DARPA’s New Approach to Situational Awareness

Making sense of multiple flows of information is a continuing challenge in conflict early warning and early response; particularly vis-a-vis decision making. How can we make overall sense of conflict data originating from different sources? DARPA’s new approach is to turn warzone data into simple stories.

From Wired:

Drone feeds, informant tips, news reports, captured phone calls — sometimes, a battlefield commander gets so much information, it’s hard to make sense of it all. So the Pentagon’s far out research arm, Darpa, is looking to distill all that data into “a form that is more suitable for human consumption.” Namely, a story.

Making sense of a complex situation is like understanding a story; one must construct, impose and extract an interpretation. This interpretation weaves a commonly understood narrative into the information in a way that captures the basic interactions of characters and the dynamics of their motivations while filling in details not explicitly mentioned in the input stream. It uses story lines with which we all have experience as analogies, and it simplifies the detail in order to communicate the crucial aspects of a situation. The story lines it uses are those the decision maker should be reminded of, because they are similar to the current situation based upon what the decision maker is trying to do.

These stories, however, would be authored by artificial intelligence (AI) algorithms courtesy of Darpa’s Information Processing Technology Office. I’m sceptical of purely AI-driven solutions for obvious reasons. What caught my interest, rather, was the idea of story telling, i.e., a qualitative, narrative approach to conflict analysis and situational awareness that may overcome some of the cognitive biases that surface during decision-making processes.

Conflict Early Warning: Any Successes?

Does any one know of any successful instances of conflict early warning and early response? Specifically, in which situations have formal conflict early warning systems provided an accurate forecast of conflict escalation which was communicated to policy/decision makers who subsequently took action?

The Dallaire cables clearly reveal that even in the most prominent case of accurate and timely conflict early warning (which was the result of human intelligence and not formal indicator-based warning systems), response was muted. Kofi Annan, at the time head of Peacekeeping, told Dallaire not to take any action but to inform the president of Rwanda about the stockpiling of weapons even though it was the president’s own inner circle that was planning the impending slaughter. Can graphing conflict trends over time elicit a more effective response?

In order for conventional early warning systems to operationalize response, they would first need to build a paper trail of analysis, which would then need to be used to lobby the UN Secretariat and other member states; these actors would then have to place political and economic pressure on offending governments and/or non-state armed groups, and the latter would have to acquiesce. Are there cases that show this has been effective in the past?

The case of Macedonia is often heralded as a successful demonstration of preventive deployment and conflict prevention. However, the operational response was not the result of indicator-based models. Furthermore, US geopolitical interests are said to have played an important role in galvanizing response.

Furthermore, as Susanna Campbell and I have argued here the decision-making process at the UN and other regional organizations is far more complex than we realize. By starting with an analysis of the existing decision-making structures and working “backwards” to formal conflict early warning systems, Campbell and I found could not find evidence that conflict early warning analysis serve as input in the decision-making process. Does this mean that the most we can expect from conflict early warning systems is analysis for the purposes of lobbying and advocacy? In other words, should conventional conflict early warning systems be thought of as tools for lobbying policy makers? Should this qualify as operational response?

That may just be part of the problem. The indicator of success for effective conflict early warning systems still seems to be timely high-quality analytical reports. Can we do better? Why not use operational response (even if the response is not successful in preventing violence) as a more appropriate indicator of success?

Any thoughts?