Size Matters! Or, How to Scale A Dao
DAOs are at the forefront of organizational development. They embody the pinnacle of decentralized and completely remote, sometimes even anonymous, teams. Not all DAOs have a distinct economic purpose, but every DAO wants to achieve something.
I wanted to know if there’s an optimal size for DAO teams. How can the fact that teams are fully remote, and contributors work as much or as little as they want, affect the optimal team size? Should teams be smaller or bigger than in traditional, remote orgs? And what forces and effects come to bear on team members?
We have found time and again that size matters. There’s a threshold where teams stop working efficiently, and the friction inherent in reaching a consensus as a group starts to lead to frustration and mistrust.
In my blog post about network effects, I introduced Dunbar’s number, or about 150 members, as a possible limit of how many members a DAO can have and still feel intimate and full of “spark” and trust between members. To recap: Robin Dunbar is a British anthropologist who studied group sizes and wanted to establish an upper limit for the number of meaningful relationships someone can have. He posited that this number is around 150 on average and results from a cognitive limit.
If humans can have about 150 real-life connections that mean something to them, can they have more or less virtual connections? And does communication technology change that number?
Follow me on a journey behind the curtain of science, going into Dunbar’s number and the research covering online communications and their effects. I found out that, most likely, remote teams need to be smaller than their real-world counterparts. Mainly because the number of connections outside the team is bigger and more meaningful for members of remote teams, and the way humans interact online is wrought with more friction. But I’m getting ahead of myself. Let’s dive in.
The science behind Dunbar’s Number
In an interview, Robin Dunbar once said that he wasn’t the one who invented the term Dunbar’s number and was surprised as the term sprouted its wings, culminating in a mention by Malcolm Gladwell in his book Tipping Point.
Dunbar’s central thesis is that the number of meaningful connections a person can maintain is a function of their brain size, which he took as a proxy for cognitive capacity. In a paper discussing Dunbar’s research and published by the Royal Society’s Biology Letters, authors Wartel and Lind summed it up:
“The number 150 was established by […] describing the relationship between group size and relative neocortex size in […] humans. That there exists a correlation between group size and relative neocortex size has been replicated in several studies […]. However, the replication studies are of somewhat limited value as most studies have used the same brain data.
The expected human group size of 150 has been substantiated by observations of human communities with group sizes ranging between 100 and 200, including hunter-gatherer communities, military units, businesses, 18th-century and Neolithic villages […].”
The authors then tear down Dunbar’s methodology by trying to replicate his findings using Bayesian tools but coming up with different numbers and huge credible intervals, where 95% of the upper group size limit numbers fall within a range between four or 300 people, for instance.
Apart from the methodological issues, I’m seeing a lot of assumptions going into Dunbar’s number, personally. My main objection is that the limits to human connections are a function of brain size. I’m confident that a cognitive limit to the number of meaningful relationships we can maintain exists. Still, I think it depends more on cultural and social patterns and personal history than on brain size.
Remote work, real friends, and the grind
Harvard Business Review author Pamela Green identifies three kinds of distance in remote collaboration:
- physical (place and time),
- operational (team size, bandwidth and skill levels) and
- affinity (values, trust, and interdependency).
She concludes that “the best way for managers to drive team performance is by focusing on reducing affinity distance” by using video calls instead of email, allowing space for personal, non-transactional interactions.
Companies often tout “work is like family here” on their hiring pages to attract talent. The word family here wants to evoke a sense of belonging, warmth, and loyalty. Especially young people who just recently moved out of their parents feel attracted to this messaging. Companies do their best to keep up appearances by providing free food, health care, massage therapists, and cozy, fun offices.
Apart from discussing whether this promise has darker edges, we can infer that family is one of the main providers of meaningful connections. While workplaces with real offices and cafeterias can credibly abscond employees into a conjured “workplace family,” the same is much harder for remote-only teams.
In addition to physical distance, hierarchies have tended to become flatter during the last decades. An Economist article on “span of control” sums up the thinking over the past decades: “For the first 60 years of the 20th century, when managers favored a structure based largely on military models, a consensus formed around the number six (employees per manager). After 1960, however, management styles began to change. Flatter, less hierarchical, and more loosely structured organizations implied larger spans of control. The consensus on the size of the ideal span rose to between 15 and 25.”
Remote teams and flat hierarchies often mean little time with superiors and add to the feelings of distance already created by a lack of physical contact. Which further reduces the illusion of work being “as a family.”
Remote-only teams have more meaningful connections outside of them and even more outside of DAOs, where contributors often have no ties and work project by project for as little or as little as they want. It is hard to expect someone who works for 15 hours a month to feel a strong sense of belonging. The focus shifts towards the work and the common goals and values in a DAO, which become much more important than in traditional organizations.
As connections outside of work become more important and numerous, the size of a team that can operate with cohesion shrinks. If Dunbar’s number has any real implications and we have 150 meaningful connections, and 80 are outside of work, the maximum viable team size is 70. Please don’t quote these numbers verbatim; they just drive home the message. The takeaway of this chapter is that there is ample reason to believe that remote teams should be smaller than teams with a physical presence and that loose ties between DAOs and their contributors mean that optimal DAO teams are even smaller.
Communication and friction
What about bandwidth? If Dunbar’s maximum number of connections is a cognitive limit, what role does the way people communicate play? And for DAOs specifically, do chats on Discord and forum posts reduce or increase viable team size?
We’ve all seen people tout the number of Facebook friends or Linkedin connections. But are these meaningful? I’d definitely not want to form a company consisting of my 6,558 Linkedin connections because I’ve never interacted with 95% of them beyond a sentence or two.
Famed media theorist Marshall McLuhan said that “the medium is the message,” so how do Discord chats compare to a physical interaction? There’s a considerable amount of research showing that non-verbal cues are essential to understanding others, and they convey meaning. Chats don’t have these cues, so they take more work to follow and make others harder to feel, for lack of a better word. At the same time, Discord channels abound as DAOs grow, and notifications add a sense of urgency and pressure at the exact moment when we’d do best to read sentences twice to make sure we understand what the other person is trying to say instead of what we interpret it to mean.
Video calls, emojis, GIFs, and articulating better help, but still lack important cues like seeing the other person move as a whole or more subliminal messages like smells and whether their shoes are polished.
We can find in all the research we’ve seen that pure online relationships tend to be more shallow, and text-only conversations introduce additional friction into communications. The biggest drawback associated with online work is the potential for loneliness or social isolation. This can devastate the mental health of contributors and lead to lower morale, reduced productivity, and increased turnover. Toxicity in chats is a great way to gauge if DAO members feel close and trust each other or not.
Toxicity is defined as hostile or sometimes even threatening behavior, often directed at individuals who do not conform to the norms of a space. It impacts everyone involved, as a single instance of toxic behavior disrupts the community dynamic and may have long-lasting effects on the group.
DAOs do well to create fun activities that bring contributors together in a way that doesn’t contribute directly to the organization’s goal. Something where contributors can interact personally or on a richer level. DAOs need spaces for meaningful interaction, so contributors form connections that last.
Even if some DAOs work towards automation and immutability of processes, they still profit from forming a community of contributors that feel involved, seen, and cherished.
Two things to look out for to judge if a team is too big are:
- Is the team on the same page? So called know-how integrity.
- How long does it take to make decisions?
Know-how integrity is when people know about the same things and agree on them. If definitions of goals, processes, and deliverables drift apart between team members, it might be time to reduce team size.
For speed of decision making, let’s take a look at an azcentral article that delves into the best ratio of managers to direct reports. The article suggests that larger ratios are better when managers are more experienced and can make decisions quickly. They should also provide clear directions and be willing to delegate authority without second-guessing employee decisions. In other words: the easier the team comes to a decision, the bigger the team size can be.
DAOs are remote teams, too
How are DAOs different from remote teams? Do the same restrictions apply?Let’s compare DAO teams with remote teams in traditional organizations side by side.
We can see that DAOs allow teams to be as loose as their members want. But DAOs have a clear common purpose and most contributors start work for free because they want to pitch in.
Traditional corporations, on the other hand, focus on team building and selecting team members so they fit in. But we all know untold stories of dysfunctional teams with toxic culture.
All in all we cannot say that DAO teams are always different from remote teams in traditional corporations. It depends on how the DAO operates. As a tendency DAO teams are probably more loosely coupled and this clearly makes relationships outside the team even more important, and reduces the maximum size for a well functioning team even further.
And that leads to the million dollar question:
What can DAOs do to scale well?
Flipside Governance has exposure to more than a dozen DAOs on a daily basis. We get to see a lot of thinking and experimentation that we scan for greatness and document.
All the great approaches to scaling that we’ve seen take a very organic approach to scaling DAOs. They take a page out of nature’s playbook and scale in the way that organisms evolved. From single cell to multi celled organisms. Some have very tight integrations of offspring, to the point where they don’t have a separate face to the outside world, while others just share a common framework.
Pods or SubDAOs
The most popular scaling system are Pods, sometimes called SubDAOs. Pods were first introduced by MolochDAO and pioneered by Orca.
Pods work entirely within the existing structure of the existing DAO, and usually add a separate tab or channel on Discord. See the screenshot from MetricsDAO below for a simple and clean example.
Bounty-ops, growth, protocol-dev and so on each have their own channel but use the same Discord server and forum. This approach is great for early stage DAOs with a decent amount of activity. Channels allow members to focus on one topic at a time and find other team members with shared interests.
As Pods grow, channels become tabs with multiple channels. See the screenshot of BanklessDAO to the right.
Gardens
The 1Hive community took an approach that takes decentralization and permissionless building to the next level. Based on a framework for token issuance, knowledge sharing and governance primitives, 1Hive DAOs can have their own missions and flavor and form so-called Gardens.
To quote the 1Hive operating manual: “Communities can use Gardens to launch their own token or, if they prefer, to import a pre-existing ERC-20 token. From there, they can use whichever tools they prefer to coordinate: chat rooms, forums, etc. The token will be used as a governance token for deciding how funds are issued and how governance parameters are updated.”
Gardens are very close to the intent of “The DAO”, the first large-scale decentralized fundraising effort on Ethereum, where users could propose projects that would then be funded from a common pool to bootstrap a decentralized economy.
1Hive is still around but has struggled to get real traction in the markets. Lack of singular focus might be partly to blame for that. While the 1Hive framework makes starting a DAO easier than before and DAO founders can participate in a community of builders and exchange tips and tricks, customers don’t have a unified experience and this leads to a dilution of uptake. In Web2, Shopify has a similar approach and has led to massive success both for itself and for an entire industry building on top. But so far 1Hive DAOs have not had the same appeal to consumers.
MetaDAOs
MakerDAO co-founder Rune Christensen recently came back to the DAO after a hiatus and proposed a massive, wide-ranging overhaul of almost everything. Focused on leading the DAO to a fully decentralized and largely autonomous future, Rune posted the details in a forum post so large that it broke the Forum limit and had to be released as two separate posts.
At its very heart is the concept of MetaDAOs, born decentralized DAOs that result from clustering in MakerCore, the engine producing DAI from crypto and real-world asset vaults.
Maker would have a DAO gestation cadence that releases a pool of new tokens every so often. Community members can then apply for the tokens and their proposals get voted on by MKR holders.
MetaDAO tokens are tightly integrated with Maker through yield farming in Balancer Pools and by emitting yield to MKR and DAI holders. They can also appeal to MakerDAO governance as a kind of decentralized supreme court, or arbiter of last resort.
Here is a true evolution of how DAOs scale, that is in some ways similar to how organisms evolve. From single cells to complex organisms with a wide range of specializations. All of it without any central control. If MetaDAOs find a market, they thrive. If not, they wither and die.
Christensen plans to put the initial proposals to vote this September 2022, so there’s a good chance that our esteemed readers will be able to watch this experiment take place in one of the largest DAOs out there.
Cells
Professor James Baron inspired a biotech company (not a DAO) to scale in a decentralized way, using Cells. In an interview with Yale Insights, he describes the rationale behind the system: “The idea the company uses to describe growth is cellular division. That’s the mindset that their scientists come to naturally, so they have adopted a model of spinning off cells that are to be no larger than 50 to 70 people — a size at which people can manage relationships with one another in a trusting way. There’s neuroscience that supports that view.”
Similar to MetaDAOs, Baron embraces the concept behind Dunbar’s number and respects that centralized decision-making comes with a cost. Longer lines of communication, executives that make decisions with limited and distorted information, and slower responses to external shocks are common attributes you can find in almost any organization with a massive scale.
Baron’s work replaces central authority with commitment, where new employees are hired who want to commit to the company for the long haul and are open to changing jobs and roles depending on what the organization needs, in return for the ability to grow with others and have direct input into the company’s goals. Much like the ideal family, if you will.
Robert Koch, one of the founders of Koch Industries, revealed a similar approach in an interview with Tim Ferris once. He said that many of the best businesses were born when employees professed unusual initiative and willingness to run with an idea they had, even if it meant considerable risk for their career.
Conclusion
Scaling human coordination is arguably the most important problem to solve as we enter this millennium in full. If humans trust they cooperate better and this reflects in their output. The ability to form tight, functional groups that are highly flexible and yet effective with quick turnaround time and relentless performance is what will differentiate the best from the also-ran’s.
If there’s one thing we want you to take away from this blog post it is this: Human connections and trust have a cognitive cost. And they are one of the most profound needs. Studies have shown that loneliness can reduce life expectancy and quality of life like little else.
DAOs offer a new way to work together and to meet people from all over the world. They can be places where great things get done or pits of toxicity. “More is more”, as guitarist Yngwie Malmsteen quipped, is not the case for DAO teams. Governance needs to keep their fingers firmly on the pulse of the community and experiment with scaling approaches when teams get bigger.
DAOs can evolve like organisms, that is their true potential. We live in an exponential age and need new forms of cooperation that can scale exponentially. Centralized control just can’t keep up.
