Eric Schmidt, former chief executive of Google, was the first chairman of the Defense Innovation Board. During his tenure the board’s recommendations have led to impactful results such as the creation of the Pentagon’s Joint Artificial Intelligence Center, the shift to an emphasis on agile software development within the Defense Department, and adoption of ethical AI principles for the military.
One of the most promising outcomes of the Defense Innovation Board (DIB) was the recommendation to eliminate the “color of money” for software development. The Software Acquisition and Practices report provides opportunities to improve software development, procurement, deployment, and maintenance. It highlighted that colors of money doom software projects.¹ Because software is in continuous development it is never “done”, and recommended Congress create pathways to smooth the funds process for long-term programs.
The most impactful recommendation on the DevSecOps environment was the creation of a new RDT&E Budget Activity BA-8 for “Software and Digital Technology Pilot Programs.” Included in PB21 was $577 million for RDT&E BA 8 with nine line item approved pilot projects. Some examples include a “Global Command and Control System” from DARPA, the Navy’s “Maritime Tactical Command and Control” program, the Air Force’s “Joint Special Operations Center Mission System”, and “Defensive Cyber” prototypes for the Army.²
Initial results look very promising and Congress has signaled that it’s ready to expand DoD’s use of “colorless” money for software. This year’s Defense spending bill includes permission for 12 pilots in 2022, up from nine the previous fiscal year. This approach is a “game changer,” said Gen Arnold Bunch, the commander of Air Force Materiel Command, during his address at the Air Force Association annual conference in National Harbor, Maryland. You no longer have sustainment, development and procurement money, particularly in the agile, secure DevOps environment the DoD is moving to.
So where does the DoD go from here? Well first the DoD must build on the trust that has been established so far with Congress. The willingness to add new pilots appears to indicate they’ve seen no significant problems so far. But Bunch said “the trust lawmakers have shown could be easy to lose if program managers aren’t careful.”³ To really enable a DevSecOps culture across the DoD leaders need to adopt radical transparency and be willing to step forward when there is a problem and not try and hide it from the Hill.
With that trust established the natural evolution is to move beyond line-item approved pilots in the budget. Allocating funds by portfolio, such as PEO Digital in the Air Force or PEO Manpower, Logistics & Business Solutions in the Navy, would enable leaders to take money that is there and use it where they need to, without the need for burdensome reprogramming actions. Not having to tie-up lawyers for six months to decide if they can use procurement of O&M funding for their requirements also helps accelerate the culture and thinking required to deliver truly agile DevSecOps.
Success in proliferating trust and use cases for more “colorless money” could fundamentally change how the budget affects broader innovation within the software ecosystem.
Enterprise tools like the Air Force’s Cloud One and Platform One fundamentally service multiple programs. Such efforts are difficult to get funded because they represent enabling technologies rather than program outputs. Dr William Roper, former Assistant Secretary of the Air Force for Acquisition, Technology and Logistics said “Airplanes look awesome. Satellites look cool. And they are made in people’s districts and flown in people’s states and employ people. Digital transformation… is harder to fit into the budget process.”4 Breaking out of a program acquisition methodology of executing a plan devised two years earlier will result in more efforts successfully navigating the “valley of death” for technology transition.
Broader expansion of “colorless money” authorities will also enable the DoD to better keep its’ “Eye on the Prize”. The nature of software development may radically change in the near future. Today DoD investments are focused on AI & ML, autonomy, quantum computing, cybersecurity, networks and complex systems. Today, computers are controlled by programs that are comprise of sets of instructions and rules written by human programmers. AI and ML change how humans teach computers. Instead of providing computers with programmed instructions, humans will train or supervise the learning algorithms being executed on the computer. Training is inherently different than programming. Data becomes more important than code. Hacking AI is very different than hacking code. The use of synthetic environments and “digital twins” many also become increasingly important tools to train a computer. The impact of AI and ML on software development will be profound and necessitates entirely new approaches and methods of developing software.5
We welcome the new budget activity and authorities that Congress has authorized. It’s exciting to see that expansion of those capabilities are being considered. However, for our nation to keep its competitive advantage we must do more. Evaluating the success of individual pilot programs over the course of years continues to erode that advantage. We are instituting policies that attempt to keep up with today’s pace of software development. These policies will be mis-aligned in a few years time with the rate of change, especially in the era of AI & ML. It’s time to be bold… instead of adding more pilot programs we urge the PPBE Reform Committee to recommend and Congress to grant authorities down to a portfolio level across all services within the DoD. Today’s best practices in agile software engineering make those old hardware lifecycle phases irrelevant. Instead, tie money to purpose with agility in how resources are applied, and without artificial barriers.
Let’s choose to be bold.