Donald Trump’s victory in the American presidential elections can be attributed to several factors: political correctness, anti-establishment sentiment, anti-incumbency, fear-mongering and nationalism being the prominent ones. But these factors are purely political. Trump tapped into something which goes beyond beliefs and ideology. If his election is a sign of things to come, it signals the end of clear objectives and ideological demarcations. The capabilities to pinpoint what it is that you seek, and finding ways to promise you exactly that, is being perfected election after election.
How did a New York billionaire, host of Celebrity Apprentice, who spoke about women in the lewdest of ways, take down the ecosystem’s candidate? She was the wife of a former president, a former secretary of state herself, she was choreographing the media’s every move, she had the backing of Wall Street and even sovereign countries, she had the administration’s help and most importantly she had the ability to rig elections, which was witnessed during the primaries. And yet she lost. Her opponent had one weapon, and we can thank our stars that her camp never understood how powerful it was and how much it could hurt them: Big data.
Everything we do online leaves digital traces. Every purchase, every like and every comment are like little windows that allow a peak into somebody’s mind. When an individual’s traces are gathered, certain co-relations about what they translate to in the real world allow the creation of models to predict, with amazing accuracy, what kind of person you are dealing with. The approach was tested by psychologists in the 1980s first, but back then the models were rudimentary and the results could not be used in a productive way. Facebook began using and perfecting this approach in 2008.
The brain behind Facebook’s attempt was Michal Kosinki. Four years into the project, he had perfected the model to such an extent that any 68 Facebook likes by one user could determine their skin colour with a 95% accuracy, their sexual orientation with an 88% accuracy, their political and religious affiliations, the level of their intelligence, whether their parents were divorced as well whether they used drugs, cigarettes or alcohol, all with an accuracy north of 80%. The model was so advanced that 70 likes by one user were enough to deduce more than their friends knew about them, 150 were enough to deduce more than their family knew about them, 300 were enough to deduce more than their partner knew about them. Anything more than 300 and you were in unchartered territory: you knew more than the user knew about himself or herself.
The making of models was not restricted to Facebook likes. The kind of posts one shares, the kind of pictures one uses on their profile, the number of friends or contacts one has, the amount of time one spends online- models providing high accuracy were created for each function. The motion sensors on our smartphones provide data about how fast we move. This is to give you an idea of how far those creating these highly accurate models have taken the game. An article on the website Motherboard stated the conclusion that Kosinki arrived at beautifully: our smartphone is a vast psychological questionnaire that we are constantly filling out, both consciously and unconsciously.
Until 2014, Kosinki was simply combining these different models and creating a massive and hugely accurate database. He was approached both for his methods as well as to gain access to this goldmine for the first time less than three years ago. Although the method and the database were never sold, it is widely believed that his models were successfully replicated by those wanting to use it. The company that made the most of these models was Strategic Communication Laboratories. The company began by purchasing personal data from a range of sources, like land registries and shopping data. Its offshoots around the world used the database for election work in Ukraine and Nigeria, and to whip up positive sentiment for the Nepalese monarch and NATO in different parts of the world. One of its offshoots Cambridge Analytica was involved in the social media campaigns of two of the most unexpected elections we have witnessed in recent years: Brexit and Trump.
Although the effect Cambridge Analytica had on the Brexit campaign is unknown, how crucial it was to the Trump campaign was explained by their CEO Alexander Nix at a conference. Every message that came out of the Trump campaign was data driven. The first and most basic idea of an election campaign- demographic messaging- was junked. Apparently, research revealed that stereotyping particular demographics and bombarding each one with the same message doesn’t work. Instead, the company’s model predicted the personality of every single adult in the United States! This was done by applying models based on behavioural psychology to big data. This process was followed by ad targeting- personalized advertising aligned as accurately to the personality of an individual as possible.
By the end of it, Trump’s team had 175,000 different ad variations. Every microscopic detail of these messages- headings, colours, captions- were fine-tuned according to what appealed to the individual. Trump’s inconsistencies and contradictions turned out to be his biggest asset, allowing him to deliver tailor-made messages to people. The campaign realised the disadvantage of television ads early- unlike social media, you had zero control over who would be on the receiving end. How the campaign targeted the Democratic voters is also fascinating. African-American voters saw videos of Hillary Clinton referring to black men as predators. People in Haiti saw videos about how the Clinton Foundation had cheated and failed them. Undecided voters leaning towards Hillary read articles about how democracy was a failed system and how there was no point in exercising one’s vote. Obviously, these were not posted from Trump’s official accounts or pages, but other sources.
The rest of course is history. A phenomenon such as this is humbling and scary at the same time. It demonstrates how insignificant and pliable we are, how simply we can be targeted, and how somebody can systematically shape our opinions without us having the slightest whiff. Big data combined with models based on behavioural psychology and followed up by ad targeting can literally manipulate the whole world. It can take advertising to a whole new level, build public sentiment, subvert democracy and spread any narrative that is backed up by enough money. On the other hand however, it did help us avert a Clinton presidency. And what if, back home, it allowed the BJP to put 2019 safely in the bag?
The BJP has always been ahead in the social media game. But tools like big data that Trump so effectively can make and break what the party has achieved in days.
Moreover, the sheer diversity of a country like India necessitates that big data be the logical next step in the social media arena. If implemented, what the party achieved in urban and semi-urban clusters in 2014, and what Prashant Kishore’s team or team 272+ achieved, will merely be a drop in the ocean come 2019. With Digital India and other schemes taking social media into the interiors of the country, the scope too would have widened immensely. Currently less than a third of the Indian population have access to the internet. Although the figure is dismal, it is more than enough to turn any election around.
With the BJP’s political stock on the rise, their rivals will look for novel ways to bring the party down. Before they employ the same highly scientific and tested methods that the likes of Trump used, the BJP must get onto the bandwagon. After all, it is better to be safe than sorry.
