Between the 5th and 4th centuries, many Greek tragedies employed the use of a mechanical crane, known as a “mechane,” to lift an actor into the air as a way of dramatically inserting a character into a scene. The character, often a deity making their first appearance in the third act, would use their omnipotence to resolve the play’s central conflict and bring about a hasty resolution.

The use of the mechane became so commonplace in Greek theatre that the Latin term, “deus ex machina” — which translates to “god from the machine” — was used to describe the unexpected last minute godly intervention in the tragedies.

The term “deus ex machine” has endured in popular culture, as modern-day storytellers mine the rapid advancements in technology — the mechane of the 21st century — and its relationship with the human mind for their art.

This relationship between human and machine is central to the future of the National Geospatial-Intelligence Agency.

In a keynote address at the 2015 GEOINT Symposium last spring, NGA Director Robert Cardillo said that geospatial intelligence, or GEOINT, is at an “inflection point.”

“In the next five years ... more than a dozen constellations, hundreds of [small satellites] will launch and continuously scan the earth,” Cardillo said. “It means our analysis of world events is going to be holistic and persistent.”

According to Cardillo, this proliferation of sensors forces intelligence agencies like NGA to pivot.

“We have to change our mindset, to investigate multiple possibilities and better understand this complex situation. We won’t need to balance a finite collection capability against a seemingly infinite set of GEOINT requirements. And I’m betting on our people — those brains — because their expertise is our ultimate value proposition.”

Holistic and persistent GEOINT analysis has been at the forefront of Jennifer Daniel’s brain for the better part of 18 months.

“How do we create coherence from chaos?” asked Daniel, the director of NGA’s Persistent GEOINT Office, a month after Cardillo’s speech.

“How do we take all this stuff that is out there now and that’s going to be out there in the future, maximize its capabilities, understand what it tells us, and connect the dots? How do we get answers faster so we can provide that information advantage to our customers who can then use it to deliver consequence and succeed in their mission?”

These are questions that Daniel and her 30-person team have been working on since her office stood up in May 2014 at the request of former director Letitia Long.

“This is not about transforming one part of our enterprise, it is not transforming one part of NGA, or one part of the NSG [National System for Geospatial Intelligence], it is about transforming the environment of how we get GEOINT and how we convey GEOINT,” said Daniel.

The first part of that transformation is clarifying, once and for all, what persistent GEOINT means.

“One understanding, one vision,” said Daniel.

So what is persistent GEOINT? According to Daniel, the best way to answer that question is to first dispel a common misconception.

“Some people have the impression that persistent GEOINT as being about a ‘dwell’ capability, like what we get from full-motion video ... where you can dwell over an area on the ground for a period of time and track activity and look at activity,” she said. “That is one part of persistence, but it’s not the only part of persistent GEOINT.”

The official definition of persistent GEOINT, as developed by Daniel and her team, is: “a focused intelligence strategy to obtain geospatial content with sufficient periodicity and duration to detect change, characterize activity, infer behavior, and discover unknowns — where the rate of information refresh equals or exceeds the rate of change.”

Simply, according to the team, persistent GEOINT is not about sensors or automation or processing or models, it’s about the synthesis of all of these things to accelerate data to answers.

According to Sherry Prewitt, defining persistence can be difficult because the term is so malleable.

“Persistence is based on your target or your issue,” said Prewitt, who runs the Frontiers division of NGA’s Office of Geospatial Intelligence Management and is tasked with looking holistically across the GEOINT missions to understand and address future analytic needs.

Daniel and her team are interested in maximizing the capability of any and all sources of information, from federal, commercial and international satellites, to airborne, terrestrial, subsurface and open source. And once you have access to all that data, how do you sort through it, organize it and tell the analyst what is important?

Enter automation.

“Machines assisting humans,” said Daniel. “It’s not machines taking the place of humans. Let machines do what they’re good at and let humans do what they are good at.”

And when we say “machine,” we are focusing on the software and coding, rather than the hardware, said Prewitt.

“Hardware is easy,” she said. “Software and code is much more difficult. There are thousands and thousands of lines of code that need to be written.”

According to Daniel, automation works in a number of ways, from connecting the data, organizing objects, tipping and cueing, and predicting outcomes.

“It’s four times faster for a human to get to the answer by using the assistance from the machine, than it is by having a number of people do that same function without a machine by searching through the images,” said Daniel. “We have examples of automation connecting the dots and discovering things that humans looking at that same set of data were not able to discover. Getting to persistence is taking these types of capabilities and building them, making them more accurate and stronger, and continuing to build them across the enterprise.”

According to Prewitt, automation is key for NGA to fully leverage the amount of information — imagery and data — that is being ingested.

“For us to fully leverage the capabilities that you would get from small satellite companies, for us to fully leverage all of our overhead capabilities, from a timeliness perspective, we have to have automation,” she said. “I’m talking about being able to process all of that information in a quick manner, being able to react to that processed information from a machine perspective. That doesn’t mean that you have to have eyes on every bit of information.”

For now, the persistent GEOINT environment of the future remains, well, in the future.

Daniel and her team, and NGA as a whole, are currently laying the foundation for that environment, which includes focusing on building up the agency’s initiatives in GEOINT Services, Next-Gen Collection, and advanced analytic techniques, and reaching out to industry, academia and government partners.

“All of these are building that foundation that we can leverage to get into the future,” said Daniel. “We have a number of prototypes at NGA as well as in the community that are working on using machines to detect activity, to detect changes, to pull data together. Many of these are in the prototype environment, many of these are at the very beginning stages. We need to continue to push forward and build on them to get to where we need to be in the future.”

Pushing an entire intelligence discipline into the future is not easy work. Nor is it instantaneous. There is no “deus ex machina” in the third act.

“Getting to a persistent GEOINT environment is not a big bang,” Daniel said. “You are not going to get there overnight and turn a switch and all of a sudden you are going to be there. We have to purposefully take what we are doing now and build to the future.”