In 1988, as part of a project that later came to be known as PANDA (an acronym representing the Protocol for the Assessment of Nonviolent Action), Dr. Doug Bond began to lead an effort at Harvard University’s Center for International Affairs to systematically assess the incidence of nonviolent struggle – also known as people power – throughout the world. Throughout the 1990s project members developed conceptual tools to advance their research, anchored around the PANDA protocol with its data lens sensitive to the contentious and coercive but not yet violent struggle in an effort to generate conflict early warnings on the likely escalation of volatile situations into violence.
The purpose of this research project was to determine under what conditions contemporary nonviolent struggle anywhere in the world had been successful in effecting social, political, or economic change, or in resisting tyranny. To the extent that nonviolent struggle was found, evidence was also sought to determine whether this form of people power was spreading, and if so, was it viable.
It is significant that this project’s systematic analyses of nonviolent struggle began well before the largely nonviolent revolutions that spread throughout Eastern Europe beginning in 1989. Indeed, the PANDA project was built upon the early, pioneering work of the late Dr. Gene Sharp, a Center Associate from the late 1960s through the early 1990s.
Several lessons became clear to Bond and his research team as they began to assess global news reports of nonviolent struggle. First, nonviolent direct action, no less than violent direct action, was reported in abundance, even by mainstream news media. Second, nonviolent direct action, like its violent counterpart, was variable in its outcomes, with the strategic performance of protagonists, as opposed to structural asymmetry, playing a pivotal role. And third, the tradition of human or hand coding of voluminous electronic news reports posed technical as well as conceptual research challenges.
In 1996 VRA was founded to develop automated tools to process the increasing volumes of global electronic news feed that had become increasingly available. VRA was awarded two patents for its natural language processing technology in 2003 and in 2007. This proprietary technology is used in VRA’s Reader, a tool for automated encoding of events data from news reports.
VRA also led the development of an expanded protocol to guide the Reader. It is called Integrated Data for Events Analysis, or IDEA. The IDEA framework superseded the PANDA protocol, though the core premises about contentiousness and coerciveness from the PANDA project remain central to its operation. In this way the Reader and the IDEA framework work together to illuminate precursors to violence that are required to produce timely early warnings on violence.
VRA’s automated approach to events data development may be characterized as offering a view from above in that reporters for global news publishers typically covered large areas, often from national capitals or large cities. Late in the 1990s VRA was approached by Nongovernmental Organizations working in conflict areas that could not reply on global news to monitor conditions in the dangerous and/or isolated areas in which they worked to delivery humanitarian assistance. These organizations wanted a view from below to be integrated into their situational awareness efforts. In response VRA developed its first-generation of its field Reporter, a tool now in its 4th generation of world-wide deployment.
The Reporter was developed by VRA as a generic, web-based tool, with many of its parameters and settings user-specifiable. This customization feature has spawned a range of customized applications for VRA’s field Reporter, from monitoring staff security to auditing child protection protocols to assessing country risk. In addition, the Reporter’s standard features include both incident and situation reporting, an understanding at a glance dashboard and an analysis module, collectively supporting situational awareness. These tools also support export and report capabilities to integrate the Reporter’s field data with other tools.
One of the early lessons learned from VRA’s news and field monitoring tools development was that context matters. A time series chart of incidents over time or volatility in situation assessments in a given country are not comparable to other countries. Thus, in the mid-2000s VRA began to develop a tool to assess structural attributes of countries to provide a measure of resilience or vulnerability as a backdrop to the dynamic measure produced by news monitoring and field observations.
VRA’s Prospects tool was designed to systematically learn from the past by ingesting historical country profiles, matching them against the outcomes with which they were associated, projecting these structural attributes into the future, and forecasting the likely outcomes going forward. These results are then interpreted in light of the specific drivers that were prominent in the forecasts to support the formulation of options for structural prevention and mitigation. These country profiles in turn serve as backdrops to inform the interpretation of the analysis of the dynamic news and field events analyses. They also allow the dynamic measures to be normalized across countries to support comparisons.
By the mid-2000s also, the rise of social media had begun to provide an alternative data source that revealed popular perceptions of people and entities. These views express sentiment toward others in the unstructured text streams of social media. To tap into this affect, VRA developed a sentiment extraction engine (the VRA Advisor) to support reputational analysis based on text with no requirement that it be structured like a news report. In other words, the Advisor was designed to operate on text-based interactions such as that carried by social media, messages and similar modes of communication.
The VRA Advisor draws upon the same automated natural language processing technology as its Reader, so it can be used to track shifting popular sentiment in near-real time. The Advisor tool is also designed a set of user-specifiable parameters and settings to address the specific needs of customers with uniquely targeted reputation requirements.