
Facebook charges more for left-wing political campaigns to advertise to people with right-wing views, and vice versa, which may be exacerbating political polarisation.
UK political parties have spent hundreds of thousands of pounds in the run-up to the general election tomorrow, and one in five dollars that went on political adverts in the US in 2018 was spent online. But new research shows that political advertisers in the US have to pay more to reach beyond their core voting base.
The researchers spent more than $13,000 to promote various adverts using Facebook鈥檚 built-in marketing tools to thousands of voters living in North Carolina in July and August. The biggest campaign the team produced targeted 50,000 voters, says Piotr Sapiezynski at Northeastern University in Boston, Massachusetts, one of the paper鈥檚 authors.
Advertisement
The researchers divided Facebook users by their political viewpoint, based on voter registration records and Facebook鈥檚 own assessments of the users鈥 political leanings. They paid to show people adverts taken from pre-existing ad campaigns run by Democratic presidential candidate Bernie Sanders, and the current Republican president, Donald Trump.
Showing a Sanders advert to Republican-leaning voters cost on average three times as much as promoting it to Democrats 鈥 and vice versa for Trump ads. 鈥淭his is a huge financial penalty for trying to show people content they might not agree with,鈥 says Sapiezynski.
Even if a political campaign wants to target all voters equally, Facebook automatically serves up its advert to people more inclined to support the point of view of the politician being promoted, the team found.
鈥淧olitical campaigns may think if they鈥檙e targeting a geographical region, they will reach an audience representative of that region,鈥 says Sapiezynski鈥檚 colleague, Aleksandra Korolova at the University of Southern California. 鈥淯nbeknownst to that campaign, Facebook delivers ads to the people who are most aligned to it.鈥
Facebook was not able to provide New 杏吧原创 with a comment before publication. A Facebook spokesperson, commenting on the research, that it served adverts to people that were relevant for them. The spokesperson added that campaigns can reach whoever they want with the right targeting and spending.
鈥淎dvertising a mattress or toothbrush is very different to the functioning of democracy,鈥 says Aaron Rieke at Upturn, a research and advocacy group focused on advertising equity, and one of the paper鈥檚 co-authors. 鈥淧olitical speech shouldn鈥檛 be shunted into little filter bubbles with the same commercial calculus that drives consumer goods.鈥
Tristan Hotham at the University of Bath, UK, who also works for Who Targets Me, a campaign group promoting transparency in political advertising, says that although the researchers studied US advertising, a similar situation is likely to occur in the UK political advertising space 鈥 something Korolova agrees with. 鈥淚t鈥檚 the same algorithm and optimisation logic being used throughout,鈥 she says.
But Hotham says it isn鈥檛 clear what effect this is having on democracy. 鈥淎sserting for certain that the algorithms are separating us politically from this study is a step too far given when the study was undertaken and who was included,鈥 he says.
Promoting adverts for such diametrically opposed politicians as Trump and Sanders means the experiment may have resulted in extreme findings 鈥 the results for more centrist candidates may be more muted, says Hotham.
However, the paper raises important questions about Facebook advertising, says Hotham. Partisan content appeals most to partisans, and Facebook鈥檚 advertising ecosystem is based around engagement. 鈥淭here are systemic problems with Facebook鈥檚 ad system, and targeting should either be made very transparent or restricted much further,鈥 says Hotham.
A spokesperson for the UK Electoral Commission declined to comment on the paper鈥檚 findings, saying it only regulates political finance.
Reference: