more. And Google no longer had to worry about content — countless small and mid-sized sites became its de facto allies, sharing fortune and misfortune, splitting the spoils — so all it had to do was focus on delivering better algorithms and better infrastructure, guaranteeing that it was the best search engine and the best ad system.
Instill a Sense of Urgency
We asked a question: in its early years, how did Google get engineers to treat the company as home and work long hours of their own free will? Sister Mei brought up making employees feel a sense of urgency. She told this story:
She was in her early twenties then, a single young woman with nothing much going on; on weekends with nothing to do she’d come into the office around 11 a.m. and get some work done — hardly a slacker. One Saturday, egged on by friends, she went to IKEA and bought some furniture, then spent the whole weekend assembling it. As it happened, that weekend everyone had something on — some were showing friends around town, some were at weddings; in short, nobody was in the office working. At two in the afternoon her boss came in as usual, found not a soul there, and blew his top; on Monday he gathered everyone and lit into them. He said: I am extremely disappointed in you. We are growing at a furious rate, and there is an endless amount of work waiting for us. Google is going to be a great company, and every one of us has a shot at becoming a millionaire, at becoming part of history. I will not allow anyone to break that dream, and I will not allow anyone to stand between us and becoming millionaires. Assembling furniture, visiting friends, going to weddings — those are all f***ing unimportant things. With me, you’re either all in or you get the hell out. Starting today, we’ll provide unlimited food and drink on weekends, and I want every damn one of you sitting at your desk, building history with your own hands!
Sister Mei said she was deeply affected — before that she had been a conscientious, mildly ambitious young woman; afterward she gradually transformed into a workaholic who put in 100 hours a week. At Google she acquired this sense of urgency, and she carried that philosophy to Yahoo as well (Author’s note: alas, to no avail).
A Mission — The North Star of a Life
At Yahoo, Sister Mei was still working furiously while pregnant, huge belly and all. One evening a senior executive chatted with her for a few minutes and could not understand why she drove herself so hard. She said that at every moment she felt a sense of mission to “make Yahoo great again” (Author’s note: Make Yahoo Great Again), and that she didn’t find it tiring. The executive scoffed, dismissed the mission out of hand, and retorted that her doing this made everyone else extremely stressed. She said she felt a sudden chill of desolation — the whole point is that generals and officers share one heart, so that the whole army fights as one. These pampered VP lords cared only about their own little patch of turf; with the mission gone, the work was just f***ing work. Employees treating a job as a job is no crime, but when executives are like that, you can guess something about the caliber of the followers they attract. (Author’s note: one cowardly soldier shames only himself; a cowardly general shames the whole army.)
At Google, she said, the early crowd was fanatical to this degree — Jeff Dean, yes, that famous Jeff Dean (https://research.google.com/pubs/jeff.html), a man whose every single achievement, any one of them, would give you or me enough material to talk about for a lifetime. Every evening around six or seven he went home and had dinner with his family; before nine he had put the kids to bed, then rushed back to the office and worked until midnight, one in the morning.
Stephen Chow said: if a person has no dreams, what’s the difference between them and a salted fish?
What Do We Dare Not Do — or Even Dream — Once We’ve Grown Up?
Children are all dreamers; they build dreams every day. Give them a pen and they can sketch out a dazzling universe; give them a lump of clay and they can conjure up a world untethered by gravity. We parents often sigh at how far short we fall, terrified that our pig-brains can’t keep up and will even cramp their imagination.
Companies are the same. A young company is like a child full of curiosity, and its thinking should not be imprisoned by the mature companies on the market. She said George (who then ran the engineering team) often told them: you must constantly ask yourselves — what we can’t do / imagine when we go big? That is the sort of thing we ought to just go and do.
She gave an example. In its early days Google’s service wasn’t stable and the algorithms were constantly being tweaked, so when crawling and indexing pages it cached the entire page, for use after a service crash or for future re-indexing, which also saved no small amount of bandwidth. If this had come up at any other established search engine company, the first reaction would have been: there are hundreds of billions of pages on the internet, and you want to store all the original content? Were you kicked in the head by a mule? Buddy, we’re a search engine — do the indexing right, find the right content when the user searches, and that’s enough! The second reaction: storage costs would be boundless — do you know what TCO means? In short, such a thing would not happen, and could not happen, at a big company.
Fortunately, Google — imaginative, childlike and innocent — didn’t worry about all that. It rolled up its sleeves and got to work, cutting roads through mountains and building bridges over rivers. Storage too expensive, couldn’t afford the pricey IOE stack (IBM servers, Oracle databases, EMC storage networks)? Then use ordinary commodity hardware — not much different from the PC in your home or mine. We know the hard drive is the weakest link in a PC, with an MTBF typically in the hundreds of thousands of hours. Call it a million hours (about 115 years), and the annualized failure rate is 0.8 percent — that is, out of a thousand disks, eight will fail on average. So Google had to develop the right algorithms to guarantee data availability under those conditions. Using commodity hardware is now the received wisdom of the internet world, but back then nobody dared to do it.
And so Google cached the entire internet at very little cost, doing what other search engines could not. What good did it do? Sister Mei said that as a search engine, results contain a title and a snippet of summary. Other engines, having thrown away the original pages, drew that snippet either from the page’s meta information or from part of the first paragraph of content — which is not necessarily the best choice, and might not even contain the search keywords. Google cached the original pages, so it could calmly pick the most fitting sentences from the page based on the query and even compose suitable content of its own to supply that summary. This instantly raised the probability that a search result would be clicked, because what people saw was a more relevant snippet. And with the original content in hand, Google could also run lots of A/B tests, post-hoc analyses, algorithm adjustments, and so on, rather than relying on whatever came out at the instant of the crawl. By the time Yahoo and the rest woke up and realized they should be doing the same, it was too late — and their entire infrastructure would have to be torn down and rebuilt to accommodate the change.
Other early Google products, like Gmail — why not an “unlimited” inbox? — were the product of the same thinking. Sister Mei said: we firmly believed in Moore’s Law, so we boldly ran the experiment. Gmail spread by word of mouth, everyone waited to be one of the lucky invitees, and the average user wouldn’t use much storage in the first year or two anyway; by the time the accumulated data grew, the price per GB of storage had already dropped by an order of magnitude. So you see, once your mindset shifts and you think what others dare not think, your thinking opens up enormously, your approach to doing things changes abruptly, and the whole cost structure comes out completely different. When Yahoo Mail — free under 4 MB, charged above it — saw users flooding into Gmail and hurried to catch up, it found that its IOE-based stack simply could not compete on cost structure. Awkward indeed: grit your teeth and bleed to keep up, or cut off your own arm and rebuild the system?
This reminds me of The Innovator’s Dilemma. Mature companies sit atop a mature pricing system and gross margins. When an innovator attacks a leader accustomed to 50 percent margins with 5 percent margins, the leader’s entire value chain takes the hit: what to do about my own cost structure (should I move off IOE?), how do I explain this to Wall Street (should I lower my margins?), do I keep up the relationships with upstream and downstream partners, how do I placate my existing paying users and customers — and so on. You just can’t afford the pain.
So the adult world is full of worries, full of people to answer to, full of compromises between standing and kneeling. Many things we dare not do — and do not even dare to imagine. George asked everyone to keep a child’s original heart, to dream wild dreams, to break the boxes, to go do things that look impossible. Gmail, Google Maps, Google Photos — all trace back to the same lineage.
An Anecdote
The London Eye. It was around 1999, and Sergey took the team to London for fun. Everyone queued beneath the London Eye, waiting for a look at the London panorama, while Sergey quietly stood counting the London Eye’s foot traffic. After a while he said: I did the math — a stupid business like this makes 80 million a year, and we only make 80 million a year too. We really ought to be doing far, far better than this.
In 2016, Google’s revenue for a single day, by the figures Sergey calculated back then, was worth three years of the London Eye.
Wisdom excels.
