What DAOs can Learn from Holacracies & Decades of Corporate Governance Research
Introduction
Decentralized Autonomous Organizations, or DAOs, have taken the web3 community by storm. It’s not hard to see the appeal — for many, DAOs offer a utopian vision for the future in which everyone has a stake and a voice. However, compared to traditional organizations today, DAOs are highly inefficient. As DAOs grow in membership and capital, they face operational gridlock where simple decisions are subject to voting and lack the muster to pass. Anyone who’s planned a group trip can attest — it’s hard enough to get a handful of people to come to a consensus, a problem that only gets exponentially harder as you scale to hundreds or even thousands of people!
Much of the governance evolution to circumvent this issue has resulted in structures that look and sound like traditional corporations, with power delegated to smaller groups that focus on a certain business function. At a higher level, DAOs also grapple with strategic questions. For example, what decisions need a vote? How do you determine who gets to vote? How do you allocate voting power? Above all, who should even be deciding the answers to these questions?
Although DAOs are new and exciting, they’re facing a lot of problems that have been around for ages. In fact, the question of productive human organization has spanned decades of experimentation and research. We did a retrospective of corporate governance research and found that, at their core, organizations use a combination of authority-, economic-, and trust-based mechanics to coordinate work. To date, most have relied heavily on autocratic power and economic incentives. DAOs are exciting because they offer a conceptual framework for an alternative. However, we see that DAOs often sacrifice practicality for idealism as they try to scale democratic consensus alongside the organization’s scope and scale. While democratic consensus has an important place in DAOs, in overpursuing that ideal, DAOs succumb to the crippling gridlock seen in many democratic institutions today.
We think that the most successful DAOs today will be smaller organizations that focus on drawing ideas from the edges, much like innovation groups within corporations. As they grow in size, DAOs will have to rely more on autocratic power, delegating absolute decision-making power to individuals at various levels of the organization. Democratic consensus will be limited to the accountability layer, acting as a failsafe mechanism to remove decision-makers from power. When it comes to distributing power and labor, humans have rarely, if ever, made massive leaps. Instead, it tends to be iterative, which feels like a snail’s pace in comparison to the speed at which technology is changing. However, this view misses the nuance of how much more efficient organizations can become at the edges by leveraging the interoperability, transparency, and economic alignment that blockchain technology enables.
Why do organizations exist?
As a whole, organizations exist because groups of people working together can achieve more than the sum of the achievements which the individuals in the organization could produce when working separately. For example, a software company benefits greatly from combining the skillsets of computer scientists, marketers, financial analysts, accountants, data scientists, etc. to create and sell its products.
However, for individuals, organizations serve a very different purpose. The why of each person is slightly different, as well as their desires, incentives, and intentions. In fact, it’s common for the individual to be incentivized in a way that’s contrary to the collective’s goals. This is known as the principal-agent problem. For example, imagine you are a struggling trader. You’re determined to make up for your underperformance, and as a result, take increasingly risky bets. Your firm bears the incremental risk while you benefit from the potential upside of the trade. While the firm ultimately wants you to make a profit for it, it might not want you to do so at the expense of such risk.
Part of the allure of DAOs is that they are trying to mitigate or even eliminate the principal-agent problem by better capturing individuals’ incentives and intentions at the collective level. At the same time, they would also benefit from the enhanced productivity that stems from increased intrinsic motivation when people are fully bought into a goal.
This is what makes DAOs so complex — trying to host all the different stakeholders and their intentions versus that of the collective. A big part of the debate is how to balance accommodating an individual’s intentions vs. trying to design the organization in a way that their intentions are more aligned with the collective.
To help better wrap our heads around this, we can break this challenge down into more specific components. Organizations focus on four core problems:
- Task Division — dividing overarching goals into smaller tasks to be accomplished.
- Task Allocation — allocating said tasks to subgroups and individuals. How do you allocate them to achieve these tasks efficiently while considering individuals’ interests and skills?
- Provision of Rewards — mapping rewards to agents to incentivize them to take costly actions to execute their allocated tasks. What kind of rewards are at the organization’s disposal and how do they compare to individuals’ desires?
- Provision of Information — coordinating and distributing the information necessary for disparate agents to work together
These problems are not unique to DAOs, and there’s a lot that we can learn from previous experiments in governance models.
Notable Organizational Structures
Over the past few decades, corporate governance researchers have proposed a variety of frameworks for analyzing governance structures. In particular, Paul Adler, a professor at USC Marshall, has advanced a popular method of analyzing organizations as one or a hybrid of three forms — hierarchies, markets, and communities — and their respective coordination mechanisms: authority, economics/price, and trust.
Over the past 100 years, most profit-seeking organizations formed into rigid, top-down hierarchies as firms. In his essay “The Nature of the Firm,” Ronald Coase states that people organize production into firms because the transaction costs of coordinating production through the market are greater than those within the firm. In the open market, coordination has inherent costs, including price discovery, the costs of negotiating and enforcing contracts, and information asymmetry between market participants. Within firms, there are far fewer contracts. Instead, employment contracts give managers absolute power over their employees. This removes the uncertainty of dealing with changing situations and requirements that arise over time. Imagine if Ford had to negotiate a contract with each of its workers for every task they performed, from negotiating with factory workers for each component they install, to the janitorial staff for cleaning the facilities, to the marketing team for each piece of copy they write. There would be so much negotiation going on that nothing would ever get done!
To get around this, firms like Ford solve the four core problems of organizing by implementing top-down decision-making, with autocratic task division/allocation & rewards provisioning. They also solve information provisioning by reducing agents’ need for information by using directives, schedules, plans, etc — individual workers in the Ford factory don’t need to know what others are doing. As long as the workers complete their assigned tasks, the managers will ensure that they coordinate in aggregate.
This hierarchical structure worked effectively in the manufacturing era, where industrial logic reigned supreme, and firms benefited greatly from economies of scale. However, as we entered the information economy and technological advancements improved our ability to communicate and coordinate more efficiently, market-hierarchy hybrid organizational structures emerged. These hybrid structures generally replaced authority-based coordination mechanisms with economic incentives. An early example is the autonomous profit center; for example, at most investment banks, each department is given its own P&L and is free to pursue profits both internally and externally. More recently, this can be seen in companies downsizing and outsourcing labor, as well as the emergence of the gig economy. Some researchers even argue that the now-prevalent practice of utilizing stock-based compensation for tech workers reflects this trend.
However, both authority and pricing mechanisms are inefficient at producing and allocating knowledge. As we’ve previously discussed, in hierarchies, information is aggregated at the top. On the other hand, reliance on pricing mechanisms for knowledge forces a trade-off between production and allocation. Knowledge production would be incentivized by strong intellectual property rights, which by definition limits the allocation and dissemination of information.
As a result, we’ve seen the emergence of a new form of organization — the community — that relies on trust as its core coordinating mechanism. In communities, people collaboratively achieve tasks and integrate their contributions in a self-organizing fashion, with control and coordination accomplished primarily through direct interactions among themselves. Work within the community is typically characterized by self-assignment to tasks and commons-based peer production. Frequently, communities are guided by shared values, rules, and protocols that help enforce trust among their members.
Successful examples of Community organizations range from software development to open-source drug discovery projects. It’s important to note that most community model examples are limited to non-profit organizations. However, in our research, we came across the concept of Holacracy, a community-hierarchy hybrid form of decentralized management that distributes authority and decision-making through self-organizing teams. Successful technology companies like Medium, Zappos, and Valve have experimented with or even fully adopted Holacracies. As we dug deeper into Holacracies, we found that many of their experiments and learnings are remarkably similar to those occurring today in DAOs. There’s a wealth of fascinating information from these companies’ experiences that can help inform what DAOs will be best suited for, how they’ll evolve, and the tools they’ll need to succeed.
Early Decentralization: Holacracies
First, a bit of history. The Holacracy system was developed at Ternary Software by founder Brian Robertson, who later published Holacracy: The New Management System for a Rapidly Changing World in 2015. Earlier in 2014, Zappos made headlines by announcing it was replacing its traditional management structure with Holacracy. Other companies like Valve Software, Medium, and Github have also experimented with Holacracies to varying degrees of success.
How do Holacracies Work?
The Holacracy system is outlined in its constitution, which is worth a read if you’re interested in learning more. We won’t cover it in full detail here, but in short, a Holacracy is a hierarchy of Circles, which are self-governing, fluid teams that focus on a common purpose. Each Circle is democratically governed, with key decision-making roles subject to elections that can be called at any time. A Circle may be comprised of sub-Circles, and there is a rigid hierarchy of authority between a Circle and its sub-Circles.
Although these Circles are similar to what Divisions / Business Units are in traditional corporate structures, Circles differ in that: A) roles within the Circle are typically collectively defined, and B) Circles tend to have more narrow scopes. As a result, there are usually more Circles than the number of Business Units you’d see in a firm. For example, at Zappos, 150 departmental units evolved into 500 Circles.
Each Role within the Circle has an explicit Purpose, a set of domains that it controls, and Accountabilities, which are the expected activities the Role will enact to support its Purpose. Importantly, every Role is automatically granted the authority to make decisions it deems useful to pursue its Purpose, as long as it doesn’t violate another Role’s explicitly defined domains. Holacracies are notable in part because of the trust they give individuals to identify, define, and accomplish the goals that both contribute to the overall purpose of the organization and match their personal skills and interests.
As DAOs have evolved, many leading DAOs are evolving into structures that sound similar to Holacracies, with sub-DAOs/nests/working groups that focus on specific functional areas, each working towards the DAO’s overarching purpose. Whether through token-enforced mechanisms or social constructs, members have a high degree of flexibility to flow in and out of these sub-structures as they see fit with their skillsets and interests. A great example of this is Index Coop and the massive reorganization they did as part of Index 2.0, where they introduced nests centered around business functions that were comprised of pods that focused on more specific tasks within the nest.
So what can DAOs learn from Holacracies?
Reliability vs. Adaptability
Perhaps one of the most critical questions for DAOs right now is “what are DAOs best suited for?” As we’ve seen, there’s no singular “best” way for people to organize. Instead, organizational design is closely tied to the organization’s purpose. That said, previous experiments with Holacracies suggest that DAOs are best suited for smaller organizations focused on being highly adaptable centers for innovation. Basically, generating and disseminating knowledge vs. building extensive products.
All organizations must strike a balance between reliability and adaptability. Most skew in favor of reliability — from generating predictable returns for shareholders to consistently meeting customer expectations, which is a good fit for the hierarchical firm. Reliability is also essential to the individual members within an organization who rely on the stability that the organization provides to focus on their work. In contrast, organizations need to be adaptable to changing market environments to give themselves the flexibility necessary to innovate and develop new offerings.
Corporate governance researchers have long observed that self-managed team structures are highly adaptable and often lead to innovative new ideas that result in jumps in productivity. For example, in 1947, organizational theorist Eric Trist identified a group of miners in England who had self-organized into autonomous multidisciplinary teams. Within teams, everybody had a stake in the proceeds of their output and was encouraged to introduce new ideas. The result was a leap in productivity when most mining operations were trending lower in productivity and safety.
Self-managed teams became more prevalent in the 1970s and 1980s in the form of participative management & industrial democracy in Europe, continuous improvement groups in Japan, and corporate innovation groups in the United States. Companies ranging from auto manufacturers like Volvo to F&B companies like General Mills saw benefits from implementing self-managed teams. However, in most situations, only a small portion of employees was involved, usually in business areas where adaptability was more important.
Over time, some companies tried to implement self-managing teams at scale. Some were more successful than others, but most found that, ironically, it resulted in more bureaucracy rather than less. For example, Medium experimented with Holacracy but found that it was too difficult to coordinate at scale:
“Our experience was that it was difficult to coordinate efforts at scale. In the purest expression of Holacracy, every team has a goal and works autonomously to deliver the best path to serve that goal. But for larger initiatives, which require coordination across functions, it can be time-consuming and divisive to gain alignment.”
A key takeaway from Medium is that the amount of documentation and process required for Holacracies was a primary gating factor:
“Holacracy also requires a deep commitment to record-keeping and governance. Every job to be done requires a role, and every role requires a set of responsibilities. While this provides helpful transparency, it takes time and discussion. More importantly, we found that the act of codifying responsibilities in explicit detail hindered a proactive attitude and sense of communal ownership.”
These findings have also been in line with our discussions with DAOs. Many of the most effective DAOs today are smaller in size and focused on generating new ideas & knowledge vs. productizing them — from service DAOs like Vector DAO, Raid Guild, and dORG to grant-giving DAOs like Vita DAO. If you think about service DAOs, they’re going after professional services firms that essentially built their businesses on gathering, generating, and selling knowledge. Additionally, their smaller size and knowledge-focused goals limit the amount of documentation required to maintain coordination.
We think that traditional organizations will likely better serve productization use cases, with DAOs participating in that upside via token mechanics. The usage of more pricing mechanics is an area where we think DAOs can really iterate over Holacracies, but there are some meaningful learnings here as well.
Bureaucracy Overload
Like Medium, when Zappos first implemented its Holacracy, employees reported being overwhelmed by the number of meetings the process generated. This created an environment where employees were more focused on fulfilling their internal processes rather than serving the customer.
In response, Zappos introduced an internal marketplace system, where teams act as micro-enterprises within the organization, each with their own P&L. Teams transact internally at market rates and compete for budgets. Rachel Murch, who implemented the system at Zappos, explained:
“Every team had a bank account, and every team had an opportunity to be a customer, a service provider, and an investor. We had service agreements between teams.
If I’m in the development team, I’m providing services to the merchandising team to let them put their goods on the website. There’s a service agreement between those teams. So that’s where you’re getting your money.
And all of that was tracked using a homegrown system. It gave teams the opportunity to reinvest in themselves to grow their team. If they wanted to charge more internally or externally, they could sell their services externally. They could reinvest the way they wanted to.”
The idea was that rather than purely relying on Holacratic processes and their vested authorities to manage the company’s full P&L, the market system removed some layers of bureaucracy and incentivized teams to move quickly to compete, all while maintaining the decentralized and fluid organizational structure. This is a notable departure from the traditional Holacracy system, where budgets are allocated top-down by a finance Circle.
Many DAOs have a similar structure to Holacracies, with a finance sub-DAO in charge of budgeting and allocating treasury to other members of the DAO. As the sub-DAO system becomes more popular, we’d be interested in models experimenting with more vigorous market competition between sub-DAOs. For example, a marketing sub-DAO could charge other members a fee for their marketing services, incentivizing them to allocate their resources to the projects they believe most likely to succeed.
Furthermore, a native token would avoid expensive gas fees and reduce the risk to the overall treasury as the currency is kept within the DAO itself. Over time, these sub-DAOs may even open directly to external parties, giving them more incentives to move quickly and develop their offerings.
Here, we also think there’s a big opportunity for tooling that is purposefully designed to streamline record-keeping and task management for DAOs. This could overcome the barriers that prevented further adoption of Holacries in larger organizations.
Where Democracy Occurs
As Zappos CEO Tony Hsieh explains, a common misconception was that their Holacracy model was a full democracy. He goes on to say that the system doesn’t seek to gain the personal consent of the workers for the organization’s decisions. Rather, Holacracy’s governance process gives everyone a space to improve the organization’s authority and expectation structure for the organization’s benefit, not for their own.
In practical terms, the democratic process occurs in defining the scope and expectations of a Circle, but most decisions are made autocratically:
“The vast majority of specific business decisions made in a Holacracy-powered company are made autocratically, by someone in a clear role with the clear authority to do so (and clear expectations that go along with that authority). However, this autocratic authority and associated expectations are granted and continually evolved via a very different mechanism than the management hierarchy we’re used to. There’s also no “manager” or “CEO” who has the power to trump an autocratic decision made by another duly-empowered role, or direct them on what to decide.”
Many DAOs today are grappling with gridlock and/or voter apathy, with some instances where DAOs have not passed meaningful proposals in months because they consistently lack the relative majority or voter quorum necessary to do so. In this sense, DAOs can try to become more efficient by rethinking when consensus is required and when it makes more sense for members to have autocratic power in their areas of responsibility.
In most cases, consensus is best left as a check on autocratic power. Rather than seeking consensus for operational decisions, it can be used as a mechanism to hold decision-makers accountable and remove them from power if necessary.
Closing Thoughts
We’re still at the very beginning of the evolution of DAOs. Already, we’ve seen DAOs accomplish incredible feats, from managing DeFi protocols that collectively account for billions in liquidity to providing design services to some of the largest enterprises in the world. However, we wanted to highlight the decades of experimentation and research that have gone into decentralized and autonomous organizations and provide examples of what DAOs can learn from them. The research is extensive, and there are mountains of information available that we’ve only scratched the surface of. We wanted to add to the conversation by making assertions that are based heavily on governance theory, which we believe is a gap in existing DAO research.
Most Holacracies failed because the coordination costs were too high. In this regard, this is actually an area where tooling can make a big difference. One of the biggest challenges Medium faced was codifying responsibilities. Tools like Sobol are emerging that help document and visualize contributors’ responsibilities. Additionally, there’s a need for better knowledge management systems for DAOs that help manage and surface the most relevant and up-to-date information across informal knowledge stored in places like Discord and formal knowledge stored in places like Notion and governance forums. Beyond tooling, we think that the coordination challenges will lead to the emergence of a supercharged project manager role that is in charge of maintaining and updating internal communications and documentation.
That said, as investors, we also recognize that regardless of how diligent we are with our own research, we’ll never be as in tune with how DAOs are pushing the boundaries of human organization as those who are on the ground, working in and with DAOs on a daily basis. If you’re working in a DAO, launching one, or building tooling to serve DAOs, we’d love to hear from you at aaron@northzone.com and wendy@northzone.com!
Special thanks to my colleagues Wendy Xiao Schadeck and Dominik Esen, as well as Harry Chiang and Dan Jaeck, for all the feedback and good conversations on this post!
