One of the latest buzzwords to agitate the Web is "community." In fact, most Web sites have less sense of community than a New York City subway car: at least people are going in the same direction on the subway. On the Web, users have very different goals, they come from all over the world, and they don't know each other.

Chat vs. Discussion Groups

Internet chat rooms are a perfect demonstration of lack of community: no serious discussions ever take place. The only application for which chat is suited is flirting: admittedly a strong human need (and responsible for countless hours of AOL use), but chat should be banished from any Web sites that do not host dating services.

The prototypical Internet chat goes along the lines "The Mac is great" - "No, Bill Gates is great" - "No, Bill is evil" - "No, you are just envious" - etc. etc.

Chat is ephemeral and scrolls by in real time, meaning that the rare posting with intellectual content will be long gone by the time a new user joins. Sure, chat can be archived, but scrolling through thousands of lines of vacuous banter is even worse than experiencing it real time. Chat is like sushi: it only works when fresh.

Discussion groups are better than chat because they are persistent and tend to encourage users to look over their contributions before posting. Also, the longer postings typically lead people to include some arguments and not just pure name-calling. Even so, most postings are fairly uninteresting. AnchorDesk has one of the few good uses of discussion groups I have seen: First, they make every article into the seed for a discussion group, meaning that discussions are integrated with the main content rather than being a distinct area for ramblings. Second, the editor selects a small number of the more interesting postings and links to them directly at the bottom of the article. This gives added prominence to the best postings and allows readers to focus their time on relevant contributions and skip flame wars or trite repetitions of weak arguments.

Guidelines for discussion groups include:

  • Allow every major page on your site to spawn an associated discussion group: you never know in advance what users will find interesting and when they will have comments to add to your site
  • Prune old postings: either by deleting irrelevant ones or by editorial promotion of the best ones
  • If you ever feel tempted to include a chat room on your site, try a discussion group first

Chat and discussion groups are both forms of user-generated content (UGC) and are better analyzed in terms of this contents' value for other users than as community-building. True, a few services like The Well have seen a genuine sense of community among its users, but such exceptional cases cannot form a model for more average sites where users will not know each other.

Participation Inequality

A major reason why user-contributed content rarely turns into a true community is that all aspects of Internet use are characterized by severe participation inequality (a term I have from Will Hill of AT&T Laboratories). A few users contribute the overwhelming majority of the content, while most users either post very rarely or not at all. Unfortunately, those people who have nothing better to do than post on the Internet all day long are rarely the ones who have the most insights. In other words, it is inherent in the nature of the Internet that any unedited stream of user-contributed content will be dominated by uninteresting material.

The key problem is the unedited nature of most user-contributed content. Any useful postings drown in the mass of "me too" and flame wars. The obvious solution is to introduce editing, filtering, or other ways of prioritizing user-contributed content. One idea is to pick a few of the best reader comments and make them prominent by posting them directly on the primary page, while other reader comments languish on a secondary page. It is also possible to promote the most interesting postings based on a vote by other readers who could click "good stuff" or "bozo" buttons.

Mega-Collaboration

The Web is not a collaborative environment in the traditional sense of a small group of people who like each other and have shared goals. On the contrary, most Web users will never meet, they have very different backgrounds and interests, and they sometimes have conflicting goals. For example, search engines suffer from authors who spike pages with "search bait" in ways that reduce the usefulness of the Web for all other users.

The model for the Web cannot be a cozy village of helpful friends. A big city of strangers is a much better model. Well-planned cities can be pleasant environments as long as there are cops to catch the bad guys and traffic lights to direct the flow of cars. In building Internet services, the city metaphor leads to the directive to rely on mega-collaboration rather than direct collaboration.

Mega-collaboration is the idea that the collective behavior of millions of people can form a constructive environment where value is derived from the mass of actions even though each individual action is done purely for the sake of the individual user. For example, a large ISP could measure what Web pages are accessed the most and use this data to pre-fetch a fresh copy whenever a user is on a page with a link to one of the popular pages. Even better, the ISP could build a probabilistic model of what links are most likely to be followed from any given page and be even better at pre-fetching pages. Once such a service has been shown to work, it could migrate to the user interface and be used to color the hypertext links depending on their popularity. In the future, ISPs may compete on value-added services derived from knowledge of the preferred behavior of their membership base.

Mega-collaboration can extend beyond pure frequency metrics. Explicit representations of quality have to become a key element in future Web user interfaces since that will be the only way for users to manage the expected flood of information in a few years. Human judgment is the only way to measure quality, so ISPs or independent quality services could add value by collecting information about the sites that thrill or disappoint users. The quality ratings shown to any individual user must be derived from other users who agree with that user and not from an average vote of all users (otherwise people could spam the ratings).

The Net.Gain Book

The book Net Gain by John Hagel and Arthur Armstrong has received much coverage in the trade press as a leading proponent of online communities. I hesitate to criticize this book since it is virtually the only one on the market to take a strategic approach to Web design. I do have to point out, however, that the authors are stretching the definition of community far beyond any reasonable practical perspective. Anything that has to do with communication between customers and sites is immediately denoted a "community," even though it would be more fruitfully analyzed as one-to-one marketing.

I wish the publisher would issue a second edition of Net Gain by simply letting a copy editor go through the manuscript and remove all occurrences of the word "community." The authors have much to say about the need to create value for Web users by taking advantage of the user-driven nature of the medium. They also understand how companies need to rethink the way they do business to prosper in the network economy. It's just a shame that the many useful messages in this book are stamped with a misleading label. I understand the zeal for promoting a book with a simplified message; flying the "community" flag has certainly gotten the authors lots of press. I simply hope that readers will not be as easily fooled but will take away the strategic messages about user-driven Web business; just forget the buzzword. Businesses will lose lots of money if they make chat rooms the focus of their sites instead of building useful features for their customers.