June 26, 2010 / by emptywheel

 

DARPA Seeking CompLit Experts

Oh boy, I’ve got a lot to say about this:

After nine years of conflict, the U.S. military is still having trouble finding common ground with warzone locals. One way to fill that breach, Darpa figures, is through “interactive stories.”

Counterinsurgency, in many ways, is a series of negotiations — over economic development, over security, over political power. And “negotiation,” Darpa explains in a new request for information, “is best served by a culturally-specific narrative that explains why we hold a position, how it relates to other parties, and how it affects all parties both positively and negatively.”

[snip]

To come up with its storytelling tools, Darpa’s Information Processing Technologies Office is hoping to go beyond its usual cadre of neuroscientists, artificial intelligence specialists, and gadgeteers.  The agency also wants contributions from “art, literature, film, dance, games development, advertising and public relations, advertising, grass roots organizing, collective decision making or any other discipline for which the respondent can make an argument that the approach bears on this task.”

Click through for the bit about Wizard of Oz.

First, a story. As I was finishing my PhD–having done research on what amounted to interactive narratives–I was hanging around MI’s School of Information some. I saw a presentation from some IBM dude who worked at their Yorktown Heights research center. He was describing how IBM’s consulting wing had started working with their clients to use narrative to better manage corporate culture. It quickly became clear to me that they hadn’t ever considered one of the basic problems of the literary sphere–irony (or, more generally, language that was not transparent, that meant something different than its plain language meaning). This IBM dude was describing, for example, how significant they thought it was that one story-collection they had done had mentioned one of their clients’ VPs a large number of times. I asked, “were they being sincere? because some of those comments sound like they were being ironic?” Because IBM was assuming language was transparent (as it often is when you’re talking to computers), they had misread that what they thought was instilling strong corporate culture but was instead a corporate-funded way to trash their own VP.

I went to ask IBM dude about this after he finished. As it turned out, he had worked with my father when he worked at Yorktown Heights; my dad was a big hero to this IBM dude (I believe he meant his compliments sincerely). And IBM dude was thrilled when I told him I was a Watson scholar; IBM had paid a big chunk of my college tuition and employed me every summer in college (I also learned to swim at the IBM pool, that’s how thoroughly I was immersed in IBM culture). So for shits and giggles, I asked whether IBM would be able to hire someone like me–someone with expertise in the subtleties of human narrative. And he sort of soured and said, “oh, no, we’re really only able to hire social scientists.”

Lucky for me, I had gotten my fill of working at IBM every summer in college.

I took two things away from this exchange.

First, the exchange reinforced my strongly held belief that our society devalues the humanities to its significant cost. As you recall, before I went to grad school, I had done a lot of documentation consulting which put me in situations (like oil refineries and credit ratings agencies) where our inability to use language in a sophisticated manner might lead to massive disaster like explosions or metaphorical Wall Street explosions. We were charging companies pretty big money to do the kind of thing DARPA now wants done. And while we offered both capacity to crank out pallet-sized documents and some particularized expertise, the underlying problem was that we, this country, doesn’t really treat humanities with the same seriousness we treat, say, math. (Not that we treat math with the seriousness it deserves either). Largely, but not entirely, because of actions of the humanities academy, the humanities have either become the feeding ground for law, or a frivolous soft pursuit usually reserved for those who could look forward to a career without big worries about paying off college loans (admittedly, thanks to IBM, I was one of those people). As a result, when business (or the military) decides they need humanities-related skills, there is both a general inability to frame the problem and a shortage of people who can apply their humanities training to practical problems.

Which leads me to the inability to frame the problem. IBM couldn’t envision that speech is–often–not sincere, and so couldn’t develop their research to account for such a probability (and, as a result, some company was paying big money to empower its employees to ironically snark their VP).

Which leads me to DARPA’s description of what they’re trying to do:

The Defense Advanced Research Projects Agency (DARPA) Information Processing Techniques Office (IPTO) is requesting information on areas of research into approaches to cross-cultural negotiation in the human terrain through narrative. The object of this research is one or more culturally-aware computer-based negotiation tools for use by parties with minimal common language skills.

Through negotiation we want others to understand our position, we want to understand theirs, and we want to come to a mutual accommodation of both. Yet understanding and accommodation are gained neither by imposing a choice nor by simply describing it. Instead negotiation is best served by a culturally-specific narrative that explains why we hold a position, how it relates to other parties, and how it affects all parties both positively and negatively. Further, negotiation is not simply achieving one’s goals but rather it is aligning all of the participant’s goals with each other so as to agree upon a set of shared goals.1 Across cultures and languages, this task is more than difficult because of possibly opposing but unknown goals, groups’ unacknowledged positions, and the unspoken stories that brought about the situation. DARPA is interested in exploring innovative approaches, which can help people cross a cultural divide, that are based upon interactive stories, the implications of decisions, and the choices and outcomes associated with alternative behaviors.

[snip]

The question of interest in such cases is what approaches and, if implemented, what tools might support human negotiators in navigating such issues, especially in non-verbal or minimum-language approaches.

[snip]

DARPA/IPTO welcomes white paper contributions from the fields of cognitive science, cultural anthropology, artificial intelligence, art, literature, film, dance, games development, advertising and public relations, advertising, grass roots organizing, collective decision making or any other discipline for which the respondent can make an argument that the approach bears on this task. Approaches must focus on the narrative aspects of negotiation and the construction or reuse of
stories. Technical areas for consideration include, but are not limited to: (1) case-based reasoning; (2) computational HSCB modeling in human terrain; (3) distributed social networks for mixed-initiative story construction and presentation; (4) psychological theories of narrative understanding and storytelling, and; (5) knowledge-based game rendering. We encourage very non-traditional approaches to this problem (e.g., a virtual tribal storyteller interacts with a human mime to produce a silent theater improvisation with audience participation). [my emphasis]

Now, I actually think DARPA’s onto something here about narrative. And given that DARPA is about research, not turnkey programs, I’m not bugged that they’re trying this.

But note the underlying assumption: “parties with minimal common language skills.”

DARPA is proposing building an entire narrative-driven negotiation system–presumably intended in the short term to help it convince Afghans to side with Americans over the Taliban. But it is assuming that we won’t go about learning Pashto before we build this narrative-driven system.

You see, the whole project seems to be an overly optimistic workaround to deal with this problem:

The Defense Department still “lacks a comprehensive strategic plan for addressing its language skills and regional proficiency capabilities,” John H. Pendleton, who wrote the GAO report, told a congressional panel.

[snip]

Gen. Stanley A. McChrystal, commander of U.S. and NATO troops in Afghanistan, decreed last year that each platoon have at least one soldier who could speak basic Dari, one of the country’s two national languages.

But the troops’ training for the unusually difficult language, which is largely spoken only by Afghanistan’s professional classes (including police), not rural farmers and shopkeepers, lasts only two weeks, soldiers say.

“I doubt you can get much from two weeks,” a Special Operations Forces veteran commented on the Captains Journal military blog last month. “It takes about one year of intense training to speak a language“ and commanders can’t spare ground troops that long.

“If 2 weeks isn’t nearly enough,” added a blog read who indicated he or she was with the 1st Brigade, 101st Airborne Infantry, “then 1st BDE 101st isn’t on the chart.”

“We’ve been issued pocket references with various vocabulary and phrases broken down phonetically,” the soldier added. “There was a half-hearted attempt at a ‘30 Key Phrases’ program, but it was never enforced.”

Want to pacify a country with neither the native speakers to speak to the local language or the willingness to learn it? Build videogames to try to communicate without language!

We neither have the competence nor are we investing in getting the competence we need to carry out our COIN project. For better or worse, we have not done what the country did during the Cold War, which is establish a bunch of Area Studies centers to gain deep competency in the culture and social science of the areas we were seeking to influence and fund people to go learn these difficult languages. Not to mention, we’re kicking out those in the military who do speak these languages.

Instead, we’re asking kids that probably didn’t join the military because of their linguistic skills to learn Dari in two-week courses and we’re trying to invent some way to successfully establish common goals through the use of computerized, mimed narrative.

This, btw, is the ultimate root of my deep skepticism about COIN. We are a country that is institutionally disinterested in learning about other cultures. One of our biggest exports (in addition to arms) is narratives that deliberately flatten culture. We don’t put much stock in listening.

So no matter which COIN genius General is in charge, no matter how much money and time we throw at the problem, we’re still going to be fighting weaknesses in our own culture and education system.

And inventing nifty videogames along the way.

Copyright © 2010 emptywheel. All rights reserved.
Originally Posted @ https://www.emptywheel.net/2010/06/26/darpa-seeking-complit-experts/