Big data and politics: The power of algorithms
In this new era of social networks
messages can be adjusted to each user’s taste.
This new form of communication filtered through algorithms
creates a political and moral challenge for communicators.
Should candidates tell each person what they want to hear?
How does one seduce citizens accustomed to messages
inviting them to be different from the rest?
How does one bring them together into a common project?
In a society divided into a myriad of individual identities,
how does one interest the majority in the country’s situation
or in understanding its economic policy?
The answers aren’t easy, but these are the questions
any politician faces in today’s world.
No more than 10% of the electorate in Latin America is informed.” That statement was made by the controversial Ecuadorean image adviser Jaime Durán Barba during an interview in the Argentinean newspaper La Nación. He added that traditional politicians “are afraid to talk about sex; they believe it’s in bad taste, that it’s not important. Up until now, one had to talk about Che Guevara.”
Durán Barba was a fundamental pillar in Mauricio Macri’s successful bids to become head of the Buenos Aires government in 2007 and President of Argentina in 2015. The consultant had by then already built his reputation among center-right candidates. He worked with the campaign that put Felipe Calderón in the presidency of Mexico in 2006 and was Ecuador’s secretary of public administration during the term of President Jamil Mahuad, who initiated the dollarization of Ecuador’s economy.
The anti-political political axis
As bad as this character may seem to those who dislike his style, he was able to steer Mauricio Macri’s Republican Proposal (PRO) ship to victory. While the previous governing party, Cristina Kirchner’s Front for Victory (FPV), talked about the State, country and people, Durán Barba, convinced that most people aren’t interested in politics, went for balloons, the “train of happiness” and smiles all around.
Of course, this political/anti-political axis on its own isn’t enough to explain Macri’s final victory, but it does encourage rethinking past campaigns along the lines of a phenomenon Durán Barba simplifies, but that had already been analyzed. “Social fragmentation is being extended to the point of making identities more and more specific, and thus harder to share,” summarized Spanish sociologist Manuel Castells back in 1996. Other authors agree: society seems more divided by consumption or lifestyles than by social class or political stances.
Modern individuals’ self-perception, above all in the 21st century, has been fragmented almost to a point of no return, as is already known by those who use marketing with specific messages to reach their potential clients. So, how does one seduce citizens accustomed to messages inviting them to be different from the rest? How does one bring them together into a common project? Should they be forced to take a political stand on key issues or must one talk about any issue that interest them?
From media to social networks
Since the rise of the press, radio, movies and television, along with the development of democracies and the need to attract more voters, politicians have understood that they have to rely on the mass media to gain access to the voters.
Like it or not, their message would be mediated by these media corporations, whose power only increased with time; it was fundamental that they feel comfortable within each one. The format of that type of dialogue isn’t some sort of controlled or even simulated interaction as much as it is one-way communication from a speaker to numerous recipients. Whatever is said has to be a catch all, because acceding to power requires seducing the majority of those recipients.
If they wanted to be understood, candidates first had to make themselves heard to establish the parameters of their discourse and to interest the public in, for example, what developmentalism is, how inflation is controlled or what taxes are for. In a way, not only was the political agenda constructed through these media but so were the cultural, economic, social, security and other agendas, thanks to a limited social dialogue that placed the debate within a certain framework.
What happens, however, when more and more people, above all the new generations, have access to communication from anywhere in the world in any format, at any time and on any screen? How does one catch their attention which, worse yet, is dispersed and loathe to be bored for a single minute? Social networks have revolutionized the way to reach this new type of individual, though in principle for publicity purposes.
Facebook and WhatsApp build and own the profiles of millions
It’s worth pausing here for a minute to understand how the business of corporations like Facebook works: it gathers large amounts of data—Big Data, as it’s called, from its users every day: their interests, places they go, their network of friends, frequent log-on times, institutions they belong to and much, much more.
With this information they create profiles that allow them to place ads selectively: diapers for mothers with infants, whiskey for those who love drinking, airline prices to wherever for those who visited a travel agency’s page... Facebook not only collects the information users put on their own pages, but also from all the sites in which a thumbs-up appears, indicating a “like.”
This company, like others, feeds off such data. That’s why it paid US$19 billion to acquire WhatsApp in 2014. This application, which was barely scraping by charging some users a dollar a year, was worth much more than consolidated multinationals. WhatsApp was a way to reach cellphones, the entryway to the intimacy of a rapidly growing portion of the global population.
In 2015, the number of mobile phones equaled the number of the planet’s inhabitants, though, of course, unequally distributed. Whoever could access that information could establish patterns and correlations and have detailed social information in real time, cropped to the needs of whoever uses it. Obviously, it’s not that there’s someone looking into people’s conversations—that’s called espionage—but computer programs can probe into the topics people are talking about, at what time, with whom, from where, etc.
The power of big data
Thanks to this monumental amount of data it has access to and the capacity to process them in real time, Facebook could be considered a publicity company with ongoing polls and surveys that allow it to prepare customized messages. It looks even better from a business standpoint when one considers that the same potential clients of these advertisements are the very ones who are also producing the contents that will keep other clients interested.
While other media have to pay producers, commentators, reporters, photographers, camera people, etc. to produce the contents that keep the flow of viewers active, social networks create a platform where the presumed users are doing the work.
Another fundamental difference with traditional media is that no one person is deciding whom to offer the diapers or whiskey. Instead the task is done by an algorithm, a program that analyzes the Big Data fed it to “learn” what to offer each person according to his/her specific interests. Those algorithms are even able to learn through trial and error how to improve their performance: if women of a certain location, age, educational level, etc. show interest in a certain product, it’s very likely that other women with the same profile would also be interested.
Facebook is one example of Big Data’s potential, but there are many more. Google Maps and other applications can tell us which road to take to avoid traffic jams thanks to those who keep the GPS on their cellphones turned on. With algorithms that learn from past happenings, they can predict what will happen now. If it makes a mistake, the algorithm will readjust to calculate better the next time. Credit cards cross data from all their clients with meteorological information to more adequately know consumers’ behavior on rainy days: is it better to launch certain sales offers when it’s raining?
Big Data can also have other social uses. For example, data journalism; putting large amounts of information like the “Panama Papers” in order, which allows for the detection of relevant cases or their presentation in an understandable way. Examples of new uses for Big Data multiply by the day.
A powerful tool
Martin Hilber, doctor in communication and a social networks specialist, explains in his article “Obama and Trump used Big Data for brainwashing”: “The availability of Big Data turned the Social Sciences, which were always ridiculed, into the richest sciences in data... We never had data, and that’s why public policies never worked. So, from one day to the next, 95% of the subjects we study got their own 24/7 sensor. Biologists always said, “that’s not science, there’s no data.” But they don’t know where the whales are in the ocean. Today we know not only where people are, but also their political ideas, what they buy and eat, when they sleep, who their friends are, how they live their social life...”
The huge numbers through Big Data allow statistical predictions of probable behavior and above all, learning from previous experiences. Information flows and new ways of processing it arise; from there, very detailed knowledge is learned about the population, from moods to consumption, including habits of movement and who their friends are. Whoever accesses this information and has the capacity to process it has a very powerful tool to influence the population.
What happens with this knowledge when it’s brought into the political arena? This question isn’t totally new. Just as for decades businesspeople did surveys to sell shampoo, for example, politicians started asking their consultants to measure society’s pulse.
With that supposedly representative, but actually biased and many times not very reliable information, candidates were able to record who their potential voters might be and carefully craft their messages to try to seduce them without scaring them away. How does one resolve this tension? The experience of the social networks provides the answer.
This is how Barack Obama won
In these times of social networks, some candidates grasped that surveys were slowing things down and, much like product marketing analysts, they needed to refine and improve their information. In a society divided into a myriad of identities (vegetarians, rockers, hippies, feminists, Catholics, surfers, athletes, hipsters, businessmen, pacifists, homosexuals, unclassifiables, etc.) and their innumerable nuances and crossover zones (the LGBT world is a good example), it ends up practically impossible to formulate messages able to seduce half the population. The social networks offer the answer to that problem.
The first one to take advantage of this resource on a large scale was Barack Obama, presidential candidate in 2008 in one of the countries where the use of digital tools in daily life is more developed. It also offers the particularity that voting is not only optional, but requires previous registration.
What Obama’s team did was classify the social network users by their political positions as revealed by their friends, to recognize 3.5 million potential Democratic voters who weren’t registered. They then studied their specific interests and customized proposals each one would see on Facebook: gender equality laws for feminists, green proposals for ecologists, withdrawal from Afghanistan for pacifists and so forth. This campaign’s level of precision ended up being greatly superior to the typical posters of smiling candidates who couldn’t say anything for fear of losing anyone who thinks differently. Instead of a “catch all”, what Obama did was more like a “catch each.” In the end, his team determined that at least a million of its targets had registered to vote.
Although it’s impossible to know the exact effectiveness of a digital campaign or who voted in the end, one can be generous with one’s assumptions. Obama won by less than 5 million votes in the country as a whole and with fewer than 70,000 votes more than his opponent in key states like Florida. He won the elections with the support of algorithms that signaled zones sensitive to his potential voters and indicated what would best seduce them.
It’s also how Donald Trump won
Donald Trump learned the lesson for his 2016 campaign. He also got profiles analyzed, but he did it with all eligible voters. This scale of work was significantly superior to any precedent. For this task, Trump hired Cambridge Analytica, a British company that had advised Senator Ted Cruz and also the campaign in favor of Brexit. With the information provided by Facebook, Twitter and also credit cards, supermarkets and all sorts of other databases, statistically reliable profiles were constructed for each citizen to detect who might vote for Trump.
One of this methodology’s findings was the discovery of a vein of potential voters in the “Rust belt” of Michigan or Wisconsin. A politician with a good eye might have been able to see these former factory workers in this sector, currently unemployed and expelled from the American dream, as potential Trump voters, but it was Big Data that in fact did it.
The use of Big Data allowed precision detection of those who fit this profile, their level of frustration with the system, their rejection of immigrants or their disappointment with the establishment governing class of recent years, which didn’t seem to take them into account. In reality, the hypotheses about why this was so weren’t nearly as important as finding messages that would speak to these different profiles to generate “likes” or share re-tweets.
A brutal blow to the mass media
How did he do that? Hilbert gave this example: “If Trump says “I defend the right to bear arms,” that phrase gives people known to be rather fearful the image of a criminal entering a house. The same phrase would give other, more “patriotic” people the image of a guy going hunting with his son. It’s the same phrase from Trump, but with two versions. They created messages for 175,000 profiles that effectively brainwash you. It has nothing to do with democracy. It’s pure populism, in which they tell you precisely what you want to hear.
The power of personalized communication guided by algorithms in a society where a good part of life goes on through networks is enormous: WhatsApp, Facebook, browsers such as Google, email, maps, tweets, Instagram, Uber, on-line buying, omnipresent cellphones, etc. allow a permanent monitoring of society and this information is tremendously valuable for whoever has access to it and can process it.
Trump and his team found a way to use that power and dealt such an unexpected and brutal blow to the US mass media’s self-esteem that most of them turned against the Republican candidate. Their rejection was so strong that Trump described them as an “opposition party.”
The 2.0 corporations, i.e. those that “use emergent social software platforms within companies, or between companies and their partners or customers,” also rejected businessman Trump’s candidacy but in addition to defeat suffered the blame for being an inexhaustible source of key information for his campaign. In fact, part of journalism and some intellectual sectors accused them of being guilty of the defeat because a large number of lies circulated through these networks aimed at mobilizing Republican voters. The magnitude of false news was so great that the Trump campaign was framed within the logic of “post-truth politics,” a term coined to describe a reality marked by what’s credible to some even though concrete facts contradict it.
Common projects are more and more difficult
This kind of digital campaign is possible in countries like the US, where social networks are the second largest news source. According to the Pew Research Center, 38% of the population is informed by these networks, second only to the television. This allows for significant manipulation of the communication to which the public has access, thanks to algorithms and paid messages directed specifically to their users.
If reality is perceived mainly through these networks and the messages we see are filtered through algorithms, two neighbors could live in totally different realities. Algorithms simply do the task of keeping whoever is looking at the screen interested and does it based on what interested them previously.
So, the fragmentation process that started some time back is deepened more and more and society itself is transformed into an accumulation of stagnant compartments, where dialogue outside of them becomes a more and more unusual phenomenon. A reality built for each person with a credible customized horizon poses enormous difficulties for a common project.
Latin America’s heading that way
As pointed out above, this case of the hyper-technological US society may not be representative of what happens in Latin America, but everything suggests we’re heading in that direction.
There is increasing evidence that the rumors about a digital communication team linked to the national government in Argentina that specializes in “building reality” in the social networks, mostly Twitter, are true. The weight of campaigns like this is relative and affects only those who inform themselves mainly through social networks. This is one of the first symptoms of the increasing weight of these networks in the construction of the social self-image and of the search to control these spaces. In such cases, social networks aren’t used only to take society’s pulse, but also to manipulate it and favor certain readings. It’s hard to know to what degree they achieve this, but surely their actions aren’t harmless.
What will happen to politics if data proves that the population simply isn’t interested in politics? In Argentina, PRO has been very skillful at exploiting specific interests and promoting them. Durán Barba frequently comments on the results of focus groups, meetings with citizens representing certain sectors of the population that are used to take society’s pulse and detect interests. This type of information would explain why science, education, health or citizen’s safety programs are getting cut, while campaigns are permanently mounted to renovate plazas or resources are used for communications that seem tangential.
An example is PRO’s insistence on pets. Shortly before winning the elections President Macri himself published pictures of his dog, “Balcarce10”—named for the street in Buenos Aires where the Presidential House is located—sitting in the presidential chair. In February 2017 pets were authorized to travel in the underground transportation in Buenos Aires. During Macri’s term as head of the Buenos Aires government, its website included a section dedicated to domestic animals.
In the same way, each sector of society can be addressed with a specific measure that in earlier times would have seemed like a deviation from the pillars upon which politics and traditional government are built.
A question every politician should think about
The power of this kind of communication could have short legs: a reality lies beneath the communication upon which stories can be woven that is very dangerous to deny.
The experience of Latin America’s recent past indicates that when no democratic channels exist for dealing with people’s real needs, pressure builds up and can end in violence and death. It’s also true that politics continue to be a powerful bond, even in the face of diversity, as has been seen in Argentina with the massive demonstrations backing the teachers or on the anniversary of the coup in March 2017.
On their own, they may not be enough to win an election, but they’re not irrelevant. On the government side, as public opinion analyst Rosendo Fraga pointed out, the march “in defense of democracy” organized by government sympathizers, allowed “Macrism” to briefly retake the initiative in the streets, a political space that usually negates it.
To simplify the discussion: if there really are people more interested in pets than politics, is it bad to take measures that address their interests? Should a good politician seek to get people interested in the country’s situation or to understand an economic policy instead? Could they do so and still win elections today? The answer isn’t easy, but it could be political suicidal not even to consider it.
Esteban Magnani is a journalist specializing in communication. This article first appeared in the May-June 2017 issue of Nueva Sociedad. Edited by envío.