RELATED STORY: The 20-Petabyte Idea

To help journalists and the public understand the origins and effects of political advertising, the Internet Archive created the Political TV Ad Archive, a freely available database of TV ads related to the 2016 U.S. presidential election. The Archive utilized the “Duplitron,” an open source audio fingerprinting tool, to track occurrences of ads backed by candidates, super PACs and other groups—all linked to fact-checks and to downloadable data on where and when ads were aired, sponsors, subjects and messages. The project used the same technology to track which clips TV news shows chose to air when reporting on the presidential debates. The project archived more than 2,000 unique ads; leading to more than 2 million views of ads, as well as logging more than 130 fact-checks of the ads and informing election reporting by major media outlets such as The Washington Post, Forbes, The Atlantic and Fox News. The Internet Archive is currently exploring new partnerships with experts in machine learning (artificial intelligence) to make such curated collections far less labor-intensive to create and analyze. By doing so, journalists, scholars and the public will have enhanced tools to establish the provenance of public statements, help assess the veracity of factual assertions, track the evolution of “talking points,” and provide deeper insight into the ow of ideas and the news.

The Rita Allen Foundation supported the initial development of the Archive’s technology through a pilot project, the Philly Political Media Watch Project, which collected ads aired in the Philadelphia region in the lead-up to the 2014 midterm election. In 2016 the Foundation helped to sponsor the primary election phase of the Political TV Ad Archive through the Knight News Challenge on Elections; a separate grant allowed the Archive to record general election advertising in eight battleground states, as well as improve its technology and extend its outreach and training to journalists, scholars and nonprofits.