Launching the 1950 Census Website: Innovation and Collaboration at the National Archives

Pamela Wright, Chief Innovation Officer, U.S. National Archives and Records Administration

Pamela Wright, Chief Innovation Officer, U.S. National Archives and Records Administration

At the stroke of midnight on April 1, 2022, the National Archives and Records Administration (NARA) launched the highly anticipated 1950 Census website. Project managers, developers, archivists, and communications staff, as well as agency executives, including the archivist of the United States, huddled around their laptops and held their breaths in anticipation.

We felt the whole world watching as years of our efforts were now focused on this moment. Weikai Zhang, our lead developer, let us know the site was live. I took a deep breath and clicked the red ‘Begin Search’ button on our launch page. I was delighted to see the homepage immediately pop up. But this was just the beginning of the test.

As the nation’s record keeper for permanent federal government documents, the National Archives holds and provides access to the decennial censuses. The U.S. Censuses have a statutory 72-year restriction on access for privacy reasons, and April 1, 2022, was the first day that the 1950 Census could be made publicly available.

In the 1970s, NARA started providing access to the Census on microfilm, the popular storage and access medium of the time. In 2010, NARA partnered with a government contractor to create an online version of the 1940 Census. Unfortunately, the 1940 Census website crashed shortly after its launch in 2012 and took days to restore. That experience marked a turning point in our approach to the then-upcoming 1950 Census. We took the lessons learned from the 1940 Census launch and crafted a strategy for the successful launch of the 1950 Census.

In the decade between launches, NARA staff diligently digitized and described the 1950 census records. We hired staff with expertise in designing and deploying cloud-hosted applications. We developed our web team and expanded their knowledge from HTML coding to Drupal coding and user experience design. We hired a web manager who had expertise in agile project management and human-centered design and who was an excellent communicator. We elicited feedback from genealogical and Native American contacts along the way and ensured that a variety of staff from around the agency was updated continuously, over dozens of sprints, as we developed the website. And we took advantage of advances in cloud hosting, which enabled us to provide the full 1950 Census dataset for download on Amazon’s Registry of Open Data.

“We decided early on to focus what the public told us should be the essential feature of the site - name search. We did our research and used a new AI-powered optical character recognition (OCR) tool to make the names searchable in the millions of pages of handwritten records.”

We decided early to focus on what the public told us should be the essential feature of the site - name search. We did our research and used a new AI-powered optical character recognition (OCR) tool to make the names searchable in the millions of pages of handwritten records. We knew the technology would not extract handwriting perfectly, so we incorporated a user-engagement tool that supports public contributions to improve the name index.

We also engaged in months of performance testing. Using traffic data from the 1940 Census and multiplying it by a factor of ten, we set benchmarks to ensure the availability of the site during peak usage times.

That night, as our team all clicked on the website, I turned to Twitter to see what the public was saying. I read out the comments from people who were beginning to use the website.

We continued to watch as the number of users came pouring in that night. Our website supported millions of clicks from around the world through the night and over the following days, weeks, and months. Diverse staff from across the agency collaborated for years on this effort, and it was a joy for all of us to see our work meeting the mission of the agency - to provide access to the permanent records of the federal government to the American people and the world.

The team is already building off the lessons learned from the 1950 website launch. We are scheduling regular cross-office meetings as early as FY23 to begin planning for the April 2032 launch of the records of the 1960 Census. After all, 2032 is just around the corner.