A few months ago, a worker monitoring a hotline for the Polaris Project, a nonprofit group dedicated to combating human trafficking, received a text message from an 18-year-old woman in distress.
The woman, a sex-trade worker, was trapped in a motel room with her pimp and she secretly used his cellphone to send a text seeking help. The Washington-based group moved quickly to alert authorities, who ultimately arrested the pimp.
For Polaris’s chief executive Brad Myles, the episode demonstrated how text messaging might offer a new channel to help victims. In the process, Polaris learned those texts are data, and collectively they can be analyzed to identify patterns in human trafficking so the group might better craft policy and awareness programs.
Polaris started its text hotline in March, through a philanthropic partnership with San Francisco-based cloud company Twilio, which powers text and voice customer service communications for clients such as Uber, Hulu, eHarmony and CocaCola. Victims can text “HELP” or “INFO” to the number 233733 (BeFree), where they are forwarded to Polaris’s hotline staff, who then respond from their computers through a messaging service called Chatter. Polaris has operated a voice hotline, at 1-888-373-7888, for a few years.
The text campaign lets a new group of victims connect with Polaris, Myles said. “There’s a population of people who are high-risk individuals, or survivors of trafficking, who would not call the phone number, and they wouldn’t send us an e-mail, and they wouldn’t fill out a Web form, but for whatever reason they would send us a text. Once we get in touch with them, it’s the same types of information we would probably learn from a call.”
Training hotline specialists to use texts to help victims in crisis has been a challenge, because of the kinds of information they contain, Myles said.
“The actual length and structure of the language you’re using is very different — you’re not speaking in full, complete sentences, you’re not able to explain context. It’s a very truncated, reductionist form of communication,” he said.
For instance, specialists have learned to interpret texting shorthand. If they ask if a victim is safe, the victim may respond “Y” or “N” instead of “yes” or “no.”
“We began to need to ask more directed, close-ended questions instead of open-ended questions,” Myles said, asking if someone is safe, for instance, instead of asking them to describe their situation.
Hotline specialists also had to adjust to what Myles describes as “a strobe-light feeling of communication” — texts are often sent sporadically, so the conversation may take longer than a phone call. With texts, “it’s not a continuous stream of discussion,” as specialists might have to wait minutes or even hours for a response, Myles said.
Despite the challenges, the texting campaign has generated large volumes of new data Polaris is trying to analyze. Salesforce, the company behind Chatter, collects data about the phone calls and texts, such as length, frequency and location. Combined with tips from callers — suspicious addresses, vehicles, or names of traffickers, for instance — Polaris has information on almost 200 variables per case.
Analyzing incidents in aggregate could help Polaris identify patterns in human trafficking. For instance, Polaris recently started receiving seemingly unrelated calls and texts throughout the country about illegal labor trafficking and abusive work conditions in carnivals. “It was something that wasn’t really on our radar as much before,” Myles said.
After searching its database, Polaris’s staff identified common recruitment sites and recruiters worldwide who were drawing immigrants into the United States to work at these carnivals. Polaris is developing interventions targeting workers and recruiters in Central America and Africa, where the workers often come from.
In the future, this data could be used to predict where incidents will occur before they do. Polaris has met with computing firms to discuss “becoming proactive and not being so reactive,” Myles said. “We could use [modeling] then to craft certain interventions that we know will target certain types of trafficking, without needing to learn about them from the calls.”