The Leader Who Creates the Conditions

There is a version of leadership that most of us have been trained to want. It has a clear silhouette: the person at the front who sees further than everyone else, who holds the vision steady when others waver, who steps in at the critical moment to call it. We promote people into leadership on the basis of how well they embody this model. We write job descriptions around it. We design accountability frameworks to test for it.

Linda Hill thinks this model is one of the primary reasons organisations fail to innovate.

Hill is a professor at Harvard Business School who spent a decade studying leadership in some of the most genuinely innovative organisations in the world. Not the ones that talk about innovation, but the ones that actually produce it repeatedly, at scale, under pressure. Her book Collective Genius, written with Greg Brandeau, Kent Lineback, and Emily Truelove, is the result. Its central argument is quietly devastating for anyone who has built their identity around the conventional leadership model: the leaders who create the conditions for collective genius are not the ones with the best ideas. They are not the decisive arbitrators. They are the people who build the environment in which better ideas than anyone could have anticipated emerge from the group itself.

This is not a rebranding of servant leadership, or a softer version of the same thing. It is a structurally different job. And the three capabilities Hill identifies as the engine of innovation (Creative Abrasion, Creative Agility, and Creative Resolution) are each, in their own way, a direct challenge to the instincts that most leadership development programmes spend years reinforcing.

Friction is not the problem

The first thing most leaders do when conflict breaks out in their teams is try to resolve it. The second thing they do, if resolution proves elusive, is attempt to suppress it. Hill's research suggests that both instincts, though understandable, are precisely wrong.

Creative Abrasion is Hill's term for the productive collision of diverse ideas through genuine discourse, debate, and conflict. Not interpersonal conflict (the kind that turns into resentment, factions, and meetings after the meeting) but cognitive conflict: people who think genuinely differently, pushing each other's assumptions hard enough that the ideas change. 

“Abrasion refers to the process of two or more substances rubbing together. That is the key dynamic of a collaborative marketplace where ideas jostle against and contend with each other. As that happens, the ideas change, improve, and perhaps even spawn other, better ideas." Linda Hill

The distinction matters because most organisations that claim to value debate are actually practising something much tamer. Brainstorming, the default format for idea generation in most settings, explicitly suspends criticism. Creative abrasion requires it. John Seely Brown, former head of Xerox PARC and one of the architects of modern innovation theory, put it directly: 

"Breakthroughs often appear in the white space between crafts... These crafts start to collide, and in that collision radically new things start to happen." John Seely Brown

The collision is not incidental to the breakthrough. It is the mechanism.

The practical problem is that genuine cognitive conflict is uncomfortable, and the line between productive disagreement and interpersonal warfare is not always obvious in the moment. Hill is clear that the leader's job is to hold that line actively, amplifying the intellectual friction while suppressing the personal. She is equally clear that most conventional leaders do the opposite. Those who seek harmony above all else limit the range of options their teams consider. Jim Morris, Pixar's general manager, put it with characteristic bluntness: 

"If you have no conflict, you're going to have something that's pretty average." Jim Morris

The Pixar RenderFarm crisis of 2008 illustrates what it actually looks like when a leader creates the conditions for Creative Abrasion rather than resolving it away. Pixar in 2008 was at an inflection point. The studio had been acquired by Disney for $7.4 billion, was scaling from one feature film a year to two or three, and was about to release Up, its first feature in 3D, with a projected value stream of $1 billion. Simultaneously, it was producing Cars Toons, four shorts for the Disney Channel tied to a merchandise franchise worth hundreds of millions. Both productions would need Pixar's RenderFarm, the data centre where the complex computer-generated imagery for each film was processed, at full capacity. At exactly the same time.

Greg Brandeau, Pixar's Senior Vice President of Systems Technology, had built his career on anticipating exactly these kinds of resource problems. When the collision became apparent in late summer, he convened meetings with both production teams to find a solution. Every option was examined. Buy additional computers? $2 million, unbudgeted, and obsolete in three years. Delay one production? Up's release date was immovable; Cars Toons had a Disney Channel air date tied to merchandise cycles. Rent computing capacity? Only 10% cheaper than buying, with the same obsolescence problem. Use cloud computing? The internet was a hundred times slower than Pixar's internal network. Borrow servers from Disney Animation? Incompatible operating systems, file servers, and software.

Every meeting ended in deadlock. Brandeau described the situation with the kind of clarity that only comes from genuine impossibility: 

"When you have these constraints, everyone is forced into a weird box. You're all friends, each of you is handcuffed to the others, and you all have knives." Greg Brandeau

When Ed Catmull, Pixar's president and co-founder of the modern CG film industry, was presented with the problem and the full list of failed options, he did not call it. He did not arbitrate. He did not deploy his authority to resolve the impasse. He said, "Neither. We need both done on schedule. See if you can figure something out."

That sentence is the structural point. Catmull refused to collapse the tension. He pushed the problem back to the group and created the condition for a different kind of thinking to emerge.

In a subsequent meeting, exhausted and running out of ideas, someone asked: "Can we borrow Disney's computers and use them here?" The idea had been dismissed earlier as insane. Disney Animation's facility was in Burbank; Pixar was in Emeryville. But someone reframed it: "We've ruled out borrowing because it's tricky. But we've bought whole new RenderFarms and installed them. How is borrowing any different?" Nobody objected. The group had stayed open, not to what was obvious, but to what had previously seemed impossible.

Disney agreed to pack and ship 250 computers, four to five tons of hardware, from Burbank to Emeryville. Thursday to Sunday: unloaded, cabled, installed, configured. A parallel team spent the same weekend reconfiguring the borrowed machines to match Pixar's environment precisely. The test: render the same frame on a Pixar machine and a Disney machine and get identical output. By Monday morning, the expanded RenderFarm was ready.

Cars Toons delivered on schedule. Up stayed on schedule, was nominated for both Best Animated Picture and Best Picture at the Academy Awards, becoming the first computer-generated film ever nominated for Best Picture overall.

The solution was not generated by Catmull. It was not generated by Brandeau. It emerged from a group that had been kept in productive friction long enough for the frame to shift. The leader's role was to refuse to short-circuit that process.

This is what Mary Parker Follett understood a century before Hill wrote it down. Follett was a political theorist and management thinker working in the 1920s whose ideas were largely ignored during her lifetime and only properly rediscovered decades later. She argued that most organisational conflicts present a false binary: the choice between letting one side win (domination) or splitting the difference (compromise). Compromise, she wrote, is a temporary solution because it leaves the underlying conflict intact. Neither party gets what they actually need, the tension resurfaces, and the cycle repeats. What Follett advocated instead was integration: finding a resolution at a higher level, one that satisfies the genuine interests of both parties rather than trading them off against each other. Integration, she insisted, is only possible when leaders resist the urge to resolve prematurely and hold the space open rather than collapsing it.

Catmull held the space. The solution that emerged was not a compromise between the needs of Up and Cars Toons. It was a genuine integration that met both.

The experiment you're actually running

The second capability Hill identifies, Creative Agility, rests on a distinction that sounds minor and turns out to be enormous. Most organisations, when they want to test a new approach, run a pilot. A pilot is the first phase of something an organisation has already committed to. It is a test run before full implementation, a chance to work out the kinks before scaling. The people running it are personally invested in its success. Their professional credibility is attached to the outcome. Consciously or otherwise, their goal is not to learn what the pilot reveals. It is to make the pilot work.

An experiment is different in one crucial respect: nobody is precommitted to the outcome. Nobody is personally damaged when it fails or points in an unexpected direction. The people running it can remain genuinely open to what the data tells them, including the possibility that the core idea was wrong.

Hill is direct about the organisational consequences: 

"Those running experiments can remain open to learning whatever the experiments tell them." Linda Hill

Those running pilots cannot, because they are not really running experiments at all. They are running confirmations. Most organisations call their confirmations experiments. The language is wrong, and the psychology follows the language.

eBay Germany's story illustrates what genuine experimentation looks like in practice, and what it can produce.

In 1999, eBay acquired Alando, a Berlin-based start-up that had launched only three months earlier and grown so fast that eBay paid $47 million for it. Philipp Justus was appointed country manager. Within a few years, eBay Germany was the fastest-growing market in eBay's global empire and the second-largest by trade volume. Then the corporate machinery tightened. Getting a medium-sized project shipped now took six to nine months. The start-up DNA that had produced the original success was being processed out of the organisation.

In spring 2004, a young product manager and a handful of marketing colleagues decided to try something on a Saturday, without permission, in violation of eBay's corporate processes. They built a Christmas treasure hunt on legacy servers: a €1,000 bonus awarded every hour. Ten million contestants logged on. The site went down. Hackers wrote scripts to exploit each hunt. German engineers worked through the weekend without stopping. By any conventional measure, a disaster.

The team's response: "It was totally crazy. We learned every second. The great thing was that even though the promotion wasn't perfect, it still was a success. We got a lot more traffic."

When Justus, by now in a corporate role that gave him oversight of the German operation, sat with the team and heard what had happened, he faced a genuine choice. The risks to eBay's global platform were real. Corporate process existed for good reasons. He could pull the plug. Instead, he recognised something: 

"I thought these small, quick experiments were exactly what we needed. It was a way to build something fast, without going through the usual development process, but have it be robust enough so as to provide a good user experience." Philipp Justus

Over the following two years, eBay Germany completed more than eighty micro-projects. One of them, EasyLister (a simplified listing form), was eventually adopted by eBay UK, then rolled out by corporate for UK, German, and US markets. From a Saturday afternoon spent breaking the rules, an innovation that reshaped the company globally.

The Pursue, Reflect, Adjust cycle that Hill identifies as the engine of Creative Agility is deceptively simple. Pursue new ideas quickly through multiple experiments. Reflect on what the outcomes actually reveal, rigorously and collaboratively. Adjust subsequent actions based on what you learned, not what you hoped. Then pursue again. The cycle is iterative and genuinely open at each stage, including the possibility that the original idea was wrong and should be set aside entirely.

Hill identifies two paradoxes that leaders have to hold simultaneously. The first is the tension between learning and performance. Genuine experimentation requires tolerance for failure, but rigorous analysis of outcomes demands exactly the kind of performance focus that makes failure feel unacceptable. Both imperatives are real. Leaders who abandon the performance lens in the name of experimentation produce chaos. Leaders who abandon the learning lens in the name of performance produce pilots. The second paradox is between structure and improvisation. Too much structure kills the exploratory instinct that makes experiments valuable; too little produces noise rather than signal. eBay Germany's micro-projects were eventually incorporated into a formal product planning portfolio, and the structure made the experimentation more powerful rather than constraining it.

The application for education leaders is pointed. The vast majority of what schools call "pilots" (new assessment approaches, timetable structures, curriculum designs, pedagogical frameworks) are pilots in Hill's sense. The conclusion has usually been reached before the evaluation begins. The evaluation instrument is designed, consciously or not, to confirm the direction already chosen. Inconvenient findings are quietly set aside. This is not dishonesty. It is the entirely predictable consequence of asking people to run experiments while making their professional credibility contingent on the results.

Genuine Creative Agility requires leaders to separate those two things. The person running the experiment cannot be the person whose job depends on it succeeding.

Hold the tension

The most counterintuitive of Hill's three capabilities is also the one most directly at odds with how we talk about leadership. Creative Resolution is not about making better decisions. It is about resisting the urge to make decisions before the conditions for genuine integration exist. When organisations face competing approaches, the default response is one of two things: compromise (split the difference, displease everyone, lose the best of both options) or dominance (the most senior voice prevails, and valuable thinking disappears with the person whose idea lost). Hill argues that genuinely innovative organisations do neither. They hold the tension deliberately, tolerating apparent duplication and ambiguity, until a synthesis emerges that neither option could have produced alone.

Bill Coughran's management of Google's storage infrastructure from 2003 to 2006 is the clearest illustration of what this looks like in practice, and what it costs.

Google's storage architecture in the mid-2000s was under enormous and accelerating pressure. The existing system, GFS (the Google File System), had been designed for web search: storing large volumes of data and streaming it back at high speed. But Gmail, Google Calendar, Google Talk, and the $1.65 billion acquisition of YouTube all required something different. They needed the ability to handle billions of small files generated asynchronously by users. The architecture was straining.

Two groups of engineers emerged with competing visions. The Big Table team wanted to build software stacks on top of GFS, adapting the existing architecture to meet new demands. The Build from Scratch team, younger, less experienced, and more radical, believed the requirements of Google's new applications were so fundamentally different from search that GFS could not be meaningfully adapted. They wanted to build an entirely new system.

Coughran, then Senior Vice President of Engineering, had a personal view. He thought the Big Table approach was probably better suited to Google's near-term needs. He had experience and instinct on his side. He did not act on it.

"I did not know for sure... No one in the organisation could know for sure which system would be best in the future. I did not want to be top down about this." Bill Coughran

He let both teams work in parallel, at full resource, for two years. He held separate engineering reviews with each group roughly every six weeks, deliberately keeping the teams apart to allow each to develop its own approach without contaminating the other's thinking. When he did intervene, he did so obliquely: 

"If one team was building the perfect left-handed thing, and the other was building the perfect right-handed thing, and you put them in the same room, you may not get anywhere, even with a respected mediator." His characteristic contribution in reviews: "I worry about... [cost, performance, latency, scalability]." Bill Coughran

Not a directive. An invitation to find the data.

His reflection on the timing of resolution is worth sitting with too: 

"I needed to take the risk of letting them run, so that they could come to the realisation, 'Oh, this isn't going to do what I thought.' I exercised personal judgment about when they would be ready to listen to a data-driven argument because people are not always ready for those arguments. You have to know when to cause a decision to happen. If there's an art to this, that's where the art is." Bill Coughran

By late 2006, the Big Table approach was implemented across Google. But the resolution was not a conventional win for one team and a loss for the other. The Build from Scratch team, whose system was not ready for deployment, was folded into the team developing the next-generation storage system, ensuring that two years of hard-won discoveries were not discarded alongside their approach. Three outcomes resulted: a better near-term solution, meaningful progress on long-term capability, and an organisation that had demonstrated it could hold competing ideas in productive tension without collapsing into premature resolution or permanent paralysis.

This is precisely what Follett meant by integration. The choice was never really Big Table versus Build from Scratch. The real question was whether to collapse the tension early, for the sake of efficiency, certainty, or managerial comfort, or hold it open long enough for a third option to become visible. Coughran chose the latter. The cost was two years of apparent duplication. The return was an organisation that understood its own problem more deeply than it could have done any other way.

The structural reason most leaders won't do this is not complexity. It is simpler and more personal. Tolerating ambiguity at scale is experienced by the people around you as indecision. The pressure to call it, to be seen to lead, to demonstrate action, is intense and continuous. Most leadership frameworks treat decisiveness as a virtue in itself. Hill's argument is that decisiveness applied too early is one of the most reliable ways to destroy the conditions for collective genius.

A different job description

Hill's closing synthesis in Collective Genius is worth quoting directly:

"The innovation leader's role is to make sure all these elements are alive and well within his or her group." Linda Hill

That sentence, in its apparent simplicity, contains a completely different model of what leadership is for.

The three capabilities (Creative Abrasion, Creative Agility, Creative Resolution) do not operate independently. They are iterative and recursive: productive friction generates ideas that experimentation refines, and resolution holds the tension between competing refinements long enough for genuine integration to emerge. Each depends on the others. And all three depend on something Hill returns to repeatedly throughout the book: a genuine sense of community. Shared purpose, common values, and rules of engagement robust enough to contain conflict without collapsing into it. Without community, abrasion becomes warfare. Without community, experimentation becomes self-serving. Without community, resolution becomes domination dressed up as consensus.

The leader's job, in this model, is to build and maintain that community and then to create the structural conditions in which the three capabilities can operate. That means amplifying cognitive conflict while suppressing interpersonal conflict. It means separating the person running an experiment from the person whose credibility depends on its success. It means tolerating duplication and ambiguity for longer than is comfortable, and resisting the impulse to resolve prematurely.

Donald Schön, the philosopher and education theorist, wrote in The Reflective Practitioner about the difference between technical rationality (the application of existing knowledge to well-defined problems) and reflective practice, which involves thinking carefully in conditions of uncertainty where the problem itself is not yet fully understood. Schön argued that the most demanding professional work happens not in the application of established answers but in what he called the "swampy lowlands": where problems are messy, stakes are high, and no algorithm exists. Most leadership preparation trains people for the high ground. Hill's three capabilities are tools for the swamp.

This has particular resonance in education, where the accountability framework has been systematically designed to reward certainty and punish ambiguity. Leaders are asked to set targets, demonstrate progress against them, and explain deviations. The system selects for people who project confidence and provide answers. The capacity to hold productive tension, to resist premature resolution, to run genuine experiments rather than confirmations, to create a marketplace of ideas rather than managing disagreement into silence, is rarely assessed and rarely rewarded.

That is not an argument for removing accountability. It is an argument for being honest about what the current model incentivises, and what it quietly trains out of the people we promote.

The most important shift Hill asks of leaders is not a technique. It is a reorientation of where they believe their value lies. The conventional model places it in their knowledge, their vision, their decisiveness. Hill's model places it somewhere harder to measure and easier to underestimate: in their capacity to create the conditions in which people who are not them produce things that could not have been produced any other way.

Ed Catmull did not solve the RenderFarm problem. The team solved it. His contribution was refusing to solve it for them.

Key takeaways

1. Creative Abrasion requires a community, not just a diverse team. Intellectual conflict only stays productive when it is contained by shared purpose, clear values, and agreed rules of engagement. Building those things is prior to encouraging the conflict itself.
   
   
2. The leader's most important intervention is sometimes refusal. Catmull's response to an impossible problem was to decline to resolve it. Pushing the decision back to the group, and keeping it there, was what created the conditions for a genuinely novel solution. Decisiveness applied too early is one of the most reliable ways to foreclose innovation.
   
   
3. Most organisational "experiments" are not experiments. They are pilots: pre-committed, personally invested, and designed to confirm rather than interrogate. Genuine Creative Agility requires separating the people running the test from the people whose credibility depends on the outcome.
   
   
4. Tolerating duplication is a leadership capability, not a management failure. Coughran's two-year parallel investment in competing approaches looked wasteful from the outside. The return, in organisational learning, near-term solution quality, and long-term capability, was only visible in retrospect.
   
   
5. Compromise is not integration. Follett's distinction between splitting the difference and finding a resolution that satisfies the genuine interests of both parties at a higher level is as relevant now as it was a century ago. Most organisations default to compromise because integration is harder and takes longer. It also produces categorically better outcomes.
   
   
6. The conditions for collective genius are built, not declared. Announcing a culture of innovation changes nothing. The three capabilities Hill identifies are structural: they emerge from deliberate decisions about how meetings are run, how experiments are evaluated, how conflict is managed, and how long decisions are held open.
   
   

Brandeau's reflection after both Up and Cars Toons were rendered on schedule speaks to something that sits beneath all three of Hill's capabilities: 

"Our sense of community enabled us to keep going and ultimately find a solution... our deep-rooted sense of responsibility for something bigger than ourselves kept our team from flying apart under this tremendous pressure." Greg Brandeau

The community is not the soft part. It is the infrastructure. Without it, abrasion destroys, agility dissipates, and resolution defaults to whoever shouts loudest. The leader who creates the conditions for collective genius is, at root, the leader who builds and sustains that infrastructure and then has the discipline to get out of the way.