Data scientists perform last rites for ‘dearly departed datasets’ in 2nd Trump administration

As federal agencies continue to reorganize under President Donald Trump’s administration, data scientists have raised alarms about what they call a concerning pattern of government datasets being eliminated, altered, or made less accessible since the president returned to office.

A group of U.S. data experts chose Halloween to release their “Dearly Departed Datasets” report, documenting federal information resources that have been modified or removed entirely. The Federation of American Scientists and other data professionals compiled the list to highlight what they describe as systematic targeting of data collections that don’t align with the administration’s policy priorities.

The review specifically points to datasets dealing with gender identity, diversity and inclusion initiatives, and climate change as being particularly vulnerable to modification or removal. While the timing of the Halloween-themed release added a touch of dark humor, the researchers emphasized the serious implications for government transparency and evidence-based policymaking.

According to the report, affected datasets fall into several categories: those completely eliminated, those with specific variables deleted, those made less accessible through removal of public-facing tools, and some that have found “second lives” outside federal platforms through preservation efforts by non-governmental organizations.

Despite the concerns, the researchers noted that the total number of completely terminated datasets remains relatively small—in the dozens—compared to the hundreds of thousands of data collections produced by the federal government. However, they warned that many more could be at risk due to staffing reductions and policy changes.

Among the most notable casualties identified in the report is a Census Bureau dataset that explored relationships between income inequality and vulnerability to natural disasters. Also eliminated was a health surveillance network that monitored drug-related emergency room visits, and a survey tracking hiring patterns and work hours at U.S. farms.

Other datasets saw targeted removals of specific elements rather than complete elimination. The review found that race and ethnicity columns were removed from federal workforce data, statistics on transgender inmates were deleted from prison population reports, and three questions relating to gender identity were struck from a crime victims’ survey.

The researchers expressed particular concern about the loss of institutional knowledge following staff departures across federal agencies. They noted that cost-cutting measures implemented under the oversight of Elon Musk, who has been tasked with government efficiency reforms, have resulted in significant personnel changes that could further jeopardize data collection and maintenance.

“When you lose the people who understand how the data is collected and structured, you risk losing the data itself, even if it’s technically still available,” said one contributor to the report who requested anonymity due to ongoing work with federal agencies.

The report suggests that datasets reflecting poorly on the administration’s policy priorities remain particularly vulnerable to alteration or removal. This selective approach to data management could create blind spots for researchers, policymakers, and the public regarding important social, economic, and environmental trends.

Some data scientists have begun creating independent backups of government datasets they consider at risk, establishing alternative repositories to preserve information for future research and policy analysis. These efforts reflect growing concern among the scientific community about potential data loss and the implications for evidence-based governance.

Federal data serves as a crucial resource for researchers, businesses, and policymakers, providing standardized information collected at scales that would be impossible for private organizations to replicate. Changes to these resources can have far-reaching impacts on everything from economic forecasting to disaster planning.

The researchers behind the “Dearly Departed Datasets” report have pledged to continue monitoring changes to federal data resources and advocating for transparency in government data management practices.