Questions in bold provided by Jason Miller. Responses by Joseph Castle.
Discuss your research on open source and what did you find about how agencies are sharing/using open source code?
The study focused on the implementation of the Federal Source Code Policy (FSCP) with whether and how agencies publish open source software (OSS). The premise being if all cabinet-level agencies have a mandate to publish OSS, an IT budget relative to agency mission, and an IT staff capable of working with software, why were some agencies able to more frequently publish OSS than others. This premise supported focus on organizational factors - cultural beliefs, public engagement, structural dimensions, and organizational location - as hindering and aiding OSS publication as one indicator of FSCP policy implementation. The findings expose systemic issues all federal agencies face when implementing new policy and technology.
The research did not focus on sharing and using open source code, however, my experience with Code.gov provides insight into these aspects of federal OSS. There is public data, largely supplied by code platform APIs showing a lot of activity with government code in the form of government repositories, copies, and contributions. There is also anecdotal evidence of code reuse among agencies (e.g., GSA's USWDS, DOE's ZFS, NSA's Walkoff).
The federal government has a long history of sharing code leading to WWII. Grace Hopper and others often attended conferences and shared code for development of computers in support of the war. Currently the federal government consumes a lot of open source software for daily operations for use with servers, data science, and application development. However, the federal government tends to purchase a good portion of COTS programs and customizes leading to duplication and wasted money. Furthermore, very little of the software purchased with taxpayer dollars is available to the public. Software purchasing and development is approximately 7% of the federal IT budget.
Agencies were implementing the FSCP with mixed results as noted on the compliance dashboard on Code.gov and we wanted to explore why some agencies were publishing OSS and others were not. We began with a simple question, "why do some agencies publish OSS and others do not?"
For data collection, we began with a metadata analysis by collecting data pertaining to agency publication of OSS. We found the 24 cabinet-level agencies were producing OSS with mixed results which required a deeper dive into what was happening in agencies.
This resulted in a qualitative study with interviews and artifact collection. The sample consisted of 25 participants from 20 agencies and no participant was from the same unit. Participants were federal employees and either managed or developed software and often resided in a software development or data science office. Participants were also from offices that published with varying frequency - minimally, intermediately, and frequently.
What surprised you about your findings? Did your research show whether the 2016 OMB policy affected more agencies sharing open source code?
Findings were established pertaining to the four main organizational factors.
-
More offices believed they should be publishing OSS even if they are not currently doing so. Participants held both cautionary and advantageous beliefs pertaining to OSS publication. Cautionary beliefs focused on publishing OSS as aligning to scope, change in practice, considering risk, and needing permission. Advantageous beliefs focused on building community, creating efficiencies, demonstrating competencies, and realizing benefits.
-
Units conducting more and varied public engagement more frequently published OSS. These units used bi-directional communication, created forums both in-person and virtually, and integrated teams and products into their work processes.
-
Structural dimensions consisted of decision-making, capabilities, diverse skills, and task coordination. Units, regardless of publishing frequency, all relied on policy. This was an interesting find and different from what literature said is often the case with technology development. All government organizations relied on policy to publish and without policy very few will publish OSS, so policy is key.
-
Organizational location showed that units with relationships with the CIO office and customers often publishing more frequently. Knowing the policies and being offered freedom to innovate allowed many to publish. This was often in coordination with CIO offices allowing for the line office and CIO office to learn at the same time and then development policy and rules to continue to publish OSS.
Summarily, the policy of 2016 or FSCP, does have an impact on agencies. It provides license to publish OSS.
Awareness and action. Awareness that OSS can help agencies with day-to-day operations and federal organizations should consider the organizational factors discussed in the study when implementing tech policy and tech itself. It would be great if action is taken to overcome some of the hurdles discussed in the paper. In my almost 20yrs in government, I have never seen or taken part in a study that holistically examines a program or technology and this study provides broad implications for federal agencies to be more responsive and efficient in the future with technology.
I am socializing through internal and external agency presentations. These include our regular Community of Practice meetings, an upcoming Federal Source Code Summit at GSA im Oct., and by releasing the full study similar to publishing code. I am a software developer and often speak at national and international conferences and use social media to share knowledge about open source.
There have been a half dozen memos trying to get agencies to use and share open source software over the last 25 years, do you feel like the tide has turned based on your research, meaning employees seem to understand the value of open source?
The study shows the policy is key for permitting or mandating pushing agencies to evolve with technology. The memos help and each one creates more awareness even if it some of the same content that has been published in the past.
Open source is often a bottom-up approach to using and building technology. Those using and contributing to open source find value in doing so and more and more individuals and agencies publish OSS each year. Consumption will continue to happen fast where publication may come at a slower pace. This is a technology and approach for agencies that continues to evolve.
Based on your research, what are among the biggest challenges to getting broader adoption of open source?
This may not be new to most in government technology but has been debated recently and that is the finding in the study of the notion of the "hollow state." The hollow state refers to the emergence of third party organizations providing taxpayer funded goods and services. While very common, it begs to question whether some federal technologists are not as equipped to implement policy and technology as they could be, meaning if federal employees do not have proper tech skills, they may be bad buyers, managers, and implementers of technology. This study found this theme where some participants did not think their offices understood OSS enough to implement the policy and publish OSS.
Data shows OSS consumption and publication continues to grow and offers benefits with higher code quality, increased security, and community participation. OSS is not for every solution and does not always have to be used but coding practices with OSS should be considered.
For federal organizations to implement policy and technology, they should consider the implications in this research. These including nurturing an understanding of OSS and newer technologies, engaging with each other and the public, continually hire individuals with diverse skills as well as train the employees already on site, consider how units are constructed with breaking down silos, and where units reside in the larger organization w/ proximity to the CIO office.