Page 44 - FCW, Nov/Dec 2017

Acquisition
It now seems possible that using agile
rather than traditional waterfall techniques have a lot of evidence yet.
cult. But one would expect that if there had been clear improvements across at least most of the board, they would be visible. As a general matter, I don’t believe we can claim that.
As I scour the procurement landscape for proven results, three areas do come to mind. All three date from 2009 or later, and two remain works in progress.
The most important performance improvement is the reduction in cost growth for developing new weapons sys- tems, the largest cost category in the government’s procure- ment budget. The fact that the systems require beyond-state- of-the-art technology creates signi cant risk in terms of cost growth and performance uncertainty, and it necessitates cost- reimbursement pricing that is not conducive to controlling costs.
Now, however, there seems to be plausible evidence that the cost growth for developing new systems has decelerated since 2009. That trend appears in evidence provided by DOD and con rmed by the Government Accountability Of ce.
Of cials believe the single most important reason for that downward trend was the spread of “should-cost” analyses for programs under development. Those analyses, which are conducted every year, are lled with concrete suggestions for how a program might reduce costs. Acquisition executives in the military services develop overall cost-savings targets for each program’s should-cost analysis, and program managers are supposed to track progress the same way they would with other performance metrics.
Those analyses have been crucial to creating counter pres- sures on acquisition of cials. Without that metric, they would focus only on the traditional metric of “getting money out the door,” which created incentives against cost control.
Impact of agile: Too soon to tell?
Two other areas that show promise involve IT, though it is still too early to tell. One is the acquisition of IT systems. The big-picture, 25-year view here shows a system still beset by signi cant underperformance, with projects regularly show- ing big cost and schedule increases and performance short- falls. In 2015, GAO put IT acquisition on its list of agencies
and program areas at high risk of waste, fraud, abuse or mismanagement.
It now seems possible that using agile software develop- ment methodologies rather than traditional waterfall tech- niques might be helping, but we don’t have a lot of evidence yet. Executives in the of ce of the U.S. CIO told me they don’t have enough data to venture even a tentative conclusion. The accounts I give here are based on reports by participants, who obviously have their own biases, and some media accounts.
The rst major use of agile in the federal government was a transition at U.S. Citizenship and Immigration Services from a traditional procurement approach to agile for an organi- zationwide business process redesign in an agency that had been mostly paper-based. It was led by CIO Mark Schwartz, who came to the agency from the private sector in 2010.
It had been a classic troubled big project. USCIS spent ve years automating a single, limited business process, which later had to be discarded because it wasn’t effective for agen- cy adjudicators. After arriving in 2010, Schwartz started doing four releases a year. In 2015, he moved to the cloud and full- edged agile development. Two years later, Schwartz reported, the agency had digitized 30 percent to 40 percent of its volume. By the time he left in 2017, the development team was doing two to three releases a week.
Dave Zvenyach, who until October ran the acquisition team at 18F and managed the General Services Administration’s blanket purchase agreement for agile services, shared what he observed during the transition from waterfall to agile for development of a searchable government spending website required under the Digital Accountability and Transparency Act of 2014 and for task orders under the agile BPA.
The Data Act requires taking elements from perhaps 600 federal data sources related to spending on contracts, grants and payments; standardizing the data; and putting it into one online, searchable system. Earlier, the agency had taken four years to develop an analogous capability for accounting data for a similar project using waterfall methodology. Yet when the new spending reports began posting to USASpending.gov in early 2017, legislators and GSA of cials praised the team for implementing the changes on schedule and without drama.
22 November/December 2017 FCW.COM