On April 1, I wrote a post on Hatebase, a new initiative from the Sentinel Project which aims to monitor hate speech as a way to detect warning signs for ethnic violence. Gwyneth Sutherlin, a doctoral candidate at the University of Bradford currently conducting field work in Uganda responded, noting the difficulties in developing applications to mine data from languages that organize information in different ways. The Sentinel Project's Timothy Quinn sent in this response to Sutherlin's post:
A few weeks ago, a Foreign Policy reader responded to an article about Hatebase, one of The Sentinel Project's recent crisis monitoring technologies, and raised several interesting points about the challenge of applying broad technology solutions to complex linguistic and societal problems. We'd like to take the opportunity to respond to some of Gwyneth Sutherlin's very valid concerns.
We would be remiss if we didn't first of all debunk the notion that any single piece of software can solve a problem as complex as geonocide. This certainly isn't a claim we make about Hatebase, which is intended to augment the existing human- and machine-driven research efforts of NGOs such as ours; in fact, it's hard to imagine any reasonable organization making such a casual boast. To logically conclude the contrary, however -- that crowdsourced speech analysis and tracking "will never gain... relevance within conflict areas" -- would seem to err as broadly in the opposite direction.
Ms. Sutherland takes issue with organizations which avail themselves of increasingly robust open mapping technologies such as Google Maps and OpenStreetMap. In northern Uganda and South Sudan, she points out, "there is no word for map" in the Acholi language. The implication, however, that aid organizations should constrain their activities to a local frame of reference seems unnecessarily restrictive. (Should an emergency room physician not run a toxicology screen on a patient who has no concept of blood chemistry? Must one understand pitch and meter to benefit from music?)
In fact, open mapping technologies such as Ushahidi have been quantifiably successful in complementing the efforts of government agencies and aid organizations. In the wake of Haiti's 2010 earthquake, a US Marine recon platoon used crowdmapping to deploy forces to a remote village where "we are now in the process of medevacing two local nationals who would not have received medical treatment in time for life or limb had we not found them." Former Secretary of State Hillary Clinton has been similarly effusive about the potential for real-time mapping, noting that "the technology community has set up interactive maps to help us identify needs and target resources. And on Monday, a seven-year-old girl and two women were pulled from the rubble of a collapsed supermarket by an American search-and-rescue team after they sent a text message calling for help."
We share Ms. Sutherlin's endorsement of orality as an unexplored opportunity for ICT, and agree that non-written interaction represents an "enormous potential to re-imagine the visual interface, to respond to cognitive cues and communication norms from the cultures crowding into the digital space." We believe this is an exciting area for exploration, and have incorporated just such ideas for alternate user interfaces into our product development roadmap.
Unfortunately, software rarely emerges as a complete, shrinkwrapped solution to a clearly defined problem. Iterative feedback from both advocates and critics, including the thoughtful comments of Ms. Sutherland, are what encourage good software to become better. Although we see significant potential for future enhancements of the Hatebase platform, we're proud of how far we've already come: Hatebase is a free, stable, data-rich platform which integrates with external crisis monitoring systems through an open API, and has attracted users and data from every corner of the world.
Of course, Hatebase is only one of the many technologies and methodologies we bring to bear on the problem of preventing genocide. We also engage locally at a participant level with affected communities, and we track mainstream and citizen media using our ThreatWiki platform. As mentioned above, we don't believe that any one approach can be a comprehensive solution to mitigating societal conflict -- rather, we feel that a combination of several integrated tools and processes offer the highest likelihood of identifying and mitigating the potential for genocide.
The Sentinel Project for Genocide Prevention
As a reminder, I'm very open to running reader responses to any of the ideas discussed on this blog. Feel free to write to me at Joshua.Keating [at] foreignpolicy.com.