One of our favorite topics is the Pentagon's Defense Advanced Research Projects Agency (DARPA). The Agency seems to be dedicated to spooky projects with the potential to destroy humanity.
Of course, that's never the stated goal.
The latest initiative that caught our attention is MAGICS, announced by DARPA this year. It stands for Methodological Advancements for Generalizable Insights into Complex Systems. It sounds impossibly complicated, but the idea is pretty simple. Apparently, DARPA is not satisfied with its ability to predict human behavior on a large scale. And MAGICS is a research project dedicated to come up with a better way.
The "complex system" they are studying is us. All of us. How we collectively act, given all the inputs and motivations and changing environmental signals that drive us one way or the other.
You'd think with the abundance of digital data and machine learning, measuring human behavior might be getting easier. There's more data to measure. But more data actually makes it harder.
The first issue is, no one knows how much each input might affect things. How influential is social media vs advertising vs advice from friends vs a news story vs the fact it's hot outside -- the list of inputs goes on and on. What underlying psychological or social phenomena drive society? Which are more important than others? We just don't know.
Even if you could assemble the entire list of influences on society, you run into a bigger problem: the list is always changing. The things driving group behavior never settle into a set pattern. It's like trying to map a spill as it's spreading across a table. It's just not a set object you can map out. Whatever drives collective behavior will change before you can analyze it.
DARPA wants to fix this. MAGICS would attempt to combine all psychological theories and psychosocial motivations into a working, predictive theory. They want to take in all data points and reliably turn them into accurate predictions. They don't just want to understand us, they want to know what we're going to do next, before we do it.
What this model is, or how it will work is to be determined. The MAGICS announcement is actually a solicitation for a proposal. They're looking for the genius who can find the new way. What DARPA does know is this research will not be done in a simulated environment.
They are going to run their models on us.
The thing about current statistical predictions (aside from the fact they're wrong) is they use data from stuff that's already happened. But the DARPA MAGICS project looks to "overcome limitations of current statistical methods". So a core goal of MAGICS is to enhance reliability by mandating the use of "open world systems". They don't want truth from a lab, they want "ground truth". Any final reports from the project are required to include "a thorough presentation of validation results" using real people in the world.
On one hand, this makes sense - there's no better test of a theory than on live people. But it also has the potential for danger, since, well, these are live people.
So what exactly are we talking about? DARPA is vague on this. They give examples of the "large scale systems" they want tested. These include how regional economies adapt to new conditions, or how populations shift in response to demographic changes.
It's one thing to come up with a "paradigm-shifting approach" to predicting collective human behavior. But how do you prove your predictions are correct in the real world, unless you somehow cause the changes to happen?
Is DARPA looking to (for example) cause demographic changes on a large scale, so their predictive models can be real-world tested? Would this involve forcibly removing certain groups, by ethnicity or age or income, to verify MAGICS predicted reactions?
And what if it works? If DARPA has a 100% accurate predictor of group behavior? Are we entering a world where uprisings are dismantled before they happen? Where armed drones appear before protests even gather? Where a lone voice who stands out in the crowd is deemed a threat, simply because he acts against the model?
This would all seem like crazy science fiction, except DARPA has done this before.
In 2011, DARPA launched the Narrative Networks (N2) program, to understand how stories influence human behavior and decision making. And like with MAGICS, they went deep into all aspects of the process -- analyzing neurobiological and cognitive mechanisms, developing computational models that could predict how stories shaped our behavior. The ultimate goal was to intervene in these behaviors. DARPA wanted to influence society and manage our perceptions of reality. Stories would become a tool in psychological warfare.
So we shouldn't be surprised if MAGICS goes the same direction. Any findings on human behavior DARPA collects will likely be weaponized.
The MAGICS solicitation was still open as of this writing. Applications to be the researcher who leads the project are due by June 30. If you get the gig, just make sure DARPA doesn't lead us minions off a cliff so the elite can have more space.