HLT Applied to Information Operations


Early in the Internet era, national leaders had a sense that Information Operations (InfoOps) would be obsolete in a connected world. With the truth only a click away, how hard could it be, for anyone with a smart device, to see through even the most lurid propaganda? Things turned out otherwise, of course, and the US now finds itself overmatched by influence operations from rival and hostile nations. At the least, a shadow was cast on the validity of the 2016 elections, and foreign influence operations crossed legal red lines delineating “acts of war”. At the most, an investment of less than $100 million may have flipped electoral results. This creates a challenge to the viability of democracy itself.

We’ve previously explored means of measuring effectiveness of InfoOps. We will further develop this theme in the context of the Department of Justice indictment of the Russian Internet Research Agency, which provides a wealth of facts on their funding, tradecraft, targeting, as well as online and physical operations with the US. We will look at applicability of the entire armamentarium of human language technology in fighting InfoOps and its necessary role in coordinated defensive and offensive operations. From the most basic functions of language identification, to the most advanced discourse analysis, we require a familiar hybrid of human and machine. As with other intelligence operations, machines excel at scalability and can provide triage for the top of the funnel, consuming as much input as the entire human race can generate. Human analysts provide the next tier, often (though not always) showing higher accuracy at language tasks. The human and machine results combine as analytics for decision makers, who will remain human for the foreseeable future.