U.S. military to trawl through 350B social media messages
(Bloomberg) --The U.S. military plans to analyze 350 billion social-media posts from around the world to help it track how popular movements evolve.
A tender for the project, based at the Naval Postgraduate School in Monterey, California, calls for screening messages from at least 200 million users from more than 100 countries in more than 60 languages to better understand “collective expression.” Messages, including user names, will be examined for comments, metadata, location and hometown identifiers.
While it’s part of an existing Department of Defense Analysis effort to harness big data for social research, “the scale and global reach of this program is striking,” Antoine Bousquet, a senior lecturer in international relations at Birkbeck, University of London, said by email.
The study’s purpose is to look at social-media messages posted publicly between July 2014 and December 2016 on a single platform, according to a solicitation request. No private communications will be included and individual users won’t be identified in the research, according to the Navy.
What Mobilizes PeopleFacebook Inc.’s primary platform has more than 1.2 billion daily active users and Twitter Inc. has more than 300 million monthly users. Printed on letter-size paper, 350 billion individual messages would make a stack about 10,000 miles high.
T. Camber Warren, the project’s principal researcher, said the data will be used for increased understanding of communication and how patterns of discourse change over time.
He previously studied internal conflicts in Africa, showing how mass media such as radio broadcasts can have a pacifying effect, while social media can inflame collective violence.
“Social media data allows us for the first time, to measure how colloquial expressions and slang evolve over time, across a diverse array of human societies, so that we can begin to understand how and why communities come to be formed around certain forms of discourse rather than others," Warren said by email.
The data can be used to train algorithms to understand “increasingly subtle shifts in cultural context,” he said.
Sowing DivisionsSuch projects are important to help defend against adversaries seeking to undermine democracy and create divisions within western societies, said William A. Carter, deputy director for tech policy at the Center for Strategic & International Studies in Washington.
“We need to better understand how narratives are shaped and communities are formed online to defend ourselves against these campaigns,” he said.
Yet collecting and analyzing large tracts of publicly available data is open to abuse.
“There is a risk that as we learn to exploit this data to manage how people interact online, it will give governments and bad actors tools that they can use to manipulate our thoughts and behavior,” Carter said.