Keeping a Civil Tongue: Web Lab's Plan to Extinguish Flame Wars

      Bookmark and Share

Imagine an online discussion about one of the most divisive issues in recent years, the impeachment of Bill Clinton, in which participants from different sides of the issue complimented one another about the quality of their arguments. Imagine one participant calling Hilary Clinton's "own agenda" "kinda scary," then earnestly asking others in the group, "Or is it? Any opinions?" Then, imagine other user posts that resembled op-ed pieces, polished (even grammatically correct) ruminations on the issue or even high-concept news parodies and humorous send-ups of mass media coverage.

Could this have been the same fire-breathing impeachment debate the rest of us lived through? The same wall-to-wall, intransigent exchanges among Republican and Democratic Party apparatchiks that passed for TV news programming during those nightmarish months? Indeed it was—at the aptly-named Realty Check, one of several experiments in conducting more civilized discussion groups in a medium infamous for its conversational "rants," "flame wars," and "drive-bys."

While the rest of us tended to yell at each other online throughout much of the late 90s, the New York Internet think tank Web Lab was evolving a better way at sites like PBS Interactive's P.O.V. and American Love Stories, Reality Check, and recently at Its Small Group Dialogue (SGD) tool is both a general method of structuring more civil online discussions and a software toolset that content providers now can license from Web Lab. As a means for raising the level of online discourse, it is already an unqualified success, enjoying rave reviews from participants and from And for the ever important bottom line, SGD communities tend to be self-regulating and thus (potentially) cheaper to maintain than monitored bulletin boards. Web Lab founder Marc Weiss proudly boasts that, after several years of experimentation across multiple sites and tens of thousands of threads, the quality of discussion has been so high and so civilized that "We have never had to delete a message for content reasons, and we've never done an internal edit on a message."

Accountability Breeds Respect
According to Web Labs, SGD works because it identifies and addresses the root causes of online nastiness: lack of focus, anonymity, and the absence of real community. SGD posters are not drive-by blowhards, but self-selected members of a group whose purpose is to discuss issues and understand diverse viewpoints. They sign up in advance at a site for focused exchanges on specific topics (impeachment, September 11 aftermath, interracial relationships, et. al.) that are scheduled to take place for two to four weeks. The pool is divided into groups of 50 to 70 people. Members can retain their formal anonymity with any username they choose, but each is required to write a short biographical statement that usually contains their age, race, religion, and often how their life experience informs their opinions.

According to Weiss and Jed Miller, the director of collaboration and community at Web Lab, this basic up-front structure goes a long way to separate online discussion from the roots of animosity. By breaking discussion areas down into limited groups that have even the most glancing familiarity, "It gives them a sense of accountability and intimacy with one another," says Miller. An online ranter generally has nothing invested in the discussion group that he harasses, no concern about who is listening, and even less about what they think.

On the other hand, SGD members are asked to introduce themselves when assigned to a group, and they then get an email with about 20 sample bios of other members. Early in the experiments with this format, Weiss recalls, "People got to know each other on a level of who they were before they started debating the issues, so there was a little bit more sense of responsibility." Members can only post to their own groups, a rule the designers feel helps to reinforce a participant's investment in the success of the ongoing discussion, but also helps isolate any brush fires that do erupt. Even in this more structured environment, the first smoldering of a good old-fashioned flame war can occur, but the group dynamic almost always extinguishes flare-ups quickly. "You'll always have a couple of soap boxers and potential demagogues," says Miller. "But when people are forced to look at what they just said by looking at other people's response to it, and are invited to declare what and who they are early on, the piñata effect is reduced."

One of the more important lessons to emerge from Web Lab's experience is just how much power a publisher can have over users' online behavior simply by establishing the right context. When signing up for SGD, users opt into a set of expectations for civil, respectful exchanges and for exposure to diverse views. In part, the self-selection process guarantees a more serious group, but the system itself encourages members to monitor the general quality of discussion. "What's really fascinating about this is that once you put the responsibility for the success of the group into their own hands, they rise to the occasion," says Weiss. Another key component is modeling. Participants can nominate posts or threads to be featured for all members of all groups to see on the board's home page. By highlighting superior discourse, the publisher collaborates with members to model and encourage good behavior and raise the bar of expectations. "People are as responsible communicators as you invite them to be," Miller has learned.

MSNBC: Beyond the Mosh Pit
This is all well and good in the otherwise genteel environs of PBS sites and Web Lab's other up-brow venues. But things are likely to be a bit different when you tackle the more mass, raucous rabble of MSNBC. com, which boasts 15 million to 20 million unique users a month. Even the site's executive producer for communities Joan Connell admits that prior to the Web Lab experiment at the site her well-populated message boards too often became a mosh pit. Despite a costly staff of human monitors, "We had a lot of trouble with hate speech and off-topic conversations," she says. That and budget cuts forced MSNBC. com to shut down its bulletin boards in early December, just after Marc Weiss approached the site about hosting a SGD on the September 11 attacks.

While Web Lab was hoping to stress test its SGD software on's massive traffic flow, the discussions only attracted a modest 1,800 participants (20 groups) over two months beginning in December 2001. "While the volumes of conversations we had were not high, the quality of the interactions were excellent, and from a community standpoint this was very important," says Connell. In fact, in the first three-week session, all ten groups exercised an option within SGD's software to extend the discussion beyond their allotted time.

About half of those who signed up actively participated in the MSNBC. com discussions, a slightly higher percentage than in previous SGD forums. "This crowd was used to a more rough and tumble environment for discussion," Miller found, but ultimately these groups followed the same patterns as earlier trials. "The investment in self-expression as opposed to pure attack was high, an eagerness not just to shout opinions, but to relate experiences."

Free-Standing, Not Free-Wheeling
Both Web Lab and learned that discussion groups like these require very active promotion and linking throughout a site, even one as well-trafficked as this one, in order to attract high participation. Getting placement on that all-important front page was critical to capturing eyeballs, but even with strong advocates among the editorial staff, SGD had trouble competing for featured links against a run of war news.

The experiment enjoyed underwriting from Web Lab's funding sources, and the site has not yet decided to make SGD permanent. Web Lab is talking with numerous media companies about licensing the product, arguing that this software produces both more civilized and cheaper online conversation.

The process is remarkably automated; it is a modified version of Web Crossing's bulletin board programming. Once a preset minimum of users signs up for a discussion, the software uses their basic profile information to assign groups, each of which reflects the demographics of the entire pool. It then emails the invitation to introduce themselves and a second email with sample bios and perhaps a starter question. Users themselves are able to modify somewhat how they see messages, and every post links to a pop up of the writer's bio. The software even polls the members at the end of the discussion about whether they want to extend it. While many groups do elect to continue, it is important that discussion sessions are finite. This seems to encourage involvement because members are willing to commit to a burst of serious conversation if an endpoint is in sight.

In the test, Web Lab had budgeted one monitor for every 30 groups, because the need for human intervention is minimal and the system lets members quickly notify the site about any tech problems or misbehavior in a group. The only editorial task involves selecting and pruning the posts that members nominate to be featured on the main page. A very comprehensive administrative tool gives monitors a detailed view of activity levels within the groups. As more sites drop their message bases because of budget or concerns over propriety, Web Lab is hoping to address both concerns with SGD.

The Value of Community
In a budget crunch, the bottom line value of a site's online community is among the most difficult things to quantify, and often it is one of the first features to go. Content providers do this at their own peril, Connell warns, because it ignores a key differentiator of this new medium. "In old media, it was a one-way street. I write it. You read it. End of story. In new media, we present the news and we provide a platform for our readers to express themselves." Without interactivity, the Web is just a radically uncomfortable way to read a newspaper or watch TV.

Miller thinks that online community has been woefully undervalued in the first and second waves of online publishing. "The missed opportunity on a lot of Web sites…was that droves of people were showing up simultaneously at the same location with a tremendous motivation to connect, with a latent instinct that needed a context." These communicators are among a site's most loyal return users and its most frequent online purchasers, he argues.

But even if an SGD member never bought a single book or pullover from a site's affiliated ecommerce partner, she still provides a site with a valuable commodity—content. Many sites have tried to capture and repurpose the free but unstructured stream of user-generated material, but generally it involves costly editorial time. Average posts within the SGD forums are inordinately polished and engaging, something closer to an informed panel discussion than the usual graffiti wall of online commentary.

Whether online discussions really can prove their worth on the bottom line, Connell feels that the SGD method is most laudable because it revisits some of Web culture's early idealism, how the medium might expand and enrich civic discourse. "Free speech is messy. Human interactions are often chaotic and unpredictable. But this kind of focused forum is like a well-maintained public park where people can feel safe about interacting with others and perhaps even learn something about themselves," she says.

For all of its democratic appeal and high ideals, however, one of the most interesting lessons from the SGD experiments is that even this more civilized online discourse is not about changing one another's minds. Post-discussion surveys find that few participants shifted their views on an issue as a result of taking part in the session. What did change was their level of understanding and tolerance of opposing views. In an online world of "drive by" exchanges, this is no mean feat.

Imagine if there were an online flame war and nobody came.