How Dunbar’s Number Makes Or Breaks A Community

We think of internet communities as one collection of people all interacting with each other. This isn’t true. People can only maintain a certain number of relationships with others. They have limited time to gather information about a group, build relationships, and stay engaged. This limits the size of the group they can interact and form relationships with, and that limit is Dunbar’s number.

Dunbar’s number is the cognitive limit of the number of stable relationships someone can have. It is commonly known as 150, although the practical number for a community is lower because people always have relationships elsewhere. 150 is the max where a large amount of time is being spent on “social grooming.” Christopher Allen’s hypothesis on the consequences of varying group sizes states:

the optimal size for active group members for creative and technical groups—as opposed to exclusively survival-oriented groups, such as villages—hovers somewhere between 25-80, but is best around 45-50. Anything more than this and the group has to spend too much time “grooming” to keep group cohesion, rather than focusing on why the people want to spend the effort on that group in the first place—say to deliver a software product, learn a technology, promote a meme, or have fun playing a game.

For in-person communities, Dunbar’s number isn’t a major problem because of physical constraints. For internet communities, it is. Through the lens of Dunbar’s number, internet communities become a collection of many groups under one banner, space, or identity. Growth causes changes to community dynamics which need to be handled to maintain quality and functionality.

A History of Communities and Dunbar’s Number

Before the internet, communities were physically constrained, both by geography and space. First, there had to be enough people in your city or town who shared the same interests. Those people had to discover each other then coordinate events or meetings. This is a lot of work, so people rarely ventured away from the communities they were a part of. If they did, they likely joined previously established ones.

When someone did manage to create or join a community, members were still constrained by space. Everything happens in person. Only so many people can fit in a space, and few spaces are designed to hold more than 150 people. Members likely interact with only a small number of the meeting attendees each time.

All of these put natural limits on a community, helping it grow sustainably and avoid problems with Dunbar’s number. The internet disrupts them.

Without physical constraints, communities can be created easily and grow rapidly. They can quickly reach and pass Dunbar’s number without realizing it. In-person communities can see and hear problems through overcapacity, noise, and complaints. Internet communities often don’t have these signs. Members leave without a reason, and problems don’t get handled.

Handling Dunbar’s Number

A common problem caused by Dunbar’s number is the quality of information degrading. A common way solution is “splitting” the community. This might be by topic, interest, content type, or permission. This happens because platforms give the option to do it and it is what others are doing. For example, take two Kanye West-related Discord servers:

You can see the evolution of a community over time. It is likely the “Kanye” server had a similar amount of channels as the “West Server Ever” when it was smaller. The “West Server Ever” has started to split, separating “general” discussion from “(kan)ye” discussion (same as “Kanye” server). They are also starting to split up less relevant topics like “politics” and “ddt-hate.” It is easy to foresee future splits as the server grows.

Splitting is an example of passively handling Dunbar’s number. It doesn’t create a significant change to the community. The structure, behaviors, habits, and policies stay much the same. Minor rule changes, roles, and stricter moderation are all similarly passive solutions.

The problem with passive solutions like splitting is that they may not solve the problems the community is having. Splitting still lowers the quality of main channel discussion as it causes the community to grow away from its original area of focus. More channels distract from the main focus. These non-focus areas are likely lower-quality because members didn’t join for these areas. The more a community splits, the more it moves away from its original purpose.

Not handling Dunbar’s number well hurts member experience. It causes members to be less satisfied and spend time elsewhere. A “sense of community” is broken. Whether they realize it or not, members will look for other communities that better handle Dunbar’s number to provide a balance of connections and knowledge.

Communities that are valuable for their members spot and handle challenges with Dunbar’s number. They continue to create connections while facilitating interactions to benefit members and create collective knowledge. A passive solution may not be the best solution, it may require a more active approach.

Handling Dunbar’s Number Actively

Although the internet has caused more communities to face challenges from growth, it also provides more solutions. There are many options beyond splitting, rule changes, and stricter moderation. Communities can apply more active solutions by changing their structure, limiting their size, and developing means of making connections.


Choosing a platform is important, among many other reasons, because it provides structural solutions to growth. The structure of the platform creates the behavior that happens there. Some platforms have more options and customizability than others. Certain structures allow for better conversation at larger sizes, better overview of activity, or better facilitation of connections. As a community grows past Dunbar’s number, the structure becomes especially important.

The most common space for internet communities to exist is the channel, whether that is a group chat, Slack, Discord, or somewhere else. The problem with channels is they isolate information into silos. It can be difficult for members to find areas that may be of interest to them. There is rarely a way to see historical activity or get an overview of what’s going on. Those silos also aren’t information-dense, when compared with forums or posts. A conversation between two people can quickly take up the whole page.

Low-density Demand Curve Growth Slack channel

Posts and forums are more information-dense. With them, members can see a variety of conversations happening at once, and figure out what is popular. They provide better access to “best-of” and historical content. Take a look at the supertalk forum. Their homepage is dense with information, with forums branching off forums. Members can, at a glance, see the areas of focus and the active areas.

High-density supertalk forum

Every structure has a trade-off. Communities must ask themselves if their structure reflects what their purpose is. Channels might be the right decision if the goal is real-time and create casual conversation. They work well when topic areas split easily. If neither of these is the case, exploring another structure may be a better option. Luckily, there are many platforms and options to explore. Being able to tailor these platforms to specific use-cases is becoming easier, meaning better structures are being found.


A second way of handling Dunbar’s number is best explained through the online education industry. In the early 2010s, MOOCs (Massive Open Online Courses) were all the rage. Millions of people took them on a range of topics. The problem is few people complete them, causing a lot of potential learning to be lost. A cause of this is the failure to solve Dunbar’s number for their community. They did little to create connections between students which motivate and help them learn more.

Recently, many courses have found success with cohorts (cohort-based courses). Rather than being “massive,” they are small to medium-sized. This allows students to build connections with other students and creates motivation to continue in the course. It also adds new ways of interacting and learning from the course materials. Since cohorts are smaller, they handle Dunbar’s number better than MOOCs and are more valuable to the average member.

Communities can learn from this example by changing the way they look at their current size and their growth plans. Being as large as possible doesn’t always align with the goals of the community or its members. Paid membership or required expertise/credentials are often more sustainable for managers while providing more value to members. On top of paid and credentialed communities, other ways to experiment with size include:

  • Limit size with DAOs and tokenized communities by forcing ownership. Align incentives through the creation of owner-members.
  • Create copies/cohorts of the same structure of the community. Add every 150 (or less) members into a copy of channels private to them. Engage each set of members separately.
  • Maximize exclusivity with the nightclub approach. Put a low cap on the number of members, let in high-status members who loudly promote the community, remove inactives or trolls, keep a waitlist, charge more. It isn’t inclusive but makes sure every member is serious about being there. It also increases loyalty through the prestige of being included.

The idea that a community needs to be open and as large as possible doesn’t always create the best experience. Smaller communities often better align with members’ goals, even if they don’t realize it. They are places for higher quality conversation and connections. I’ve written before about why people need to explore higher friction communities, but they need to exist for people to join them.


Finally, communities can create connections at the individual level. They can create systems to connect individuals or small groups of members. This makes a big community small.

By joining, members show their interest or that they share an identity. Connecting these individuals can lead to good conversation, potential friends, and collaborations. It also creates a deeper connection with the community by making the members real. People generalize the community on the members they know. They think “the person I met was thoughtful and smart, so the rest of the members are also this way.” This creates a better overall perception of the community. Examples of tools that facilitate these face-to-face connections include Luma, Gatheround (fka Icebreaker), randomcoffee, Zoom, and many more.

Voice rooms and small, non-public events are other ways to facilitate these connections. I’ve seen communities do “cafes” or “work sessions” where members work together in a Discord, Zoom, or Clubhouse room. I’ve seen spontaneous listen or watch parties. Not every event needs to be a big public spectacle. Internal events to build connections should happen more.

All of this works well because it makes the community very small for a short period. This allows people to make deeper connections without the broader context interfering. When you are interacting with one person, you don’t need to think about how your actions fit into the broader community. It provides focus, something many communities lack when they get too large.


Dunbar’s number exists and internet communities must handle it. Most handle it passively, barely realizing the changes it causes. This may work for some, but many will have to actively handle Dunbar’s number or see a loss in quality.

As a member, you should be wary about joining a community that doesn’t handle Dunbar’s number well. Their stated goal might be much different from their actual goal. They may state they want to help you accomplish a task, make connections in a specific industry, or create valuable conversations. They may have succeeded prior to reaching Dunbar’s number, but as they grow they fail to handle the challenges arising from it. Know your goals when joining. Look for communities with the structure, size, and connections you want. The platform, habits, and artifacts of the community are the best clues to figuring this out.

As more attention is paid to internet communities, we will see new ways of handling Dunbar’s number. This helps people find and benefit from communities matching their goals. It creates healthier communities and, overall, a better online environment.

Thanks to Kate Canniff, Abe Winter, Casey Rosengren, and Nick Drage from Foster for the feedback.

Let me know what you think on Twitter.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s