New Delhi, India – In October 2020, in the lead-up to the assembly elections in the eastern Indian state of Bihar, the Bharatiya Janata Party (BJP) ran an advertisement on Facebook insinuating that rival Rashtriya Janata Dal’s chief ministerial hopeful Tejashwi Yadav was instrumental in getting a politician killed.
The advertisement headline read: “Tejashwi Yadav threatened RJD worker Shakti Malik and said, ‘I am Lalu Prasad’s son and deputy chief minister, if you raise your voice I shall get you killed.’ The threat was real. Shakti Malik was killed.”
Bihar police later found out Malik was murdered by his business rivals. But in just one day, Facebook showed the advertisement 150,000 to 175,000 times, mainly to Bihar’s male voters. For that, the BJP paid Facebook a minuscule 4,250 rupees ($56), or less than three paise (less than a cent) per view, and helped make it go viral.
A viral BJP ad insinuated that the rival chief ministerial hopeful in a elections was instrumental in getting a politician killed.
This is not the only instance where the BJP got voters’ eyeballs on Facebook for cheap. Part 3 of this series yesterday revealed that the BJP has consistently received less expensive ad rates than the Indian National Congress – colloquially known as Congress – and other parties in nine out of 10 elections over 22 months, allowing it to reach more voters for less money than the opposition.
But how does Facebook’s advertising platform favour the BJP?
While the Wall Street Journal had reported earlier on instances of individuals within Facebook’s management favouring the ruling party in India, the leg up that BJP gets while advertising on the platform may not be dependent on individuals in the company. Previously unreported evidence shows the advantage that BJP’s ads get is possibly due to Facebook’s algorithm, designed to keep users hooked to their newsfeeds.
A review of Facebook’s advertising policies suggests the company’s pricing algorithm favours advertisements that are likely to generate more “engagement”- number of likes, shares, comments. So if a political party and its proxies have pumped in enough advertisements and campaigned heavily, often with emotionally – or politically – charged content to increase engagement on Facebook, its advertisements would automatically work out cheaper. A similar reach would be costlier for smaller parties.
In the United States elections in late 2020, this arrangement seems to have helped former US President Donald Trump get lower advertising prices than rival Joe Biden.
In India, the BJP appears to be reaping the benefits of the same system.The BJP and its affiliates are the biggest direct publisher of political ads on Facebook.
The social media platform has also allowed a large number of BJP’s ghost and surrogate advertisers to campaign for the party, bypassing, in the process, India’s election laws and Facebook’s own guidelines.
These surrogate advertisers almost doubled BJP’s visibility during elections, often with disinformation and communally charged content, as we reported in Part 1 and Part 2 of this series.
BJP likely draws favourable rates from Facebook’s advertisement algorithm by sheer virtue of its dominance and its polarising content, experts say. But its dubious proxies and surrogate advertisers provide a multiplier effect, further bringing down the rates at which BJP is able to buy advertisements on Facebook.
In the past Facebook’s top executives in India have closely worked with the BJP’s campaigns to train the party workers on how best to use Facebook to turbocharge its campaign.
Unlike television and print media, Facebook does not have a predefined rate card for advertisers. It auctions viewing slots – the opportunity to show an advertisement to a set of target audiences on Facebook.
Typically, auctions are meant to discover the highest bidder. But the game-changer is Facebook’s algorithm, which can pick a lower bidder if it deems the advertisement more “relevant” to targeted audience and subsidises the ad to make it cheaper.
This audience can be narrowly defined on the basis of demographics, behaviours and other attributes chosen by the advertiser, or can be broadly defined to achieve intended outcomes – like getting 1,000 app downloads or one million link clicks – that are then executed by Facebook based on the data it collects about users. Advertisers can outsource finding the desired audience to Facebook, making it one of the most attractive platforms for them.
During an auction, when two advertisers vie to get into the Facebook timelines of a group of people who display particular traits, normally, the highest bidder wins. When demand falls, it’s happy hour for advertisers.
But in Facebook’s case “an ad that’s relevant to a person could win an auction against ads with higher bids”, it says on its Business Help Center page for advertisers.
A Q&A on how ad auctions work on Meta's Facebook platform
Meta’s algorithm subsidises ads it believes are more ‘relevant’ to targeted users, making them cheaper [File: Meta Business Help Center website]
Facebook’s algorithm determines the “relevance” of a piece of content based on its estimates of how much it expects the targeted audience to “engage” with it.
Most popular social media platforms provide focused advertisements to clients; that is the value they offer advertisers. Facebook’s business also benefits because advertisements that hook users, keep them glued to the timeline. Facebook sells it as a win-win to both the advertiser and its users.
Facebook is not unique in offering discounts systematically to its biggest clients. Many brick-and-mortar businesses have always done this. So why have Facebook’s ad pricing policy raised questions across several democracies?
The strategy that works as a perfect business model is not merely about advertising goods and services to be sold. Instead, it leads to unfair and unequal opportunities for political parties in electoral contests and skews the field.
“If a political party learns how it can game Facebook’s algorithm, then it can get its content out to a larger number of people for much cheaper, thereby amplifying its own narrative and gaining politically,” said Shivam Shankar Singh, a former political consultant and author of the books The Art of Conjuring Alternate Realities and How to Win an Indian Election. “Gaming this algorithm often involves moving from informative content to emotionally inflammatory content. Such content is amplified by the algorithms, and it also yields greater benefit to political parties looking to polarise an electorate.”
To ensure that differential ad pricing does not undercut competition in favour of any political candidate in the US, a law there mandates that all TV and broadcast media stations must charge the same rate to all candidates. Social media is yet not regulated under this particular law but the country is debating this. In India too the Election Commission requires transparency over political advertisements but hasn’t applied that standard to social media.
“If it was about commercial advertisements, it’s completely the prerogative of the businesses and the publishers to decide what price they want to charge for the advertisements,” former chief election commissioner of India SY Quraishi told Al Jazeera. “But political advertisements need to be regulated in the context of the election laws,” he said adding, that campaigns on social media in India do need better regulations, “at least of the same level as the traditional print and broadcast media.”
Advantage, dominant and polarising party
“The primary interest of a platform like Facebook is to maximise the engagement and duration that users spend on its platform, so that it can sell more advertising,” says former political consultant Singh. “Due to this, the Facebook algorithm often ends up promoting the most outrageous and inflammatory content that would get emotional responses from users.”
A study by researchers at Northeastern University, the University of Southern California and non-profit tech justice organisation Upturn in the US showed in December 2019 that Facebook’s ad-delivery algorithm promotes political polarisation.
“For campaigns, such delivery may inhibit them [politicians] from reaching beyond their existing ‘base’ on Facebook, as getting ads delivered to users the platform believes are not aligned with their views may become prohibitively expensive,” the study said.
In effect, Facebook encourages putting blinkers on its users by diminishing access to diverse views through its pricing algorithm. So, in India, anyone Facebook infers to be inclined towards pro-Hindutva politics and Prime Minister Narendra Modi can be targeted at a lower price if the content of the advertisement also eulogises the two. But it would be more expensive to show the same audience the advertisements that disparage pro-Hindutva politics and Modi.
Northeastern University’s Piotr Sapiezynski, one of the authors of the study, said the price of advertisements may vary significantly for two parties depending on how they target the audience. But “in a situation where both parties are targeting the same audience, advertising of the party which has much more support on Facebook in that audience would be cheaper in general than the party which has much less support,” he said.
BJP’s official page has 16.7 million followers and its leader, Narendra Modi, has 46.8 million Facebook followers. The sheer size of supporters on Facebook gives BJP’s content more interactions. In comparison, the Congress page has just 6.2 million followers, and its leader Rahul Gandhi is followed by only 4.7 million people.
In what is a bonanza for the ruling party, Facebook’s advertisement algorithm is likely to show its ads to more people for less money because of its dominance, elbowing out campaigns of others, such as Congress, with fewer Facebook followers.
By looking the other way while the illegal proxies and surrogates of BJP proliferated unchecked, Meta Platforms – the owner of Facebook – ensured that pro-BJP content got discounted rates on its platform.
This arrangement, if left intact, could allow an already dominant BJP to grow its support base on Facebook exponentially while putting other political parties in an increasingly disadvantaged position with each election campaign.
In response to a list of detailed questions over email, Meta said: “We apply our policies uniformly without regard to anyone’s political positions or party affiliations. The decisions around integrity work or content escalations cannot and are not made unilaterally by just one person; rather, they are inclusive of different views from around the company, a process that is critical to making sure we consider, understand and account for both local and global contexts.”
Meta, however, did not respond to the specific question on differential pricing charged by its algorithm on BJP’s and other parties’ ads or on the study linking the algorithm with political polarisation. Meta’s full response can be read here (PDF).
ECI did not respond to questions despite repeated reminders. BJP’s chief spokesman Anil Baluni and IT and social media head Amit Malviya too did not respond to queries.