把乐高交出去：Molly Graham 的超高速增长生存手册

Graham starts by quoting a friend who said all advice is just someone telling you what they did. Her own version is that she made every mistake in the book and then started inventing new ones. She began in tech in 2007 at Google, in a communications department that grew from 25 to 125 people in nine months, which was her first experience of growing very fast.

She then spent five years at Facebook, joining in 2008 when it had 80 million users, around 500 employees and was smaller than MySpace. When she left it had 5,500 employees, 5 billion dollars in revenue and over a billion users. After that she joined Bret Taylor's startup Quip to learn how to build something from nothing, then helped Mark Zuckerberg and Priscilla Chan start the Chan Zuckerberg Initiative, which grew from 30 to 250 people in a year.

Across those jobs she says she has studied what it takes to thrive, not just survive, inside growing companies, and what separates businesses that plateau from generational ones. Her pattern is to take only jobs she is highly unqualified for: "I like being scared." Lenny relays a similar line from another guest, that if you feel comfortable at work something has gone wrong.

Graham's best-known metaphor imagines a giant pile of Legos put in front of kindergarteners with the instruction to build something. You build a house, get good at it and decide you were put on earth to build houses. Then someone says it is actually a neighborhood, and you must hand the half-built house to a new hire while you build dog parks and streets.

The emotions that come with that are territorial feelings, fear of the new person messing things up, and worry that the new work will be less fun. Graham says she still feels them after 18 to 20 years, but you should give the Legos away anyway. The torment of scaling companies is that the ground moves under your feet, and the opportunity is going from someone who builds houses to someone who can build entire worlds.

The reason to do it is that the Lego pile grows as fast as the company, so anyone who holds on ends up buried. Graham says that at Facebook she gave away her job about every three weeks, constantly rehiring herself. You have to grow as fast as the company grows to take advantage of what it offers.

To deal with the irrational side, Graham externalizes the emotions of change into a little monster she calls Bob, whose job is to make her the worst version of herself. Bob is the one who wants to grab the fun Legos back or send the rage email at 9 p.m. Bob never goes away, and it can get worse as you get more senior and feel you ought to know what you are doing.

Her rule is to let Bob do his thing without acting on the emotions, which are normal but are not a compass. Sleeping on it rarely helps because you still want to send the email at 8 a.m., so her rule of thumb from Facebook is to give it two weeks. Reactions that come in waves and fade are Bob; anything that persists beyond two weeks is real and worth discussing with a manager, friend or coach.

Asked when you should not give away your Legos, Graham says there is no rule of thumb, but embracing change is generally better than fighting it. Leaders should help people look at the future rather than the past, and she has seen even being layered under a new manager turn out well. She adds that careers are long and a hard moment is one chapter or even part of one.

Graham, who gave a TED talk on this, tells the origin story. After two years in HR at Facebook she planned to stay until the IPO, but Chamath Palihapitiya, who ran growth and mobile, told her over lunch she was wasted there and offered several roles, ending with building a mobile phone. She knew nothing about mobile, advisers told her not to, and she could not stop thinking about it.

She took the job and spent six months feeling like an idiot, ending with the lowest performance rating of her life. Then, after trips to Taiwan on the hardware, she drew a phone layout on a whiteboard to explain why something was not possible and realized she actually knew things. Over the next three years she became a mobile expert; the phone was a costly failure for Facebook but not for her.

Chamath had drawn her a picture: most people do careers as stairs, getting promoted every two years, but the more fun careers are like jumping off cliffs. You fall, which she says takes about six to nine months, and then you climb out on a J-curve far beyond where stairs could take you. Fast-growing companies are the place to do this, she says, because if you prove yourself people hand you opportunities you are unqualified for. She ended Facebook in product after doing business development, hardware and more, simply by saying yes.

Lenny asks how to tell a promising cliff from a bad one. Graham distinguishes kinds of fear. Financial fear is real: before leaving a job or taking time off, work out the number you need so you are not constantly terrified, which varies a lot by person. Specific financial anxiety, she says, is much more useful than existential financial anxiety, and the number is often smaller than your brain makes it.

The other kind of fear is being scared you cannot do it or will fail, which she treats as a flashing green light. Even failure teaches you who you are: her stint in product at Facebook taught her she has a good product mindset but is not the person who cares about the button. Lenny stresses framing the leap as proving it to yourself rather than showing others, and Graham says nothing accelerates self-knowledge faster than trying something you are scared of.

During the fall phase, Graham says, the most valuable thing is to embrace being a professional idiot, the person who asks what a word means or why the meeting is happening. No one can take your learning away even if the project fails, and often the dumb question is one everyone had but nobody dared ask. She says her superpower is walking into a room and asking whether there are goals, which is often what she was hired to do: bring clarity.

Lenny shares a young product manager who realized he should aim to be a 10x learner rather than a 10x PM. Graham connects this to AI and her friend's high-school senior: the anchor is soft skills like grit, hard work and learning how to learn. What you know today is far less valuable than what you can learn by tomorrow, she says, noting that Google was said to rewrite its codebase every eight years, and evolution is the point of the Legos idea.

Graham explains that the waterline model is not hers; she learned it leading 75-day wilderness trips for the National Outdoor Leadership School. A team is a boat on an ocean, trying to reach its goals, and the waterline asks what under the water makes that harder or easier. There are four layers in descending order: structural (goals, vision, roles, expectations), dynamics (culture, decision-making, conflict), interpersonal (relationships between two people) and intrapersonal (issues within one person).

People usually jump to the bottom and blame individuals, but her rule is to snorkel before you scuba. About 80% of team problems, she says, stem from structure or dynamics, and if a manager does nothing else they should provide clear roles and clear expectations. She has often taken over teams where nobody knew what their job was or what success looked like.

In practice she takes a two-sided approach: here is what I am seeing, what is going on for you, and what do you think your job is. Often the person's picture differs from yours: you described an elephant and they produced a tiger. Re-describing the elephant over and over feels like being a broken record, but it is one of the hardest and most important parts of leading.

Graham says she has a bone to pick with OKRs, because leaders often hand her a hundred-line spreadsheet that creates clarity for no one. Goals, she argues, are a communication tool, and the right cadence depends on stage: every two months at seed, annual once the business is established. Her first rule is that no company needs more than three goals. Facebook had three for the five years she was there: growth measured by monthly active users, engagement and revenue.

The second rule is that one goal must win in a fight; at Facebook that was engagement, since buying bots to inflate user counts did not help it. The third is that an intern starting Monday should understand the goals, with acronyms explained. The fourth, borrowed from Claire Hughes Johnson, is that strategy should hurt: if you are not making painful trade-offs, people will choose priorities for you.

The fifth is one goal, one owner, a number with a name next to it, since two people owning a goal means no one owns it. Lenny adds that having someone's neck on the line is a powerful lever. The sixth is that goals alone are not enough: citing James Clear that winners and losers have the same goals, she says you need a process to follow up, hold people accountable and learn what it takes to move the number.

When working with founders, Graham uses their goals to learn the business and asks dumb questions until they can state the one sentence and one number they mean. It is easier to write the hundred-line spreadsheet than to say what the three drivers of the business are. Lenny notes that Facebook's businesses were just as complex as anyone's, so three goals can work elsewhere.

Graham says her focus is not management 101 but leading through change. The first lesson is that a leader's job is not to have all the answers but to get good at finding them and bringing people together. Saying "I don't know, let's go figure it out" is scary, but the most experienced leaders say it all the time, and Lenny notes Bob will insist everyone can see through you.

The second is not to promise what you cannot control. It is tempting to tell recruits onboarding will be smooth or tell an executive they will be CMO forever, but promises about stability, titles or never hiring over someone demoralize high performers when broken. Lenny recalls a late project where he promised it would never happen again; Graham quotes Claire Hughes Johnson that such promises are letter bombs you mail yourself.

Third, firing is as important as hiring: she says the best hirers are right only about half the time, so half of hires may need to go, and keeping them leaves barnacles dragging the ship. Fourth, her mantra is to serve the business, not the people, since everyone is better off if the company succeeds. A tool from her manager is to ask what you would do if there were no emotions involved, do that, and then work out how to communicate it as kindly as possible.

Graham says managers get dragged into spending huge energy on strugglers, yet high performers are the future of the company and where a 10x return on time lives. Many managers leave a high performer alone because they are doing well. She instead builds a whole system for working with them designed to draw out their potential.

Her method is to run experiments. She forms a theory about what someone is capable of and tests it in increments: less guidance, a bigger project, more visibility, less oversight. Then she pairs the person with company needs, asking where this person's strengths are most useful. She has benefited from the same treatment at Facebook, where people saw potential in her and asked her to help with something that unlocked a lot.

The first lesson from working with Zuckerberg, Sandberg, Larry Page, Sergey Brin and Bret Taylor was humbling. Having worked on culture at Facebook, Graham concluded that 80% of a company's culture is defined by the founder's personality: Facebook is Mark, and Google when she was there felt like a university. Her version of founder mode is to build a company that makes decisions the way the founder would when absent, not to try to shape culture yourself.

Culture, she says, is the way we do things around here, and it lives in how you hire, fire and decide, not in values on a wall. Imposing values that do not match the founder, such as move fast for a founder who is happy not to decide, produces cultural dissonance. A second lesson from Zuckerberg is that people do not escalate enough: escalation is not tattling, it is two people who disagree going together to someone with the power to decide.

The third, from Sheryl Sandberg, is that growing more than 100% a year is a bad idea and 50% is the happiest rate. Growing that fast means you cannot de-duplicate issues, so two people are hired into the same job without talking to each other. More people does not make you faster, Graham says, so be scared of adding headcount and ask many questions as you open roles.

Lenny relays a former colleague's observation that Graham's ambitions shifted away from big C-level roles toward community building. She describes a model in progress: everyone has a proving phase in which they show they are good at things and do what career advice says they should, until a wall forces the question of what you want to do with it. After 10 to 15 years of helping brilliant founders realize their vision, she found it no longer fed her, even though it was hard to walk away from jobs with fancy titles.

She calls the attraction to such jobs a "LinkedIn crush" and says she repeated to herself, "What does this get you that you don't already have?" After a long journey of experiments she landed on building safe spaces where leaders can learn and grow, and find sanity and connection in a crazy world. The question, she says, is what do I want versus what do I think people expect of me.

Closing with the advice she gave her partner Sarah on starting at OpenAI, she says to get to know the customer, because the customer never changes, and to find such constants as guiding lights in a storm. Humans crave stability, but in fast-growing companies you should expect instability: assume you will have a new boss in six months. What you take from any company is people who want to work with you again and what you learned. She can be found at Glue Club (glueclub.com), on LinkedIn and on her Substack.