(by Nick Chiosie) On 28 June, the Center for Technology an National Security Policy, among them TIDES members, hosted Jason Matheny from the Intelligence Advanced Research Projects Activity (IARPA), a United States research agency under the Director of National Intelligence’s responsibility.
The speaker, Jason Matheny of IARPA spoke about two projects that IARPA is currently funding. The projects are focusing on social media and open source information that will enhance a decision maker’s situational awareness in near real-time. This will hopefully create an environment that will enable commanders and local authorities use their limited resources more effectively during a crisis.
One project is called Open Source Information (OSI) which centers on automated analysis of publicly available data to anticipate and/or detect societal events; civil unrest, economic crisis, disease outbreaks, etc. By analyzing all publically available data and cross examining specific trends it is hoped that certain patterns can be detected within open source outlets. If the specific event is dangerous to the local population they can be warned before it occurs and encouraged to take precautions. By automating the analysis the programmers can tweak the system to add/lower value to specific sources of information over time. Alerts are sent to an analyst and then the analyst can send out relevant and accurate information to decision makers in a speedily manner.
For example, if large groups of people start using Google maps to find the closest hospital, while others are self-diagnosing themselves on medical websites, while there is a sharp decline in traffic in specific areas (less people driving to work because they are sick) accompanied with a high volume of Tweets about symptoms, and retail sales of thermometers increase over a short period, it is hoped that a prediction can be ascertained as to “X” sickness is going to hit “Y” city within “Z” days. It is hoped that by predicting an event, local authorities will have more time to deal with and prepare for a crisis.
The other project is called Aggregative Contingent Estimation (ACE) which adds a human perspective when attempting to create conditional forecasts of events transpiring. Simply put, if correlating events transpire, what is the probability that a crisis will occur. The project focuses on the informed judgments of over 10,000 people to reduce random error. Some people are weighted more or less depending on varying factors. The results have been mixed but so far, but the project was able to glean some insightful material; Accuracy is explained less by what people know and more by how people think, prediction markets and interactive pools increase accuracy of probability, domain expertise is not highly predictive of forecasting accuracy. It is the hope that this project will lead to increased situational awareness of commanders and local authorities so they can use specific information to mitigate damage and avert crisis in the future.