ATD Links Archive
Issue Map
ATD Links Archive
ATD Links

How Can Social Learning Scale Massively? Lessons from World of Warcraft

If people work in groups separate from each other, how can we mitigate the silo effect and make sure we capture all of the learning on a collective basis? The answer may be in massively multiplayer online role playing games.

Much fuss is made of class-size effects in schools, but I often get blank stares when I talk about the dangers of putting 10,000 people together in an online learning environment where we are trying to foster social learning.

Increasingly, the call is going out for some sort form of social learning to be a part of online learning initiatives. And it's quite right to do so. We're well aware of evidence linking social context and long-term learning retention. We know how important social comparison is to our lives. We see the opportunities of the web 2.0 to create both a push and pull of knowledge throughout our organizations. But we also have vast numbers of people to include in these processes if we are to make social learning a full part of our workplaces.

Social learning vs. crowdsourcing

A chief concern about the implementation of social learning within the enterprise is how to scale the benefits of social learning to meet the large numbers of employees that make up its audience. For me, there is a mix-up between the power of social learning and the power of crowdsourcing.

Crowdsourcing suggests that the more people we throw at a problem, the easier that problem is to solve. Social learning is more concerned with the meaningful relationships we build with others and how they help provide the context for learning. Social media is a tool which sits at the confluence of these two ideas; articulate your ideas using social media and they have the power to not only influence your close followers, but also the wider world.

Robin Dunbar theorized Dunbar's number, which contends that there is a limit to the number of other people with whom one can maintain a relationship. The number is said to be 150, give or take a few. Dunbar based his findings not on observations of daily lives, but on an evolutionary perspective to account for the optimal number of relationships an individual should have in order to thrive.

Other studies, like those conducted by McCarty et al, have sought to estimate network size empirically. These methods have yielded higher numbers than Dunbar's; a mean of 291 was found in the McCarty study. However, even these measurements have their flaws. Most notably, the McCarty study relied on people to estimate their own network size.

Facebook is rapidly becoming a better measure in the opinions of many including myself. According to Facebook, 130 would be the average number of 'friend' relationships a person has on the platform. It would be fair to say this number is conservative at the moment; not everyone is on Facebook and many people keep a separation of their friends, family, and co-workers, which means their complete network is not accounted for by a number. However, this number is also likely to be skewed by the number of "nonfriend" friends we tend to have on Facebook; mostly old school acquaintances, who we might like to spy on for social comparison reasons, but wouldn't otherwise count as friends.

Of course, the real answer here is that there is no single number to succinctly articulate how big a social network can be; the number will be slightly different according to our behaviors and situation. But, whatever that number is, it is probably in the low hundreds.

It is important to remember that we already have a number of relationships before we set foot in a social learning environment. Our capacity to make more meaningful relationships is going to be limited by the number of these relationships that already exist. In other words, we probably only have a few slots left open. So when you are faced with a room of 10,000 people, where will you start focusing your effort in order to start building these few new meaningful relationships without wasting your time?

The answer is that you probably won't. Most people don't. Less than 1 in 5000 visitors to Wikipedia actually makes an edit each month. Similarly, when we run smaller classes of 15 or so people on the 3,000 member community platform the social interaction flows readily. But it is impossible to do this on a grander scale while fostering true relationships. Sure, people contribute to large news websites with comments, but that's more about expressing opinions than about building relationships.

Modeling game relationships


So, what can we do to address this issue? Certainly, just adding a social media facility to your learning platform and expecting relationships to flourish isn't going to work. I often say that there is nothing sadder than an empty forum - and I've seen enough of them in the various back alleys of company intranets and LMS's to last a lifetime.

The answer, for me, lies in breaking down the whole population into smaller parts on an autonomous basis. We can model these sorts of relationships on massively multiplayer online role playing games (MMORPGs) such as World of Warcraft (WoW).

WoW has millions of players entering its "world" every day. Players choose a realm to play within when they enter the game. Each realm is an individual copy of the game, perhaps characterized by running in a different language (French realms instead of English, for example). Within each realm is a series of playable areas that take the form of continents; these can be explored autonomously and alone, with players taking on challenges that exist within them as they go.

However, many of the more complex challenges that exist within each realm require a team effort to complete. This is where things get interesting. Small groups of players, banded together autonomously as "guilds" form to take on the bigger challenges.

Of course, you don't want to form a guild to take on a challenge and find some other group already in the dungeon (how often does that happen at work), so each challenge has the ability to provide a unique instance of itself for your group. This instance is a copy of the same challenge others can take, but only your group has access to it. This way many groups can take on the same challenge at the same time.

Each person within the guild needs to be engaged in order to tackle the challenge. There is rarely room for freeloaders as the challenges are often limited in terms of the number of players who can be in the group. Everyone contributes.

Guilds are often fairly tight-knit groups. Some of the more serious ones go on to meet each in real life, and many guild participants would readily accept that some of their relationship slots are occupied by those which they play games online with. In addition, there has emerged a huge community of guilds talking with each other; sometimes on friendly terms, sometimes more competitively. But the ability to showcase skill and discuss tactics with other guilds is one of the biggest drivers of online communities outside of the actual game environment.

Transferring to social learning

This model can help to overcome the scalability issues that social learning often faces. Asking people to make an impact on the world as a whole is difficult, but it's easy to be influential within your group. Hiding in the big wide world is easy, but it is difficult within a smaller group. Making meaningful relationships with everyone in your organization is beyond the realm of possibility, but you can select a few people from which to learn within a smaller group.

In short, the answer lies in breaking down the enormous mass of your workforce into smaller groups, working together to improve both themselves and the organization. The limitation in this approach is in the crowdsourcing approach to problem solving. If people work in groups separate from each other, how can we mitigate the silo effect and make sure we capture all of the learning on a collective basis?

  • I would suggest that silo effects can be countered by simple measures to ensure the groups are diverse in nature; only a certain number of people per department in each group for instance.
  • I wouldn't stop anyone from being members of different groups for different topics, allowing insights to spread virally between groups.
  • I would look to the groups to curate the best content to be pooled into a single, enterprise-wide access area. Instead of trying to aggregate everyone together on every topic, have groups nominate their best insights to be part of the company's best insights and use a voting system within the realm to showcase the very best content.

Bottom line

To be sure, there's a lot more work to be done in this area, and at the moment I'm looking to talk to those who have implemented social learning initiatives within their organizations to research deeper on what the 'ideal number' might be. But for now, let me suggest five lessons from WoW to help your Social Learning initiatives scale massively:

  1. break down online social interactions into smaller realms and instances for groups
  2. ensure everyone in a group needs to contribute in order for the group to succeed
  3. create areas for groups to interact with other groups
  4. don't allow groups to match up identically with organizational structure; instead, diversify
  5. curate the best bits of each group to deliver real insight back to the rest of the organization.

Ben Betts is managing director of HT2, creators of innovative learning technologies. To add your thoughts on this topic, visit Betts' blog at

About the Author
Ben Betts is an entrepreneur, technologist, and social learning expert. He and his team at HT2 develop products and services for online learning. In his role as CEO of HT2, Ben developed the Curatr Social Learning platform and the open source learning record store Learning Locker. He has worked with Google, Accenture, KPMG, Pearson Education, Oxford University, Cambridge University, BP, Twitter, Nike, Barclays, Shell, Xerox, and many more. Ben writes and speaks about social and peer-to-peer learning around the world, and has made an appearance at TEDx. He completed a doctorate at the University of Warwick (UK), where he studied the application of social learning in the workplace. He also holds an MBA and is a Fellow of the Learning & Performance Institute.
Be the first to comment
Sign In to Post a Comment
Sorry! Something went wrong on our end. Please try again later.