DARPA - Education Dominance

Training and human effectiveness

Program Manager: William D. Casebeer, Ph.D.

Education Dominance set out to demonstrate that education could be improved by an order of magnitude. Instead of just looking at the classroom, the program focused on the end-goal of education—becoming the expert in your field.

We know how to create experts, as Anders Ericsson and Malcolm Gladwell have articulated—a decade of very hard work and a host of fortuitous circumstances. The objective of this DARPA project was to demonstrate that we could create experts in months rather than years, and to get every student to that level of competence and confidence.

A second major objective of this program was to recognize that this level of education would be impossible to scale—the resources required, from exceptional (and thus scarce) resources such as the world’s best teacher to the support costs of one-to-one tutoring where every student had access to, for example, an enterprise scale network would be prohibitive. Thus, the second objective was to demonstrate economic scalability—in this case to demonstrate that technology could replicate an exceptional educational environment.

The third element of the program was to identify a pilot project where the value to exceptional education was nearly unlimited, where the potential ROI to even an expensive school would also be at least an order of magnitude. The Navy presented such an opportunity. Their computer networks are arguably the most challenging in the world—with nearly 300,000 nodes and over 200 networks that are constantly being reorganized and often becoming autonomous. The challenge to the current Fleet ITs is so great that typically at least a problem a week cannot be solved onboard ship, and needs to be escalated off-ship for solution by elite teams.

Over 6,000 tickets are escalated off-ship; many take weeks to solve; several hundred represent over a million dollars each in operational costs.

Education Dominance was chartered to demonstrate an order of magnitude gain in educational outcomes, to replicate these results using technology, and to demonstrate a path to driving the Navy’s trouble tickets to zero (among other operational objectives).

The program was structured in four phases: create a school that demonstrated the factor of ten efficacy gain, run rigorous field trials to validate the gains, replicate the school using technology, and revalidate.

The initial school was run in Monterey, with fifteen entry-level Navy students with no prior IT experience. The school was predominantly one-to-one tutoring with two dozen carefully selected IT experts serving as tutors; all were gifted teachers who had also done something in their career to demonstrate that they were truly exceptional at IT.

The Institute for Defense Analyses (IDA) then conducted five weeks of intense, often grueling competitions between these graduates and Fleet experts, many with over ten years of experience and more than 40 advanced schools among them.

The Education Dominance graduates won all competitions, often by wide margins, demonstrating not only a mastery of their subject but also exceptional problem solving and reasoning skills—applied to the subject domain, exceptional confidence and very deliberate precision. All are characteristics of experts that are rarely engendered in a classroom.

The second phase of Education Dominance focused on porting this school to the Digital Tutor, technology that had been developed in advance of the program. This artificial intelligence platform is designed primarily to replicate the behaviors of exceptional tutors; an immediate indicator of its success is how often students refer to their tutor as “he” rather than “it”.

The Digital Tutor, in phase three, was installed in the Navy’s entry-level school for ITs in Pensacola, and continually tested while the content was under development in steadily increasing increments. Again, IDA ran tests after each cycle, against sailors taking an eLearning course offered by the Navy, the school’s instructors, and a classroom using Microsoft and CISCO curricula. In all trials the graduates using the Digital Tutor dominated, often by margins that were described as “huge”.

Finally, IDA ran two weeks of competitions between the Fleet and Education Dominance/Digital Tutor graduates—encompassing very challenging, real-world trouble shooting problems, building enterprise-caliber networks from scratch, an advanced knowledge test and an oral review board. Again, the Education Dominance students dominated in all competitions against Fleet experts with an average of ten years of experience.

Back to what we do