Alea Cook is the Digital Preservation Analyst at Cambridge University Libraries and Archives (CULA).
When we talk about digital preservation today, the scope goes beyond scanning historical volumes or maintaining local drives. Modern preservation practices are fundamentally digital, and increasingly, they include open public data. Public datasets form the statistical backbone of modern society - underpinning democratic representation, public infrastructure, and research worldwide.
 |
 |
Because government data exists in the public domain without copyright, it provides open baselines for evidence-based study and decisions. Yet this record is increasingly fragile. In the United States, public data faces continuous threats from site restructuring, broken links, policy shifts, and quiet removals. With no mandate or funding to maintain permanent web archives, vital records can vanish overnight. Having research data disappear is a genuine horror story for the public - when these records vanish, civic trust, research, and democratic accountability take an immediate hit.
That is where the Data Rescue Project (DRP) steps in, cataloguing at-risk collections and mobilising volunteers to secure open backups before access is lost. To mark World Digital Preservation Day 2026 and Data Horror week, the Digital Preservation and Open Research teams hosted a Data Rescue Hackathon at Cambridge University Library.
What Is at Risk?
Data rescue is a direct defense of empirical evidence. Across US portals, key categories of public data face targeted restrictions:
-
LGBTQ+ Community Data: Agencies have scrubbed questions on sexual orientation and gender identity (SOGI) from major surveys, obscuring the empirical baselines needed to enforce civil rights laws and track inequities in housing, healthcare, and employment.
-
Climate and Environmental Records: Emissions tracking, climate modeling, and impact assessments have been altered or removed. Open records from centres like the National Snow and Ice Data Center (NSIDC) underpin global scientific research.
-
Reproductive and Public Health: Vital public health tracking and maternal care statistics face mounting restriction. Surveillance systems like the CDC's Behavioral Risk Factor Surveillance System are essential for monitoring long-term disease trends and health disparities.
This threat is especially clear in the proposed changes to the 2030 U.S. Census (Docket No. 260903-0005), highlighted by the Hands Off Our Census initiative. These proposals threaten foundational baselines by excluding resident non-citizens from apportionment counts, replacing direct enumeration with flawed tax-record matching, and erasing race and ethnicity questions asked since 1790. Standardized census counts guide hundreds of billions of dollars for schools, hospitals, and transit - when compromised, the entire public ecosystem suffers.
Digital Investigation
Working from DRP guidelines and priority selections, our group worked through the civic archiving workflow. Participants claimed unassigned datasets, downloaded raw files, made html copies of webpages, and deposited complete packages into DataLumos, the open repository hosted by ICPSR.
Investigation skills were challenged when several listed datasets were found to already be missing on agency sites, and attendees became digital detectives - tracking down original tables and documentation through web archives, secondary mirrors, and research repositories. After finding the dataset copies, they uploaded fresh backups to DataLumos and logged the permanent records into the DRP catalog.
This work puts the LOCKSS principle into action - Lots of Copies Keep Stuff Safe. When vital public data lives on a single federal server, it has a single point of failure. By deliberately moving these datasets into independent, open-access repositories like DataLumos, distributed community archiving creates the practical redundancy needed to withstand sudden policy shifts or technical failures. It is easy to look at spreadsheets and codebooks as just dry information, but statistical baselines tell vital human stories. They are the empirical proof of communities living, working, and facing systemic challenges. When we build redundant, decentralized backups, we aren't just hoarding files - we are actively protecting public truth and ensuring the evidence future generations will need remains accessible, verifiable, and safe.
If you'd like to get involved:
-
Explore rescued data: Browse catalogued records at portal.datarescueproject.org.
-
Join DRP community hours: Participate in regular virtual working sessions on dataset tracking and metadata cleanup.
-
Host a hackathon: Use DRP open resources to run an event in your own department or library.