The Geek Way: The Radical Mindset that Drives Extraordinary Results

by Andrew McAfee

“Well, when geek means that you’re willing to study things,… then I plead guilty, gladly.”

Here’s my answer: the geek way is the right way for modern companies, because modern companies have to move faster and innovate more than their predecessors did.

Former Google CEO Eric Schmidt explained to me one of the biggest consequences of this shift: “In the classic corporate model, everything is run in a hierarchical way, the offices get bigger over time, and bureaucracies abound. Companies like this were actually successful for a long time because they have some strengths: they’re predictable and they serve their customers well, as long as customers keep needing the same thing. The reason that culture doesn’t work very well in the information age is that the customers need changes, and you have to be able to change more quickly than, you know, every five years.” The geek way is a set of solutions for thriving in this faster-moving world. They’re cultural solutions, not technological ones. As much as the business geeks love technology they don’t think that any particular suite of technologies holds the key to building a great modern company. There’s no killer app for business success. What there is instead is what there have always been: people, groups, and the challenge of getting them to work well together in pursuit of goals. After thinking long and hard about this challenge, the business geeks have figured out some powerful and unconventional ways to tackle it. They’ve taken the standard corporate culture of the industrial era and given it an upgrade.

As we’ll see, the geek way leans into arguments and loathes bureaucracy. It favors iteration over planning, shuns coordination, and tolerates some chaos.

I can’t recall ever hearing any manager, geek or not, talk about many of the key concepts we’ll cover in the pages ahead: ultrasociality, prestige versus dominance, ultimate versus proximate questions, the press secretary module, and so on. I’ve also not come across any of these concepts in a business book written for a general audience (as opposed to an academic one). And while some concepts we’ll explore in these pages are more familiar, like plausible deniability, observability, norms, common knowledge, Nash equilibria, and the prisoner’s dilemma, they still aren’t given enough emphasis given how important they turn out to be. In short, the new science of cultural evolution hasn’t yet been transferred from university seminar rooms to corporate boardrooms and meeting rooms. It’s not what they teach you at Harvard Business School5 (trust me, I was there for fifteen years). The really exciting opportunity for me as the author of this book—and (I hope) for you as its reader—is to create a new understanding of some key business issues by bringing together theory and practice. Or, to be more specific, some new theory from the field of cultural evolution and some new practices from geek companies.

The geek way leans into arguments and loathes bureaucracy. It favors iteration over planning, shuns coordination, and tolerates some chaos. Its practitioners are vocal and egalitarian, and they’re not afraid to fail, challenge the boss, or be proven wrong. Instead of respecting hierarchy and credentials, they respect helpfulness and chops.

This advantage comes from three main sources. The first two are Planet’s willingness to use cheap, commercially available components, and its tolerance for failure.

“At NASA I was part of five spacecraft missions in six years. And that was considered prolific; most people do one or two in their NASA career. Subsequently, in the ten years of Planet, we’ve launched five hundred spacecraft on thirty-five rockets, with eighteen design/build iterations of the spacecraft concept. It’s a different pace of innovation.”

said, “I can’t find out who can approve this.” He said, “What are you talking about?” And I said, “I’m getting ready to change the Amazon.jobs website. There’s gotta be some kind of management review committee. Somebody has to look at this.” He said, “Somebody already did—you. You’re telling me that we’ve got legal approval. We also have business approval; that’s you. So why don’t you push the button and make the change?”

“What happens if this change you’re proposing is a mess?” And I said, “Well, we turn it off.” He said, “How long will that take?” And I said, “Less than twenty-four hours.”

When I asked Williams how long it would have taken one of her previous employers to make the website change she proposed, she answered, “About three months.” But at Amazon, when projects were done people just pushed buttons.

He finished with “So what do you think?” and sat down. In my experience, this was a cue for employees to tell their CEO that what they’d just heard from him was great,

Instead, the first person to speak, who looked too young to have been anything except a recent hire, opened with “There are a couple things that I don’t like” and continued on. I felt a little bit sorry for the kid. Flatly contradicting your CEO in public might not be a career-ender, but it would at least serve as a teachable moment for him and the rest of the people in the room. Of course, Halligan wouldn’t respond with anything as obvious as “Watch yourself, youngster,” but would he take the usual approach of making a self-deprecating comment that everyone in the room would implicitly interpret as a warning? Would the other attendees rush in with “What I think you mean to say…,” or “I hear a lot of agreement here!,” or any of the other standard lines intended to restore all-important harmony? What actually happened was that Halligan said, “Yeah—good point.

His body language didn’t change, the tension in the room didn’t rise, and no one except me looked the least bit surprised.

The first great geek norm, which is epitomized by Will Marshall’s journey from NASA to Planet, is speed: a preference for achieving results by iterating rapidly instead of planning extensively. Ardine Williams’s experience early in her time at Amazon illustrates the second norm, which is ownership. Compared to industrial-era organizations, geek companies have higher levels of personal autonomy, empowerment, and responsibility; fewer cross-functional processes; and less coordination. Doug Bowman left Google because that company decided to make design decisions based not on judgment or expertise, but instead on the norm of science: conducting experiments, generating data, and debating how to interpret evidence. In the interaction I witnessed with his employees, Brian Halligan demonstrated the fourth and final great geek norm, which is openness.

The geek norm of science, based on producing and evaluating evidence, seemed alien to Katzenberg. Whitman observed in an interview that his ideas didn’t need to be supported by a lot of evidence: “I say, ‘Where’s your data?’ He says, ‘There is none. You just have to go with your gut.’” In another interview, an unnamed “high-level Quibi colleague who is actually several years younger than Katzenberg” commented further on his evidence-free confidence. As this person put it, “I don’t pretend to know what kids are going to like anymore. But Jeffrey is, like many entrepreneurs, a true believer in [his] own underlying ideas.”

“Unless you agree with [Katzenberg and Whitman], you’re a troublemaker. Meg believes she’s a marketing genius; Jeffrey believes he’s a content genius. So you end up in shitty jobs where you’re there to execute their vision, which no one else there believes in.”

Hastings had long believed that openness was essential for making good decisions. The culture deck, for example, addresses an ideal colleague and says “You say what you think even if it’s controversial” and “You are known for candor and directness.” Such colleagues, however, were in short supply just a couple years after the culture deck was published.

In the wake of all this Hastings wondered why he didn’t get more internal opposition to Qwikster before it launched. It was a dumb idea after all (dumb, dumb, dumb), and he had worked to build a culture in which it was okay, even expected, for colleagues to give negative feedback when warranted, even to those higher up on the org chart—when, for example, an executive came up with a dumb idea. So he asked around to find out what had happened. Why had the culture of openness failed, despite what the culture deck said? He learned that people within the company had hesitated to tell him what they thought about Qwikster because he was so passionate about the idea and so sure of himself. As one of his colleagues told him, “I knew it was going to be a disaster, but I thought, ‘Reed is always right,’ so I kept quiet.”

The differences between these two types are largely cultural.

They are (in alphabetical order) agility, collaboration, customer, diversity, execution, innovation, integrity, performance, and respect.

The second aspect of the two experiments I want to highlight is close to the opposite of observability; it’s plausible deniability. Can a person make a claim like “that wasn’t me” or “that’s not what happened” or “I didn’t know / see / hear / do that” and have any hope of being believed?

The challenge for entrepreneurs, executives, and managers is to build healthy, fast-learning companies.

There are plenty of other ways to put the Homo ultrasocialis perspective to use. For example, this perspective causes leaders to ask different questions about the critical topic of ethics. In addition to asking, “What kind of ethics training should we have for our people?,” they’ll also ask, “What are ethical behaviors we want to encourage? How can we turn these behaviors into norms? How can we increase observability that these norms are being followed, and reduce plausible deniability that they’re not?” As this example shows, taking the Homo ultrasocialis perspective focuses us on group-level interventions, while the Homo sapiens perspective often focuses on individuals.

At that point, Kahneman had the smart idea to use a technique—one that would be included in the textbook—for surveying the members of a group. During a team meeting he asked everyone to write down how long they thought it would take to finish the book. (He hadn’t discussed the topic with anyone else before the meeting.) These independent estimates ranged from one and a half to two and a half years, with an average right in the middle. Kahneman then had another good idea. He tried to find out what the base rate was: how long did previous projects take to create a first textbook for a brand-new curriculum?

The statistics that Seymour provided were treated as base rates normally are—noted and promptly set aside… After a few minutes of desultory debate, we gathered ourselves together and carried on as if nothing had happened.

They proposed that the press secretary—and overconfidence, in large part—exist because confidence is so important for us Homo ultrasocialis. Confident people have an easier time attracting allies, psyching out opponents, and attracting mates.

“I think part of it is that he’s confidently inconsistent. If he were inconsistently confident, on the other hand, he wouldn’t be nearly as popular.”

To see how, let’s go back to the ultimate geek ground rule: Shape the ultrasociality of group members so that the group’s cultural evolution is as rapid as possible in the desired direction. Since we know what kind of cultural evolution we’re seeking here, we can be more specific: Shape the ultrasociality of group members so that the group makes better decisions and predictions. We also know that overconfidence and confirmation bias are two of the main things that keep us from getting better at these tasks. Finally, we know that while the Homo sapiens approach of training ourselves to be better, more rational thinkers is helpful, it’s unlikely to make our press secretaries stand down, and might even give them new material to work with as they draft their biased memos.

We saw in chapter 3 that evolution crowdsourced the knowledge about making fire from the individual to the group. Something very similar has happened with ideas: evolution has crowdsourced evaluating them to the group, not to the individual who originated them. Our minds are deeply justificatory with their own ideas, and just as deeply argumentative with others’. This is the viewpoint laid out by cognitive scientists Hugo Mercier and Dan Sperber, who offer an ultimate perspective on how reasoning works in us Homo ultrasocialis. In this perspective, individuals’ ideas get evaluated by group-level discussion and debate. Bad ideas get rejected, and sound ones accepted. A lot of us walk around with the stereotype that (other) people will believe almost anything, but the evidence says otherwise. Most of the time we’re actually quite good at the back-and-forth of probing somebody else’s argument and evaluating its quality.9 In fact, Mercier and a team of colleagues found that we become better at ignoring our press secretaries and evaluating our own ideas if they’re repackaged and fed back to us as coming from someone else.

Data, Demos, Debates

Things, “More shocking, Marc and I eventually became friends; we’ve been friends and business partners ever since… With Marc and me, even after eighteen years, he upsets me almost every day by finding something wrong in my thinking, and I do the same for him. It works.”

So one [thing that demonstrated Horowitz’s value as a colleague] is he would talk back to me… He wouldn’t just roll over. He would argue right back. And if you just observe a lot of companies over time or investment firms or whatever, there’s the temptation of everything wants to become a hierarchy. And people have trepidations about speaking truth to power.

All the alpha geeks I talked to when researching this book stressed the importance of argumentation at the highest levels of the company.

Can There Be Too Much Science?

There are a couple reasons for this. One of them is that the geeks believe Goodhart’s aphorism is not so much a law as a caution against overly simplistic measurement and reward schemes.

Because of the press secretary, we’re bad at evaluating our own ideas. But we’re excellent at evaluating the ideas of others. Our minds are inherently justificatory about their own ideas, and argumentative about the ideas of others. The geek norm of science is all about arguing. It also specifies how to win arguments: with evidence. Not seniority or charisma or past performance or rhetoric or philosophizing or appeals to morality or aesthetics, but evidence. For the norm of science, the ultimate geek ground rule is: Conduct evidence-based arguments so that the group makes better decisions and predictions, and estimates.

Economist Larry Summers warns about the dangers of the “promiscuous distribution of the power to hold things up,” which is a concise description of excessive bureaucracy.

No one at HP would get out a blank sheet of paper and design a spending approval process that distributed the power to hold up an important project among twenty different people spread across multiple countries. No organization wants that. But the organization is the wrong unit of analysis here. The right one is the individual within the organization. And instead of thinking about those individuals as wise Homo sapiens who realize that high productivity and healthy profits will be beneficial to them, we should instead think of them as Homo ultrasocialis who know way deep down that high status will be beneficial to them.

Veteran investor Charlie Munger says, “Show me the incentive and I’ll show you the outcome.” This guidance becomes even more helpful when we realize that for us Homo ultrasocialis some of the strongest incentives are social.

Microsoft had no shortage of the kind of dysfunction we explored in the previous chapter: bad decisions made by overconfident executives who listened to their internal press secretaries instead of testing their beliefs with evidence.

More employees seeking management slots led to more managers, more managers led to more meetings, more meetings led to more memos, and more red tape led to less innovation.

other. A software designer remembered that “people planned their days and their years around the review, rather than around products. You really had to focus on the six-month performance, rather than on doing what was right for the company.”

Eichenwald, “I wanted to build a team of people who would work together and whose only focus would be on making great software. But you can’t do that at Microsoft.”

To reduce bureaucracy, take away opportunities to gain status that aren’t aligned with the goals and values of the company.

To prevent such a collapse, the business geeks take a radical step: they stop a lot of coordination, collaboration, and communication. It’s hard to overstate how big a change this is from the dominant approach of the late industrial era. When I was starting my academic career in the mid-1990s, business process reengineering was all the rage. This approach to improving performance was based on defining key business processes that cut across several functions (taking an order for a hundred widgets from a customer, for example, involved looking up inventory at the warehouse, having finance do a credit check, telling accounts receivable to generate an invoice, and so on), then specifying in great detail all the coordination required to make the process run smoothly. Reengineering and other approaches that emphasized cross-functional processes and well-defined coordination and collaboration were extremely popular. Executives loved the idea of having a tightly synchronized company that operated like an orchestra in which everyone was playing from the same sheet of music.

A common example is process as proxy. Good process serves you so you can serve customers. But if you’re not watchful, the process can become the thing. This can happen very easily in large organizations. The process becomes the proxy for the result you want. You stop looking at outcomes and just make sure you’re doing the process right. Gulp… The process is not the thing. It’s always worth asking, do we own the process or does the process own us? In a Day 2 company, you might find it’s the second.

One example of this norm is Amazon’s emphasis on “single-threaded leaders.”

Colin Bryar recalls, “In my tenure at Amazon, I heard [Bezos] say many times that if we wanted Amazon to be a place where builders can build, we needed to eliminate communication, not encourage it.”

Cash registers, for example, don’t make it impossible for cashiers to steal money; the bills and coins are right there every time the register is opened. Registers reduce theft not by making it impossible to commit, but instead by making it easier to observe after the fact. They keep a record of sales over the course of a cashier’s shift.