It's no joke. The 1950 census [1] is out today!
For genealogists, the addition of another decade of Census Records to the public domain means the continuation of family histories that have been waiting for the latest records release. For some, it will be the first time they will see a relative's name in the census that they may have known in their own lifetime.
Since it captures the start of the Baby Boomer generation, the 1950 Census has also been the single largest census data pool managed to date -- over 165 terabytes -- approximately 10 times the size of the 1940 Census dataset.
“It’s taken innovation and creativity to keep this project on track throughout the pandemic and to continue to meet our project milestones," said Project Manager Carol Lagundo, who leads the 1950 Census project at the National Archives. "We hope the public will benefit from our hard work.”
This release of records also comes with many of the records already digitized and organized via machine learning.
To develop the initial name index, the archives are using Amazon Web Services’ artificial intelligence / optical character recognition (AI/OCR) Textract tool to extract the handwritten names from the digitized 1950 Census population schedules.
“The OCR being used to transcribe the handwritten names from the census rolls is about as good as the human eye,” said Project Management Director Rodney Payne. “Some of the pages are legible, and others are difficult to decipher."
Because this technology is not 100 percent accurate, the National Archives is asking for the public's help in submitting name updates to the index via the 1950 Census website. Family members and researchers can help improve the accuracy of the name index and make the records more accessible.
“This is an exciting project for the National Archives, and we know it is important information for so many Americans,” said Chief Innovation Officer Pamela Wright. "We are looking forward to collaborating with the public to refine and enhance the first draft of OCR-created names. This is a great example of automating as much as we can and then collaborating with the public to make access happen."
The National Archives suggests the following tips for searching the Census records:
- Search for a surname plus state and county of residence. Remember that common names such as Smith could yield tens of thousands of results.
- Search for the first and last name of the head of household (plus state and county of residence if known) because the surname was written on the census form only on the line for the head of household and other persons in the household with a different surname.
- If a person has an unusual or unique first name, search for that person by first name plus state and county of residence.
- Do you know the names of your relative’s nearby neighbors? Did they have a less common surname? Search for the neighbor’s head of household or surname (plus state and county of residence).
- Don’t ignore search results that aren’t an exact match - the name may have been extracted incorrectly because of the handwriting. For example, a surname like Gross might have been extracted as Grass or Gress.
Read more about the 1950 Census [2], and the U.S. Census [3] record collection, at the National Archives [4] website.
Image: Portfolio Control Label for the 17th Decennial Census, 1950. View in the National Archives catalog [5].