Winning Principles for Defense Transportation Digital Modernization Strategies
The Defense Transportation Enterprise—military and industrial components—is in a race to capitalize on the data and knowledge revolution, which is the normative and technical follow-on to the digital revolution of the nineties. Not everyone in the defense transportation community has the same perspective as to how to score relative gains against competitors in this race. To that end, this article draws on my year as the first-ever Air Mobility Command (AMC) Chief Data Officer. It also attempts to persuade the reader that winning principles are human-capital-intensive, inculcate a shift from automation to ideation, cause a normative shift around integration semantics, push science out of programs and into pure services, decentralize analytics, and combine with an increasingly ubiquitous corporate framework to empower the Commander.
First, observers around the enterprise notice that as Machine Learning (ML) and Artificial Intelligence (AI) applications mature, the associated workforce tends to grow. Whereas some believe that computer programming and data science ought to allow one’s community to do more with fewer employees, which would be a sign of capitalizing on an increasing degree of automation, the opposite is occurring. One finds oneself in need of more and more people to do data science work. Since the emphasis is on augmenting human efforts with ML/AI (not replacing human operators with those capabilities) and making systems more capable by making them more complex, one requires an increasing number of human workers to achieve the imagined effects.
Some of those imagined effects are the applications of things like deep learning, neural networks, natural language processing, or computer vision to an existing group of systems that hold archived records. To implement that vision, one must link previously unlinked systems and add applications and layers of code to existing infrastructure. Leaders understand that by doing those things—linking previously unlinked systems and adding things to them—one increases the complexity of what one depends on. A leader who understands that achieving his or her vision for digital modernization will necessitate an increase in system complexity will reject or qualify support from partners who advertise that they will simplify the system. The leader is really setting out to magnify effectiveness by making systems less complicated, but more complex.
In addition to magnifying effectiveness while growing in complexity, key stakeholders consider yet another seemingly disingenuous but valid paradigm shift. Boomers on the event horizon of the digital revolution wanted to automate workflow processes; however, later generations on the edge of the data and knowledge revolution want to augment creativity and ideation, which is an inherently complex interaction of human and machine competencies.
In addition to magnifying effectiveness while growing in complexity, key stakeholders consider yet another seemingly disingenuous but valid paradigm shift. Boomers on the event horizon of the digital revolution wanted to automate workflow processes; however, later generations on the edge of the data and knowledge revolution want to augment creativity and ideation, which is an inherently complex interaction of human and machine competencies. Human beings can imagine, and machines cannot. Machines can reason faster than humans—apply the scientific method to human ideations and model the ramifications of those ideations—years faster than humans. Those data science professionals who can create the architecture necessary to make this pairing accessible to the average person are purple unicorns, and many in the defense transportation community are latent to integrate them and slow to capitalize on their unique knowledge, skills, and abilities.
Abundantly clear from one’s inertia in capitalizing on data science technologies is that one is not primarily limited in achieving desired end states by technology unavailability. Indeed, there are many advanced and diverse data and knowledge technologies in the world, each one an outgrowth of one imagining what one can do with statistics and computer programming by applying those skills to a natural condition. Strategic leaders remain wanting, however, for a cadre of operational and tactical leaders who consistently imagine and reimagine how to provide responsibility and accountability over these emerging technologies’ applications to military problems.
Responding to that impoverishment, strategic decision-makers in the Defense Transportation Enterprise use the term integration in different ways than before. Whereas the defense industrial complex used the word in the early 2000s primarily to describe the intertwining of people and processes, in the last two years the term has been increasingly used to imply the stitching together of disparate subsystems so that the data contained in each becomes part of a larger, more comprehensive system. What used to be called integration is increasingly referred to as convergence, implying an interests-based approach to cooperation. Integration increasingly means subsystem linkage and accessibility, aimed again at augmenting the human effort, and it also implies the responsibility and accountability necessary to operationalize that integration.
Adding on to the newly technical bend of integration, federal data cataloguing is ending some data-gentrified companies’ preeminence and instead rewarding a new cavalcade of data scientists, statisticians, and computer programmers. By making authoritative source cataloguing a by-law, the federal government has allowed the Department of Defense to improve data understanding beyond previous requirements for contract services aimed at the same competencies. Armed with a federal data catalogue and moderately mature data understanding, however, the defense side of the complex remains in relative paucity of data preparation skills. Companies that advertised helping one get ahold of one’s data are making the transition to helping one prepare one’s data. To the extent that a company can make the move, it preserves its relative market share.
Companies that advertised helping one get ahold of one’s data are making the transition to helping one prepare one’s data. To the extent that a company can make the move, it preserves its relative market share.
Even this lucrative window of opportunity in data preparation is ephemeral. As initiatives like computer programming language incentive pay and data analytics as a military core competency combine with the rapid commoditization of data preparation web applications, data professionals are increasingly chased to higher ground. Since the dev-sec-ops pipeline used to increasingly commoditize data analytics applications will also increasingly speed and commoditize the dev-sec-ops process itself, the view of acquisitions as systems- and programs-centric is dematerializing. Using a process to develop applications of AI in order to speed up and streamline processes, including those used to field better AI, is a wicked cycle of increasingly expeditious innovation that ultimately leads to the dissolution of many things exogenous to the process’s essence.
The process itself is a science or scientific methodology, increasingly an instance of computational social science. Since AI in a dev-sec-ops pipeline eventually outmodes science as a system or program, one is left with Science as a Service (ScaaS). Since the Defense Transportation Enterprise has a strong institutional memory of evolving Software as a Service (SaaS), Enterprise Information Technology as a Service (EITaaS), and Data as a Service (DaaS), the community is poised to adapt faster to ScaaS than its predecessors. The Commander’s ability to contract for data science as a service outside of a program allows for what Dr. Brian Keegan calls “decentralized execution” of the data science application to mission.[1] Whereas data governance frameworks allow for centralized control of processes, the centralizing programmatic paradigm is cut from execution, and the Commander reaches out to industry on an as-needed basis for scientific services.
This lowering of the level of execution of data science projects and partnering with industry dovetails with an increasing amount of training in basic data analytics across a growing number of military specializations. Dr. Chris Provan describes this as an informal emergence of analytic solutions advisors at lower levels in the chain of command, springing in tandem with a trend in the commercial sector to “decentralize analytics.”[2]
Further on from ScaaS, the service components’ grappling with the costs and benefits of layering a corporate framework over the top of a board and council process is coming to a close. Whereas the J1, J2, J3, etc., would meet in council to inform the Commander’s decisions; or in industry, the Chief Information Officer (CIO), Chief Data Officer (CDO), Chief Analytics Officer (CAO), Chief Financial Officer (CFO), Chief Technology Officer (CTO), and Chief Knowledge Officer (CKO) would meet in council to inform leadership … on military staffs it is increasingly both. Instead of failing in a haphazardly competitive knowledge and information confluence, the two structures supporting data-informed decisions aqueduct as one structure. Having been adapted for defense knowledge management, the corporate framework is increasingly supplementary and integral to the Commander’s battle rhythm. This integration of a specially adopted corporate framework on military staffs has doubled the rigor of data-informed decisions without doubling the cost of decision making.
In summary, the winning strategy for both sides of the military industrial complex is to first aggressively validate, fund, and hire additional personnel who provide ideation-centric applications of data science to mid-level problems. Second, focus high-level strategy documents and mid-level implementation plans on an orientation toward dev-sec-ops highways. Third, embrace complexity, both in systems and in governance. Complexity does not necessarily imply complication or difficulty, and humans remain near infinitely more complex than the machines or governance structures they ordinarily feel compelled to simplify.
By Maj John Biszko, USAF
[1] Dr. Brian Keegan, (Data Scientist, MicroStrategy), Interview by John P. Biszko, 26 May 20.
[2] Dr. Chris Provan (Data Scientist, Mosaic), Interview by John P. Biszko, 28 May 20.