What’s Next for Defense Software Development?

In February 2018, a Defense Science Board Task Force released a final report on the “Design and Acquisition of Software for Defense Systems” that contains recommendations which will shape defense software development and acquisition in the years to come. Vendors working in this space should be aware of them.

Report Rationale

The DSB created its review Task Force to achieve the following objectives:

1. Examine the current state of Department of Defense (DOD) software acquisition and recommend actions for the Department of Defense and its suppliers;
2. Consider the development, test, and evaluation of learning systems;
3. Contrast and compare DOD and commercial software development and determine what commercial software development capabilities military systems should embrace;
4. Identify impediments in DOD requirements, contracting, and program management and how they might be removed;
5. Determine if “Agile” software techniques are being used effectively and identify impediments;
6. Determine if the commercial concept of a minimum viable product should be adopted by the DOD;
7. Determine best management approaches to achieve rapid and effective software upgrades, including an analysis of modular, open architecture;
8. Look at lessons learned from recent software challenges (e.g., OCX, F-35); and

Provide recommendations to ensure rapid adoption of cognitive capabilities as they mature.

Recommendations

To meet these objectives, the DSB Task Force made seven recommendations how to improve software acquisition for defense systems. These are for the DOD to:

1. Implement a “Software Factory” approach and incorporate it as a key evaluation criterion in the source selection process.
2. Adopt continuous iterative development best practices.
3. Adopt risk reduction best practices for program acquisition strategies.
4. Adopt strategies for software sustainment that are better incorporated in current and legacy programs from development, to production, and sustainment.
5. Improve the software and acquisition workforce.
6. Frame contracts for long-term software sustainment.
7. Implement an independent verification and validation (IV&V) process for machine learning and autonomy in software systems.

Potential Industry Impact

If adopted, the recommendations of the DSB Task Force could play an important role in shaping the defense software market. The demand for a Software Factory approach could in particular determine future industry winners and losers. The factory demanded would basically comprise each vendor’s application of agile/iterative processes to the development of software for the DOD. Under the DSB recommendation, no software factory process, no contract award.

The adoption of agile/iterative software development techniques informs several other recommendations as well, potentially affecting post-award work. Vendors that have not already adopted agile techniques for defense work should consider doing so immediately in order to remain competitive and to retain work they have already won. Not doing so could be the difference between retaining contracts or losing recompetes.

Industry partners currently have a period of time to adopt iterative software development processes because, as the DSB report points out, the DOD itself lacks the workforce – both on the program and acquisition sides – that is familiar with agile development. Prime contractors are likely to be called upon to provide this expertise.

Concerning machine learning, the DSB recognizes the DOD’s lack of “research and experimentation programs around the practical use of machine learning in defense systems” despite the growing viability of the technology. The DSB recommends, therefore, that the DOD establish “efficient testing, IV&V, and cybersecurity resiliency” methods to ensure that machine learning technology is properly incorporated into the defense software ecosystem. Programs along these lines are likely to appear under the Under Secretary of Defense for Research and Engineering, at the Defense Advanced Research Projects Agency (DARPA), the Software Engineering Institute Federally Funded Research and Development Center (FFRDC), and at DOD laboratories. Contractor support will undoubtedly be required.