Recent events have focused our collective attention on the internet's rich bouquet of dickitry. I am not myself a blogger, but I do have some expertise in the realm of online behavior: I recently completed a Master's program in what I like to call "network psychology." Over the last three years, I've reaped a bountiful harvest of solid, scientific insights into the ways that technology and psychology interact to shape the online world.

In this essay, I'd like to share some research on antisocial behavior. Education is its own reward, I know, but I'm also hoping that this information will help guide social software designers as they work to promote civility in the blogosphere and beyond. So, on to the pearls of wisdom...

Real Names Are Not Magic

Anonymity does promote antisocial behavior, it's true, but only because it shields abusers from the consequences of their actions. Requiring commenters to provide their real identities, or just a valid email address, won't curtail abuse. That crap only works on Rumpelstiltskin.

Simply put, eliminating anonymity is not the same as creating accountability. Study after study has shown that accountability is the key. Bernard Guerin (1999) argues, from high atop a mountain of evidence, that accountability is the on/off switch for both antisocial behavior (threatening language, favoritism) and social loafing (slacking off when working in groups). Research on in-group bias confirms these findings (Dobbs & Crano 2001). Accountability plus group membership actually leads to better behavior than when people act alone.

The trick is that accountability comes from consequences, and consequences are not the internet's forte.

Mods are Mediocre

Most social software relies on the old censorship stand-bys: deleting posts and banning users. These are toothless solutions for many reasons, but the one most pertinent to bloggers is a catch-22 between bias and inefficiency. Lone moderators are as likely to act on personal biases as on community standards. Even when censorship is even-handed, abusers can console themselves with the thought that the mod is "just out to get me."

The usual solution is to amalgamate the opinions of multiple moderators. That's a great way to filter out personal bias, but it also slows things to a crawl. The more opinions you gather, the longer it takes for censorship to actually happen. In the meantime, discussion proceeds as if there were no moderation at all. One survey of Slashdot traffic showed that most discussions are half over before any censorship takes effect (Lampe & Resnick 2004). You might as well let anarchy reign.

The thing is, ostracism can be a powerful punishment, even when delivered through a computer screen. Kipling Williams (2002) created a three-player virtual ball toss game that gradually excludes its human player. (The other two players are bots.) After being ostracized by "people" they didn't even know, in an exceedingly trivial context, study participants reported feelings of rejection every bit as strong as when they'd been ignored face-to-face. Anyone who's ever had a forum post go unnoticed knows the feeling.

The challenge for software designers is to find a way to let community members ostracize abusers without the inefficiencies associated with moderator systems.

Invest in Identities

As long as people can create and discard new identities at will, true accountability will remain out of reach. Trying to validate people's real names, however, is both unwise and unnecessary. You don't have to know who "MadHater99" really is in order to punish (or reward) them as long as the identity itself has value.

My prescription for blog comments is to place new posters on probation: filter their posts until such a time as the blog owner can review and approve them. Note that I said "filter," not "hold." Look no further than Digg's comment system for an excellent example of how filtering can minimize unwanted content. It's ostracism without censorship.

After a user has made a few decent contributions, they go off probation and their posts are displayed upon submission. At this point, peer review should take over. As in Digg's system, users with established identities should be able to rate each others posts; if a user's recent average drops below a certain threshold, they go back on probation.

In this way, identities gain value over time and as the community at large positively evaluates them. Abusers are steadily filtered out and switching to a new identity just puts you on probation. A centralized service could add even more value by tracking users across multiple blogs, collecting peer ratings from their every post, and then sharing that data with any blogs they post to in the future.

In Summation...

• Accountability only flows from meaningful consequences.
• Human moderation is caught in a catch-22 between bias an inefficiency.
• Stable, valuable identities will allow communities to police themselves.

Thus, we come full circle. When users must invest in their online identities, even pseudonymous ones, it becomes possible to apply real consequences for abusive behavior. Accountability breeds civility.

Sincerely,

-- Daniel Bayn
dan@bayn.org

P.S. For more on filtering and other visualization techniques, see
Shaping Online Behavior through User Interface Design.

P.P.S. If you dig that, check out Memes, Communication Technology, & Cultural Change.

References

Dobbs, M. & Crano, D. (2001). "Outgroup accountability in the minimal group paradigm: Implications for aversive discrimination and social identity theory." Personality & Social Psychology Bulletin, vol 27(3), 355-364.

Guerin, B. (1999). "Social behaviors as determined by different arrangements of social consequences: Social loafing, social facilitation, deindividuation, and a modified social loafing." Psychological Record, vol 49(4), 565-578.

Lampe, C. & Resnick, P. (2004). "Slash(dot) and Burn: Distributed Moderation in a Large Online Conversation Space." Proceedings of ACM Computer Human Interaction Conference, Vienna Austria.

Williams, Kipling D. et al. (2002). "Cyberostracism: Effects of Being Ignored Over the Internet." Journal of Personality and Social Psychology, vol 79, 748-761.