Help Us Help You: an IT Procurement Policy Wishlist for the GSA and USDS

One vendor’s view of how federal policymakers could hold underperforming contractors accountable, create digital experiences taxpayers are proud of, and bring every agency and level of US government along with them.

The US federal government currently spends $80 billion per year on maintaining and operating outdated digital infrastructure, and has accumulated trillions in overall technical debt. As taxpayers first and vendors second, we want the whole of American government—from core federal agencies to local school boards—to have software that advances national abundance and makes the public feel like their money is being well spent. 

We have five proposals for the GSA and USDS as they evaluate further reforms to how the government buys and develops software—to ensure that vendors have the right support and accountability to do their part in shoring up America’s digital excellence. 

1. Expose our performance reviews to the public

The worst legacy vendors have long operated in legally protected obscurity. While contract values are public, results are hidden. FAR 42.1503(d) was written to ensure performance reviews were only visible to select internal parties, thus protecting every vendor’s “competitive position”. In practice this has meant that serial offenders have faced few consequences while continuing to be rewarded with new contracts.

CPARS / PPIRS could be reformed to be a public Yelp for vendors. When we take taxpayer dollars for a contract, they should be able to trivially access concise reports on whether we lived up to our end of the bargain. If we underdeliver, that fact should be visible to every procurement officer and citizen alike. These reviews could also be streamlined to come much faster, with stricter enforcement on completion dates. Cycle speed matters.

2. Demand that we propose better options

In a typical RFP process, we're asked to bid on fulfilling requirements that reflect lobbying pressures, bureaucratic sprawl, and deeply outdated views of technology, not what makes for good software. Taken literally, these requirements reliably lead to worse outcomes.

Vendors can assist the current FAR reform effort by proposing alterations that account for all governing legislation in more efficient ways. But we can only do this if the RFP process allows us to be vocal when we believe that requirements don’t support those interests. By mandating that we provide these suggestions through public correspondence, this would realign our interests as vendors with those of the people ultimately paying our fees.

FAR 15.304 still limits this natural feedback loop by forcing all RFP responses to account for identical requirements, even where some requirements don't make sense. This feedback round needs to happen earlier on. By encouraging procurement officers to leverage the modular contracting flexibility of FAR 39.103, getting vendors to submit thoughtful pushback could be its own distinct, mandatory, and high-ROI stage of each RFP process.

3. Let us work backwards from real user needs

Many billions of dollars in bad software investments could be avoided if vendors were able to talk to users (on both the government and constituent side) much earlier on. 

This can be done cost-effectively, and will always pay for itself in multiples. The best private sector companies don’t design software by bureaucracy. We start with users, and return to them at every point in the development cycle. Our best guesses will never outperform real-world data about how people actually intend to (and do) use our products.

While FAR allows for user research, many procurement officers are still unaware of how much latitude they have. Requiring this research as a consideration during every RFP stage would greatly advance it as a standard best practice.  

4. Make the GSA the first buyer

When individual agencies procure and develop software, they naturally construct their RFPs to meet their own immediate needs. This basic tendency has led to a sprawl of many thousands of independent systems that (1) don’t talk to each other, (2) share little to no common architectures, and (3) each come with high independent maintenance burdens. 

The GSA, in concert with the USDS, is already reasserting itself as a centralized buyer for digital services. They could go a step further by becoming a first buyer for foundation-layer architectures that can them be commonly adopted across the government. While each agency will need some customizations and toolsets, expanding common libraries would ensure that the work each agency pays for can be leveraged by all. This would massively reduce the cost of new development while standardizing usability and feedback cycles.

5. Make it easy for state and local governments to follow

Beyond just cost duplication, fundamental interoperability between software and data systems at different levels of American government has massively hindered cross-party efforts to deliver timely services and stamp out fraud.

The federal government can be a tide that lifts all ships by expanding opt-in programs that allow interested states and local districts to easily adopt battle-tested assets like policy sections, buying guides, and reusable code libraries.

The recent passage of the SHARE IT Act was a needful step in both.guaranteeing federal code ownership and axing historical restrictions on running, improving, and sharing the code that the public has already paid for. But as some of this code may sit in private federal code repositories, access rights could be extended to all levels of government—ensuring that benefits from these forward-looking policies extend as broadly as possible.

GovTron connects top talent from Silicon Valley to all levels of American government. Learn more here.