Laura Bliss is a staff writer at CityLab, covering transportation, infrastructure, and the environment. She also authors MapLab, a biweekly newsletter about maps that reveal and shape urban spaces (subscribe here). Her work has appeared in the New York Times, The Atlantic, Los Angeles, GOOD, L.A. Review of Books, and beyond.
The agency has led major advances in data science since 1890.
The U.S. Census Bureau is on track to go digital in its 2020 survey, the Pew Research Center reports. The agency is aiming to get 55 percent of American households to respond to the next decennial census digitally, using personal computers or mobile devices. (Neighborhoods with low internet usage rates and elderly populations will still receive paper surveys through the mail, as will people who don’t respond online.)
Excepting a handful of households in 2000, the 2020 census would be the first in which Americans haven’t had to fill out a paper survey, assuming all goes to plan. It’s a huge cost-saving move, as the more citizens who respond on their own, the fewer enumerators the bureau has to send door-to-door. There will also be far fewer employees compiling the mailing list: Rather than send employees to all 11 million census blocks in the country to gather addresses (as in the past), the bureau plans to cover just 25 percent of those blocks using on-the-ground employees, and the rest using aerial imagery and GIS data. Between cutting labor, printing, and mailing costs, the agency expects to save $5.2 billion compared to the 2010 census.
The advance could also put the bureau back on track with its historic position at the cutting edge of data processing. After all, this is the agency that has been helping advance information technology since 1890, when it leased a fleet of “tabulating machines” to turn cardboard punch-cards printed with survey responses into machine-readable data. Not only did they radically accelerate census processing, but within a few years the devices were being used by department stores, electric and gas utilities, drug manufacturers, oil companies and railroads. Their inventor, Herman Hollerith, went on to found the Tabulating Machine Company, which eventually became IBM.
Versions of Hollerith’s machines (eventually electronic ones) dominated the landscape of data processing until the 1950s and the arrival of early computers. Which the Census Bureau was also all about: In 1950, the agency first processed the census with a gigantic electronic computer—the first ever commissioned for civilian purposes—which tabulated twice as fast as the old machines. For the following census, the Bureau switched to an optical device that “read” census results on microfilm and transferred the data to magnetic tape, which was then fed to computers—a technology not unlike to the optical scanners the agency has used since 2000.
But concerns about fraud, delays, and glitches have caused the bureau to shy from going online over the past several decades. In that sense, it has fallen from the fore of data science. While moving online isn’t exactly on par with inventing a precursor to the computer, the bureau’s plans for 2020 could include software and IT infrastructure that advances data processing once more.
Or not. 2020 is still four years away, and whatever technology the bureau is developing now could be nearing obsolescence by the time it hits “send” on millions of digital surveys. But as the rest of the world moves online, it’s time for the once-revolutionary agency to keep up—and hopefully, push technology a small step further.