Scale effects in enterprise social software

When people think about social software, they think big. Thomas Vander Wal’s slide on tagging is a good example. At the bottom of the scale is personal use, and the largest scale is shown as a “mature system.” (see slide 25 below).

Organizations come in various sizes and shapes, from small businesses and workgroups to very large enterprises. Even large enterprises with tens of thousands of employees are tiny compared to the scale of public social software services such as Delicious (5 million users a year ago), and SlideShare (a million users in December)

The tools and properties of social software act differently at different sizes and scales. Bigger can be better, worse, or just different.


At a large scale, with very large numbers of users and assets, tagging reveals the knowledge of crowds. A tag cloud in a large, active community reveals clusters of interests and enables discovery as a byproduct of many tiny acts of personal organization.

At a small scale, tagging doesn’t have these same effects. There isn’t enough distribution to discover interesting things about the topic space, and there isn’t enough density to make it a really efficient browsing tool on its own (although it does help with search).

On the other hand, smaller-scale tagging is useful. At a small scale, with good processes, tagging can be used to create workflows and update feeds that are useful and also adaptible.

Ranking and Rating

At the level of Amazon and NetFlix, ratings from a large base of participants quickly reveal the hits and the dogs. Rating is a quick action that lets casual participants contribute quickly and get something back in the form of recommendations to match their tastes. In a good-sized knowledgebase, rating is valuable to indicate which content is service its purpose and what needs improvement.

At a smaller scale, ratings are a very different tool than at larger scale. Ratings can be an excellent decision support tool to assess the popularity and priority of ideas in a finite period of time. Ideas are generated and fleshed out. Then, ranking and voting are used to prioritize the ideas.

There are also ways that ratings can be counterproductive at a small scale. In a discussion community where there are typically a handful of contributors or commenters per topic, having one or two ratings on each post is a meaningless waste of space (see an example of this anti-pattern on the Personal Democracy Forum, an otherwise excellent blog.)In communities where participation is high, asking for “ranking” – a low-engagement activity – can serve the purpose of reducing contribution. If you see something that you have a chance to make better, why make it easier to point out that it is broken than to go ahead and fix it?


Participation is an area where smaller communities have very different dynamics than large ones. In consumer communities, there are typically a very small number of very active contributors. The vast majority of people are consumers, and only a small percent do any amount of contributing. In a large community, that small number results in vast creativity, but it’s still a small percent of the whole. By contrast, healthy collaborative intranets have much higher active participation, often in the low tens to 50 percent or more.


Large social systems have significant challenges with social misbehavior. Large crowds where people don’t know each other bring out anti-social behavior (trolling, spam, just plain idiocy). So large communities need to implement explicit reputation systems to reduce the noise caused by anti-social participants.

In smaller communities, where people use their real names and have real-world accountability for their actions, misbehavior is exceedingly rare. Facilitation is helpful to foster productive interaction and help a group head in a common direction, but explicit reputation isn’t needed. More implicit reputation revealed by contributions and participation can be interesting and relevant.


There are two social software principles that are at odds with each other:

  • social software gains value as more people use it.
  • social software is that which can be spammed

On the one hand, large scale networks can yield valuable insights, where large numbers of people tag, rate, and link information and each other. On the other hand smaller groups have higher participation, collaboration, and civility.

Administrators and champions who foster social networks should think about the scale of their community, and use tools and techniques appropriate for the scale.

Leave a Reply

Your email address will not be published. Required fields are marked *