Government hasn’t kept pace with advances in technology. Only 13% of major government software projects succeed, and the successful and failed ones alike cost 5–10 times more than they should. When these projects fail, so too do the public policy initiatives that depend on them — unemployment insurance, DMVs, healthcare exchanges, paid family & medical leave, etc. — leaving behind the millions of Americans who rely on those programs.
We’ve seen in the response to the COVID-19 crisis that even if lawmakers move quickly to pass legislation to get money to laid-off workers, small businesses, and hospitals, those policies can’t be implemented effectively when the technology tools used to apply for, distribute, and track funds can’t be easily modified or don’t work.
This is particularly evident at the state level. States all provide the same general services to the public, but they rarely hold the legal rights to the software underlying those services. The result is that when a state manages to build software that works well, they cannot easily share that software with other states Some software sharing is done on an ad-hoc basis, but governance structures are difficult to set up, and are reinvented each time. The most common way that software is “shared” is when the vendor that built it then sells it to other agencies as “commercial, off-the-shelf” software that still requires extensive customization. The result is that states end up with several sub-optimal outcomes: 1) they don’t own the software needed to carry out their mission; 2) they’ve committed to a customized system that can’t be easily or automatically updated; and 3) they lose the ability to benefit from competition among vendors for better price and service terms.
States’ over-complex procurement processes result in insurmountable barriers for new software providers and insufficient competition. To date, the federal government, which funds a substantial portion of state technology modernization efforts, has been slow to intervene or create better incentives.