Technology Planning Hurdles Continue for 2020 Census

Published: December 16, 2015

DOCCybersecurityDigital GovernmentInnovation

Planning and operational testing is underway for the Census Bureau’s next Decennial Census, which is estimated to cost $12.5 billion and deliver $5.2 billion in cost avoidance through improvements.

The operational plan for the 2020 Census was released in early October 2015, outlining various challenges against which the project will contend. According to the test schedule laid out in that document, 2016 will kick off production system development and testing along with testing integration of self-response and non-response. Since optimization of self-response is one of the key innovation areas touted as part of this new approach, the execution of improvements to the self-response strategy warrants careful consideration.

While plans promise to deliver cost savings, a recent report notes the inherent, significant cybersecurity risks associated with greater reliance on technology for enumeration. The threats posed by the web-based approach range in degree of severity including input integrity issues, potential phishing scams, security of confidential data, and attempts to affect apportionment of the House of Representatives. Additionally, the 2020 Census “will take place against a background of long-term increasing distrust of government, and also increased awareness of the government’s collection of data on the population.” The JASON report offers a more granular view of some of the issues related to the census technology strategy, particularly validating respondents and combatting fraud. If these aspects are addressed appropriately, the Census will be well positioned to detect and mitigate attempted fraudulent use. To that end, the report provides over 20 specific recommendations related to ID processing, administrative records, levels of validation for non-ID responses, mobile devices, fraud detection and mitigation, and algorithm improvements.

Specifically, the report offers recommendations to improve the current record-linkage and matching algorithms in use and to expand engagement with academia. Somewhat surprisingly, the report suggests that the organization has some of the necessary capabilities to make applied research enhancements, “but that the appropriate people and computer resources are now assigned to operational tasks.” In the current technology employment environment, it’s not unusual for agencies struggle to get sufficient in-house talent. Suggesting that the Census Bureau already has the resources it needs is promising, but the question becomes whether they’ll manage those resources optimally. In that vein, the report suggests revisiting the idea of increasing resources for applied research to enable Census Bureau personnel to better support the 2020 Census.

Although the report does not delve into the cybersecurity issues around the new approach for the census, the combination of recent breaches of government information systems and concerns for potential for identity theft are likely to add to the ranks of individuals that are skittish of handing over personal data. The JASON report mentions the importance of maintaining “fallback processes” to address computer-illiteracy and “computer-hostility.” It seems possible that, at some point, levels of distrust in the government’s handling of data may eat away at the promise efficiencies of the new, innovative approach.