Cooperation and Competition In Organizations: A Way To Get Both

Written by in Bucket list
- 19 min read

In their new book, "Humanocracy", Zanini & Hamel describe how thousands of scientists organized themselves in the Atlas project. This project was to build one of the most important parts of the world's largest-ever machine, CERN's Large Hadron Collider. They organized themselves in a bottom-up structure that relied on peer-to-peer coordination rather than traditional command-and-control. This reminded me of how the former VISA organization was structured. Let me explain how such membership cooperatives work.

What is Atlas?

First, the Atlas project. This is how Gary Hamel & Michele Zanini describe it:

"Consider the Atlas project, one of four research initiatives that make up the Large Hadron Collider. Encompassing more than 3,000 scientists from 180 institutions, Atlas was launched in 1992 in a bid to uncover the deepest secrets of the universe. To that end, the Atlas team built one of the most sophisticated machines ever constructed—a giant particle detector, 45 meters high and 25 meters long, with more than 10 million parts that had to be assembled deep beneath the soil of a bucolic Swiss village.

In the early stages of the project, the Atlas consortium struggled to find the right organizational design. Given the novelty of the undertaking, design and development would need to be broken down into subprojects that could be tackled by small teams. On the other hand, all the subsystems, and there were hundreds of them, had to fuse seamlessly together. Therein lay the dilemma. While autonomous teams would excel at creative problem solving, they’d struggle with high-level coordination. A centralized organization, by contrast, might be better at system integration, but would be overwhelmed by the sheer number of new-to-the-world problems that would need to be addressed. A top-down structure would also provoke resistance from the fiercely independent scientists whose expertise was crucial for success.

In the end, the consortium opted for a bottom-up structure that relied on peer-to-peer coordination rather than a cadre of senior project managers. Every subsystem had its own board, which included all the scientists working on that component of the project. The deliberations within these boards were open and collegial, but could also be heated. In the case of an impasse, opposing teams debated the issue in front of colleagues who then voted for what they believed was the best option. As cross-system issues arose, temporary working groups were convened to hammer out solutions. When, for example, the design of the primary detecting magnet turned out to require more space than originally envisioned, thus shrinking the room for other equipment, a task force was mustered to invent a workaround. Throughout the project, subsystem boards published real-time information on their progress, and relevant experts were encouraged to comment online. At the strategic level, a collaboration board handled major decisions. Every participating institution had a seat on the board, and a two-thirds majority was required to greenlight a decision.

Bringing the Atlas detector to life required tons of leadership and creativity. What it didn’t require was a pyramid. No one within the Atlas consortium had the power to give an order. Everyone was a colleague and no one was a boss. Despite this, the Atlas detector was completed on time and within budget."

From Atlas to VISA

Hamel & Zanini's story about the Atlas organizational design reminded me of one of our most intriguing Bucket List visits last year: our day-long meeting with Dee Hock. We visited him in Olympia, near Seattle, on a beautiful autumn day. Hock shared with us the almost forgotten story of how the former VISA organization worked and was structured.

Hock was the founding force, and CEO Emeritus, of VISA – the worldwide credit card company millions of people rely on daily. His ideas are inspiring, but surprisingly little has been said about them in ‘future of work’ conversations. He proved it’s possible to create a large organization that can cope with a fast-changing environment.

In 1990, Peter Senge described VISA as "one of the business world's best-kept secrets" in his bestseller "The Fifth Discipline". Senge writes: "VISA simply doesn't look like a typical global corporation. It has no publicly traded stock because it is owned by its twenty thousands member organizations. It has no large corporate headquarters because it is a network governed by those members and a written constitution that spells out its purpose and operating principles and the decision-making authority vested in its elected governing boards. It lacks a typical high-profile CEO who guides its strategy (and pulls down an astronomical compensation) because its strategy is in fact many strategies that arise from the thousands of autonomous businesses in the network.

In short, VISA was inspired by abandoning our 'old perspective and mechanistic model of reality' and embracing principles of living systems as a basis for organizing. Eventually Hock even coined a term for the type of organizing he envisioned, 'chaordic,' because in nature 'order continually emerges from seeming chaos, while in management we always try to impose order because we fear that chaos will take over.'"

Hock's design for the organization was built on the same fascinating paradoxes the Atlas consortium faced. VISA and Atlas are both structured as membership cooperatives (MCs). The MCs harmoniously blend cooperation and competition amongst members, without traditional hierarchcies and bureaucracies.

But what does an MC look like in reality? How does an MC work? Let me try to explain, using the former VISA organization as inspiration.

What was VISA?

Let me first explain what VISA actually was. It was not, and is not, a bank. Nor does it keep bank accounts or offer credit to customers. It does not even issue the cards that bear its logo.

So, what does it do? VISA’s product was to create an environment in which member banks could compete and cooperate, to coordinate the efforts of thousands of members, and to lower transaction costs and prevent fraudulent transactions.

To achieve all this, VISA structured itself as an MC in which the members were also the joint owners. From then (early 70s) VISA grew into the world's largest credit card organization, owned by its 16,600 members.

Eventually VISA abandoned the MC principles in 2008. That's why I write about the 'former VISA organization'. Nevertheless, the design principles of the former VISA are an excellent illustration of how MCs can work at scale.

(FYI: VISA restructured into a public stock-issuing corporation in a then-record-breaking IPO which raised $17.9 billion. At the time, they essentially abandoned all MC principles.)

Membership Cooperative (MC)

In MCs ownership and participation of members are forever linked with the goal to focus thinking and actions of different members on the performance of the whole organization—not just their individual performance.

In MCs, clear membership qualifications are set by a Board, but after that any organization meeting those criteria has to be allowed membership. Clear rules of membership fees and voting rights are needed. These are often related to the amount of added value a member generates for the MC.

Moreover, members are not required, nor are they be forced, to purchase services from the MC. As such, the MC is motivated to innovate and stay relevant to earn services fees. Simply, the MC is forced to be supportive of members, and not the other way around.

There are some interesting concepts that make an MC work:

  • Set of principles
  • Collection of meta-rules
  • Fractal organizational structure
  • Fractal governance process
  • Cooperative competition
  • Dispute resolution process

Set of principles

Every MC formulates a set of principles to guide how it achieves its purpose. Hock defines a principle as follows: "By principle I mean a behavioural aspiration of the community: a clear, unambiguous statement of a fundamental belief about how the whole and all its parts intend to conduct themselves in pursuit of the purpose. A principle is a rule against which all structures, decisions, actions, and results will be judged."

Here are five examples of principles the former VISA had in place:

1. The organization must be equitably owned by all participants.

This means that ownership of the organization should be in the form of irrevocable rights of participation. This also means there will be no stock. The rights of participation should not be raided, traded or sold. They should only be acquired by application and acceptance by means of membership.

2. Power and function must be distributive to the maximum degree.

This means that there should be no power vested in, or function performed by, any part of the organization that could reasonably be exercised by any more peripheral part.

3. Authority must be distributed within each governing entity.

This means that governance should be distributed, with no individual institution, particularly management, able to dominate deliberations or control decisions at any scale.

4. The organization must be self-organizing.

This means that participants should have the right to self-organize at any time, for any reason, and at any scale with irrevocable rights of participation in governance at any greater scale.

5. The organization must seamlessly blend cooperation and competition.

This means all parts of the organization should be free to compete in unique, independent ways, yet able to yield self-interest and co-operate when necessary for the good of the whole.

A collection of meta-rules

Every MC should have a collection of meta-rules (based on its principles) to govern the organization on its largest level. These meta-rules regulate governance between the members and state where the authority to amend those regulations resides. They also act as a kind of constitution for the organization and should lock the purpose and guiding principles into the organization.

The meta-rules can at any time be amended by the Board of Directors of the MC. However, this often requires a majority vote of members, which means meta-rules can be changed only when there is an overwhelming consensus among members. (e.g. at the former VISA it required a three-fourths majority vote, at Atlas two-thirds.)

Next, MCs also establish a set of operating regulations. These are the rules of the games that members must follow. They state, for example, how members should coordinate their work, how decisions are to be made, and how conflicts should be settled.

Rules of the game are the coordination mechanisms for the entire organization. MCs ensure they are never set in stone. They should be able to be modified without renegotiating contracts. This is where the fractal organizational structure and fractal governance process play important roles.

Fractal organizational structure

The organizational structure of MCs is not based on a traditional hierarchy, but is inspired by fractal structures found in nature. These structures follow self-similar, fractal patterns and are seen in flowers, trees, plants—even droplets, streams, rivers, waterfalls, mountains, and coastlines.

article photo

In nature, fractals repeat a pattern at different scales. Examples are everywhere. Take the fern example above. But also in forests, trees, branches, leaves-each is a copy of the one that came before.

Fractal organizational structures are based on the same principle. Below I sketch a simplified representation of the how a fractal structure would translate to organizations.

article photo

In such fractal structures the Level I circle (e.g. the international level) represents the whole organization, functioning as an umbrella for Level II (e.g. the regional level) and Level III (e.g. the national level). These are separate legal entities, with relative autonomy and their own Boards of Directors.

(For this example I chose three levels. No fractal organizational structure is bound by this number of levels, nor is a fractal in nature.)

MCs are designed to balance the interests and powers of all members, regardless of their size or location-which ensures no single member or person can dominate the whole organization. This divests power from the centre, and allows local parts to address their unique needs, create their own policies, and develop local product and services.

Fractal governance process

The fractal organizational structure goes hand in hand with a fractal governance process. This enables each level to have similar relationships to other levels, smaller and larger. It also means each level has similar rules governing behaviour and distribution of authority. This helps different levels to coordinate. I sketch a simplified view of a fractal governance process in the image below (for the fractal organization shown above).

article photo

In this fractal governance process, every level had its own Board of Directors. Each level designs the meta-rules for its level, to help members reach consensus among themselves. The Boards of Directors are often elected starting at the member level all the way up to the largest circle. This means, in our example, individual members elect the 'Level III directors' , ‘Level III directors' elect the ‘Level II directors', and finally ‘Level II directors’ elected the Board of Directors at Level I.

Meta-rules should be designed and enforced in the opposite direction, from Level I to the member level. Still, every level should amend its own meta-rules, but they must conform to the meta-rules of the levels they are part of. This enables the organization to enforce standardization while respecting local deviations.

The whole loop of this complex structure is closed by the fact that the ‘largest level’ of the fractal (Level I) grants its authority to the ‘smallest level’ (members) of the fractal by agreeing to the collection of meta-rules that demand that the Level I Board of Directors must be made up of people from the members.

The fractal governance structure means that the activities done at each level are largely identical. All levels are responsible for their own behaviour and activities. They can, for example, set and enforce their own membership rules.

However, the individual members are still the most important part of the MC. They must stay autonomous in the sense that the MC should not regulate the relationships members have with their customers. The members should keep the right to develop and market any service or product they wish, as long as they conform to the meta-rules.

But note, in MCs the members also hold the primary responsibility and accountability to their customers and to their respective local government institutions.

Govern by consensus

All levels of the fractal should use a standard governance process, where members make and implement decisions based on the authority outlined in their respective meta-rules. Obviously, there will be some differences between, for example, the responsibility and accountability of Level III versus the Level I, but the overall processes should look largely the same. Such standardized processes have another benefit. Board members can easily serve on any other level.

Often, next to the Board of Directors there are ‘committees’ run by subject-matter experts. These provide Boards with the necessary knowledge in specific areas (e.g. marketing, sales, technology, etc.). New issues often start with, and are developed by, the committees before they are presented to the Board for final approval. This has the advantage that committees can “take the temperature” among different members about potentially controversial issues without immediately initiating a widespread debate. It also means issues and ideas come from, and are advocated by, the general membership—not just the Board.

During a Board governance process, all issues (initiatives, approvals, budgets, etc.) are decided by democratic vote, forcing members to govern by consensus. Hock, for example, designed the process at VISA as deliberately democratic to make sure no one bank, nation, group, or region could become a dominant player, and able to act solely out of self-interest. This doesn’t mean some players did not have more influence than others. It just made sure this influence didn’t reach unhealthy levels.

Hock was more than aware that this process of consensus seeking was not necessarily efficient. But he was convinced that it was extremely effective. Hocks says: “We always take the word consensus to mean everybody has to agree. But consensus is really the agreement of the people that care, with the acquiescence of those who don’t particularly care.”

He knew consensus usually means the ultimate solution is sub-optimal at some level for everyone in the organization. But he also knew that, due to the fractal structure of the organization, most decisions would be pushed to the smallest level possible, meaning a level where the problem could be handled most effectively. This also meant issues that reached the largest level were few: they had to be of importance to all members in the organization.

Hock says: “It is a very counterintuitive. When you look at such a structure with eyes conditioned by mechanistic, industrial age organizations, the assumption is that such an organization would just simply disintegrate, because it is too complex to make any serious decisions. But quite the opposite occurred. By pushing all, or most, of the authority and activity to the smallest and most peripheral places quite the opposite occurred. And it has to do with self-organization and self-regulation."

Cooperative competition

It is not hard to imagine that creating harmony in these communities of cooperating competitors is a delicate process. On the one hand, members want to cooperate with other members to increase their collective performance. On the other, each member also wants to beat members competing in their market. This creates an almost perfect example of the so-called Prisoner’s Dilemma: members have incentives to compete, but each would be better off if all decided to cooperate.

That’s why the MCs build cooperative and competitive forces in at all levels. The cooperative ones fosters effectiveness, and the competitive ones efficiency. Interactions between these seemingly opposite forces make the organization resilient, robust and successful. They try to achieve a minimum of order, and a maximum of chaos.

Take the example of a product innovation at the former VISA. Members enjoyed enough autonomy to develop products for their customers. But they had to make sure these conformed to the guidelines and standards developed by VISA. This was needed to ensure inter-operability with other products in the system, potentially built by other members. As a result, members developed a wide variety of products and let customers decided what they liked most. This meant the failure of a product could lead to a serious disaster for a single member, but leave the organization undamaged.

Thus, product innovation at VISA was not coordinated within the organization, but rather by market forces outside. It was said to increase the chance of finding successful solutions, because many members innovated along the same lines. This avoided lots of coordination and consensus-building in innovation. It became an emergent and efficient process. This was important. It allowed the organization to develop very effective solutions. At the same time, it preserved each member’s autonomy and the incentive to compete—albeit with the same members with whom they cooperated.

Hock says: “If any bank innovated in some way and its innovation was not effective, it died out at the very small level and didn’t affect the whole. But if a bank, anywhere, innovated by providing something that was truly effective, knowledge of it would propagate through the system in a matter of weeks and would be emulated by other banks. So, innovation was continual and bad judgement died out without noticeably affecting the system. But successes propagated and continually built it.

And so, the more that each individual bank innovated and competed effectively, the more need it had to cooperate to participate in the global system. And the more it cooperated in its essential parts, the more it could compete. So, competition and cooperation existed simultaneously and drove one another. It became an incredibly successful program. In its early days it was growing at 50% compounded annually. Which is almost hard to believe, but that is exactly what happened.”

Dispute resolution

It is also not difficult to imagine that MCs need a solid dispute resolution process to keep running smoothly. In fact, this is a critical mechanism in most organizations that try to harmoniously blend cooperation and competition. Simply, whenever competitors engage in cooperative activities, there is potential for conflict.

The former VISA organization, for example, had a clear and formal arbitration process to resolve disputes as fairly as possible. It consisted (again simplified) of a few steps:

  • First, all members had to agree to follow the meta-rules as stated above. This set of rules also dictated how disputes were to be resolved, and included a simple mechanism to resolve things among themselves.
  • Second, when the first step proved unsuccessful, members could adopt a formal arbitration process administered by a ‘dispute resolution group’, a jury of their peers. Members submitted their compaint, along with supporting documentation, to what acted as an internal ‘VISA court’, which would review it and decide. Bringing a case to the internal court came with a fee to discourage frivolous use.
  • Third, these decisions were ‘final’. But members could appeal a decision, with the Board of Directors as the supreme arbiter.
  • Fourth, members had the right to bring a suit against one another in their country’s legal system if they did not like the result of the arbitration.

Different MCs have different kinds of dispute resolution processes. But no matter how the process works exactly, it should be in the interest of all members to make it as efficient as possible and to minimize the number of arbitration cases. So, if specific rules are causing many disputes, a ‘committee’ should suggest changes to clarify them. This committee should also educate members about rule changes in the dispute resolution process, to minimize future disputes.


The Atlas project and the former VISA show that MCs can be a viable, alternative design for large organizations that aim to seamlessly blend cooperation and competition among members, be they teams on the Atlas project, or companies at the former VISA organization.

What do you think? Do you think this kind of model deserves more attention? Or are they simply too complex?


leave a reply

Replies (13)

Sjaak Evers

Sjaak Evers

More attention, no doubt about that. Applicable in public organizations like national states and the European Union?

| | 3 | Flag
David Collins

David Collins

This is a brilliant organisation design Joost, and mimics my own experience of what really works in large many-actor systems.

| | 2 | Flag
Peter Brownell

Peter Brownell

VISA's story has been my inspiration for years. I think that Dee Hock's work, and all the Chaordic Commons, has somehow been left out of the contemporary dialogue on organizational design.

Reading Dee's book, I suspect that VISA's principles also meant that they were successfully using agile principles a decade or two before the manifesto! (I've not seen any further description of the Dirty Coffee Cup methodology anywhere, but it does seem to mirror the basics of Kanban)

| | 2 | Flag


Great article. The multiple levels of boards and meta-rules are a great example of co-operation and empowerment.
This seems quite similar to the story of Gujarat Co-Operative Milk Marketing Federation (GCMMF) in India. This is the largest co-operative organization in the world with revenues of over USD 5 Billion. They market its product under the brand name AMUL.
GCMMF was founded around the time India gained independence in 1947 and today has more than 3.6 million members in the co-operative. Their pioneering work led India to become the largest milk producing country in the world. Something for you to look at.

| | 6 | Flag


Great article. The multiple levels of boards and meta-rules are a great example of co-operation and empowerment.
This seems quite similar to the story of Gujarat Co-Operative Milk Marketing Federation (GCMMF) in India. This is the largest co-operative organization in the world with revenues of over USD 5 Billion. They market its product under the brand name AMUL.
GCMMF was founded around the time India gained independence in 1947 and today has more than 3.6 million members in the co-operative. Their pioneering work led India to become the largest milk producing country in the world. Something for you to look at.


Thanks Balesh for this tip! I will definitely start looking into GCMMF. Are there any resources you would recommend reading?

| | 1 | Flag
Hannes Alton

Hannes Alton

Great article Joost. The model itself deserves absolutely more attention. Is the VISA model based on the Viable System Model of Stafford Beer? There are some very interesting parallels.

| | 2 | Flag


I'm curious on why did VISA abandon their MC module. And that also makes me think - would be interesting to research companies who have been holocratic or bottom-top, but then have failed or changed and became more hierarchical. What kind of similarities do they have and have there been similar reasons to fail?

| | 5 | Flag
Kirsten Gibbs

Kirsten Gibbs

Thanks for this one Joost. I love it, and am off to find out more about how this works.

I've been enjoying your book, and I have another candidate for your bucket list - a small manufacturing company on the south coast of England:

I think you'll enjoy meeting them.

Let me know how you get on.

| | 1 | Flag
Dil Green

Dil Green

Governance in the service of enabling participants to do what they consider the most useful work they can do:

| | 0 | Flag
Flags are private, only visible to forum moderators. Be specific: "It's spam/off-topic/inappropriate because..."

Leave a reply

This field is required. Only use letters, digits, spaces and minus characters.
This field is required.
This field is required.
We only use your email for spam detection purposes.