Far From the Mad Crowd
By Lisa Hammitt
“Mission Accomplished.” So read the post on fifteen-year old Phoebe Prince’s Facebook page, written by one of her tormentors hours after the fifteen-year-old hanged herself by a scarf in the stairway of her family’s apartment.
This happened in early 2010. Phoebe had moved from Ireland to South Hadley High School in Massachusetts, where she was immediately set upon by a clique of girls who were apparently threatened by her charm, good looks, and popularity. They harassed her in every way imaginable: they followed her from class to class, called her vile names, threw things at her, physically threatened her via texts, and posted hateful comments on MySpace, Craigslist, Formspring, and Facebook. “Go kill yourself” was one of their final taunts.
Five students pled guilty to criminal harassment, but served no time. Phoebe’s parents settled with the school, which had been aware of the harassment but didn’t appear to take it as seriously as they should, for $225,000.
As the mother of two boys — at the time, ages four and one — I had already formulated a rough plan on how to limit their exposure to the many unsavory corners of the internet.
The story horrified me and I had to do something about it. I had seen plenty of bullying growing up in the Bay Area, but it was confined mainly to school and school-related activities. As awful as it was, at least a bullied person could find some refuge at home.
Researching this more, I discovered that 80% of kids report being bullied online. That there is very little unfriending or blocking of bullies on social media. And that bullying of kids is usually a continuation of what’s happening at school (while adults get bullied online, they usually don’t get harassed at work). In other words, today’s teenage bullying is 24x7, pervasive, and mean beyond belief. Taunts and rumors go viral in minutes. Instigators sling their mud from the safe distance of a tablet, and may further insulate themselves with an anonymous identity.
Just as the mother in me was appalled, the technologist in me wanted to fight back, so I started to assemble an arsenal of potential counter measures:
- Homology Matching is a technique borrowed from computational genomics, where it discovers biomarkers in proteins and maps them to phenotypes. Added to Natural Language Processing, it could help distinguish teasing from harassment and hate speech from sarcasm.
- Collaborative Filtering could personalize posts and push positive content to community members.
- Social, Interest and Knowledge Graphs could help identify bullies and their victims before things spiral out of control.
- Constraint-based logic, when combined with machine learning, could help automatically enforce codes of conduct.
- Gamification can score good and bad content, and reward positive contributions to the community.
Could we get enough data to train the algorithms? No problem. Even a forum with 20 posts per day and 7 comments per post would be sufficient. In short, we have a variety of technology tools to bring to bear, many of which have been time tested for 30+ years.
But as much as I’d like to try out these tools, I’ve come to realize that they’re inadequate on their own.
To understand why, just look at Phoebe Prince’s school in Massachusetts. Although school administrators and counselors didn’t know the entire scope of what was happening to Phoebe, she had sought help from them, others had reported the incidents, and at least one student was briefly suspended for her actions. So it seems as if the school should have been able to mount a more effective response to the toxic social environment swirling around the new, pretty freshman from Ireland.
It turns out that civil culture and determined human intervention are the critical factors to making the online world safer, both of which appear to have been lacking at South Hadley High School.
Nobody knows this better than David Spinks, the CEO of CMX Media and the leader of a 7,000-member community called CMX Hub. David has tirelessly raised awareness about how to limit online harassment. He notes that while a respectful culture is a good broad-brush way to reduce the potential for harassment, “even the best culture will quickly turn into Lord of the Flies” if community managers are lax.
He backs up his observations with example after example, ranging from 7 Cups of Tea to the World Economic Forum. The WEF’s TopLink Community has perfectly implemented David’s playbook. Thousands of hours of moderator training, online resources from Harvard and curated communities have achieved 99% adoption among their world leaders. Pre-Davos debate is civil, voting is anonymous and knowledge sharing is a central tenet, while flaming, trolling, sockpuppeting, and meatpuppeting are strictly controlled by moderators packing some slick AI. The success of Toplink has led the Forum to extend conversations to young leaders and they enjoy the same adoption rates with a strong code of conduct.
As I go to SXSW to discuss this topic as a panelist at both John Greenblatt’s and Amy Poehler’s SmartGirls’ sessions, I have never been prouder to be an IBMer. We have the best cognitive capabilities in the world and they can help to combat online harassment. Our core philosophy, though, is to use it to help humans do better. As David’s very willing partner, it is an honor to be that ambassador for future generations.
