In a recent conversation, Denice Ross — who previously held the post of U.S. Chief Data Scientist — described her role to a casual acquaintance by likening herself to the Lorax, only with federal data as her cause rather than trees. Through her role leading the Federation of American Scientists' nascent Data Policy Institute, Ross has positioned herself as an advocate for the preservation and accessibility of federal data.

On the DailyScoop podcast, Ross argued that federal data weaves through nearly every corner of everyday life — though she noted her analogy might not fully hold — and that the subject draws less public scrutiny than many other policy areas, leaving room for a dedicated advocate to raise its profile.

Launched earlier this year as a dedicated division within a nonprofit science think tank, the Data Policy Institute carries an explicit organisational mission — advancing the field of data policy — and its staff have been working to bring that purpose into plain view rather than letting it remain an unspoken given. Four pillars structure that effort: monitoring federal datasets for changes, identifying openings for public participation, and charting the disconnected terrain of federal data policy to reveal where coordination is absent.

The institute's founding goals have remained broadly consistent over time, yet the organisation itself took shape during — and was influenced by — a period of notable turbulence in federal data. Records tracking federal data activity, maintained by the institute through a newly established monitoring tool called the Federal Data Terminations Tracker, show that the Trump administration has removed or modified hundreds of data categories while dozens of datasets have been taken offline.

Those alterations often took place without public notice, directed at datasets that fell outside the administration's preferred policy positions, and the erosion of institutional capacity within federal agencies has compounded the damage — a loss Ross described as potentially the hardest to reverse, adding that her broader concern was whether people would come to recognise the role that government-collected data plays as an unseen but foundational layer of public infrastructure.

For now, the institute is recording those changes, though Ross noted her aim is to move people beyond simply relying on the data toward actively championing it whenever openings allow. Attendance at webinars covering notices about particular datasets in the U.S. Federal Register had reached into the hundreds, she noted, pointing to that as a marker of progress on the effort.

Efforts on that front align with a broader federal initiative to implement the bipartisan OPEN Government Data Act — a legislative area the organisation is anticipated to address at a later stage. Enacted in 2019, that law put in place much of the administrative structure now guiding federal operations, among them the positions of agency chief data officers and the interagency council through which those officials coordinate.

During her time in the Biden administration, Ross held the role of U.S. chief data scientist — only the second person to receive that appointment — and has since outlined a vision for building what she terms "civic infrastructure," a foundational layer intended to prepare communities for broader participation as chief data officers extend their public-facing work in the coming years. The position, brought into existence during the Obama years, can shift in its priorities depending on whoever holds it, she noted.

Her goal was to use federal data to understand how policies and programmes were affecting the public, and Ross described the role of chief data scientist as one focused on expanding the capacity of federal agencies to break down data in ways that could reveal which communities were underserved or carrying a disproportionate share of burdens, allowing those agencies to take steps toward addressing those gaps.

Clarity arrived for Ross only at her second White House posting, she noted — pointing out that two and a half decades of practical work in data policy had gone by before she could put a name to her own professional territory, partly because the field itself had never settled into a firm definition; the route to that understanding had wound through a Presidential Innovation Fellowship and a senior advisory position during the Obama administration.

Ross has noted that public policy programmes can be completed without any exposure to the frameworks governing federal data, and that data scientists already employed in federal roles often encounter those same policies only after they have begun working. One goal driving the institute's work, she explained, involves building data policy into a more structured discipline capable of addressing what the public actually needs.

That work from Ross predates the formal establishment of the Data Policy Institute. That same month, a Federal Data Field Guide took shape through a joint effort between Ross and Christopher Marcum, who had previously held positions as both a senior statistician and a senior scientist within the White House Office of Management and Budget.

Produced with backing from a UC Berkeley technology policy fellowship, the document organises federal data through a nature field-guide structure — pairing animals and insects with each data category as mascots — a format Ross characterised as an attempt to inject some levity into how the federal data ecosystem is navigated and understood. According to Ross, the guide lays groundwork for broadening data thinking beyond purely statistical and administrative sources, while also acting as an entry point for those learning about data policy. The material served as the foundation for a Data Skills for Congress cohort — made up of congressional staffers — when she led the sessions that ran over the summer.

Shortly after assuming the presidency, Trump signed executive orders instructing federal agencies to dismantle diversity, equity, and inclusion programmes and to substitute "sex" for official references to "gender" and "gender identity" — a set of directives that was followed by the removal of thousands of datasets from government systems. Ross noted that although the bulk of the removed records were restored, datasets tied to gender identity, racial classification, ethnicity, and climate measurement did not come back.

Initiatives aimed at retaining that body of information started taking shape during the same period. Volunteers coordinating through the Data Rescue Project set out to build a crowdsourced archive of datasets at risk of deletion or alteration. Tracking federal policy shifts became the foundation of the Impact Project's data work, and before that year closed, Ross had assembled the first version of what would grow into the Terminations Tracker.

The project's underlying aim, Ross noted, was to establish a uniform method for tracking datasets that had been taken down or discontinued — a framework intended to let researchers gauge actual losses with greater precision after early circulation figures had overstated the extent of the problem — and the effort itself had grown from a broader collaborative push to crowdsource a working inventory of what had gone missing. She noted that the numbers pointing to dataset losses — which gained traction in the opening stretch of the administration — traced back largely to readers misinterpreting routine shifts in how many datasets appeared on platforms such as data.gov, a problem compounded by the absence of any shared framework for distinguishing between products, websites, and tools as separate categories.

A structured approach to monitoring federal data took shape when Ross and her colleagues determined that a standing, formalised process would serve a clear purpose — leading to the August debut of the Terminations Tracker. As of the current count, the tracker holds 38 discontinued datasets alongside 337 individual data elements pulled from availability since January 2025.

Melanie Klein, who works on federal data policy at the Federation of American Scientists, described the tool's purpose as building on an existing resource called Dearly Departed, with the aim of producing the version of data terminations most useful to policymakers. Klein outlined two criteria guiding the assessment: whether termination rates exceed the baseline turnover normally seen between administrations, and whether the affected datasets, analytical models, and tools are so specific to federal operations that no outside entity would be positioned to repair or reproduce them.

Forms and surveys turn out to be the easiest category to monitor, Klein explained, because any revisions to those documents leave a visible trail in the Federal Register and on Reginfo.gov. Layoff notices are set to go out from a number of those agencies as well. Beyond those automated methods, Klein explained that the project depends heavily on its network of data specialists to flag datasets that have stopped being collected — a gap that daily searches across news outlets, social media, and federal websites can identify but not fully close on their own.

Hart, who leads the Data Foundation as its chief executive — an organisation operating without partisan affiliation or profit motive — told a broadcast outlet that broader public involvement in data policy was a development he viewed positively, adding that the national data infrastructure had gone without adequate funding across multiple administrations from both parties and that no single organisation possessed the capacity to address that gap on its own. Hart noted that even routine dataset retirement or modification carries an obligation to the people who depend on that data — they warrant advance notice and a meaningful say in any such decision.

Erie Meyer, who helped establish the U.S. Digital Service and previously served as chief technologist at the Consumer Financial Protection Bureau, said the volume of ongoing disruption makes it all the more necessary to understand what information is being withheld, adding that those working to document and catalogue the changes — Denise among them — are doing work that will matter for any eventual recovery.

Beyond its Terminations Tracker, the Data Policy Institute operates a separate data-quality platform at DataIndex.us and has entered the public debate over far-reaching proposed revisions to the decennial census, including participation in the formal comment period. Among the adjustments proposed, questions relating to race and ethnicity would be dropped from the form.

Ross outlined plans to broaden the organisation's reach into hard-to-monitor domains — among them satellite imagery and sensor data — while also keeping the public better informed about data-related matters. Ross expressed a desire to see data-policy specialists deliberately embedded within federal agencies that handle substantial data responsibilities, noting that whoever takes office next would benefit from having those experts positioned to contribute from the outset.

Ross reflected that her tenure as the nation's chief data scientist left her wishing more had been done to communicate to ordinary people why government-held data carries real relevance to their daily lives, a form of outreach she described as genuinely rewarding work. When asked about public awareness of government data, Ross indicated that a reduction in federal capacity — including workforce losses, contractor departures, and the disbanding of advisory committees that help keep data collections methodologically grounded — places all federal data collections at risk of deterioration, with downstream effects on the quality of life for people who depend on them.