From the Department of Pre-crime, a guy the Trump administration might appoint to run the 2020 Census might try to cook the books. It seems a little unfair to accuse someone of a crime they haven’t even had an opportunity to consider committing yet, but I found it interesting to consider how it could be done. This sort of thing definitely happens in some countries, for example to perpetuate minority rule in spite of demographic change.

Each census starts with a simple questionnaire sent to every household. In 1970 and 1980, over 75% of those queried sent back responses. In 2010 that figure dropped to 63.5%, and in 2020, with distrust of government at an all-time high and increasing fears of data breach, the response rate will likely be significantly worse—current estimates range from 55% to 60%. To identify the non-respondents—at least 40% of Americans—the Census Bureau will have to exert considerable energy.

Thomas Brunell will determine how vigorously to track down these unidentified people in diverse locations. In rural areas that commonly vote Republican, he could direct workers to scour the trailer parks, while in urban Democratic strongholds, he could order census takers to visit non-responding households only during working hours. He could spend his advertising budget wisely in some places and less so in others. He could dispatch non-Spanish-speaking personnel into Hispanic neighborhoods. He could feed fears of deportation in immigrant communities. He could use credit rating companies to locate non-respondents, although many of the poor will never appear on such registries.

There is a simpler route Brunell could take. He might choose to do little, a tool almost as effective as the nefarious schemes detailed above. So far, the Census Bureau’s budget has been held to its 2010 level, despite a significant increase in the population and the expected rise in the percentage of those who do not respond to the initial questionnaire. Without greater resources and dedicated will, the Census Bureau could leave tens of millions of Americans uncounted. The GAO has warned that the 2020 Census is at “high risk of failure,” but requests to add funds have not yet been granted by Congress. New coping technologies are being introduced, yet trial runs have been curtailed due to lack of financing. Plans to test a Spanish-language questionnaire have also lapsed. Such constraints raise the stakes. When resources are limited, how to allocate those resources becomes paramount.