Do you enjoy going to the pool, love MAC cosmetics and sometimes enjoy spiral fries? Then you are a life-loving intellectual with homosexual inclinations. Interested in Kurt Cobain, Adidas Originals and biology? Then you are a heterosexual neurotic and you have many friends.
If, after reading this, you are now shaking your head disapprovingly, considering this prejudice, then you are mistaken. Because these are not prejudices, but exaggerated, it should be recognized, examples of correlations – connections between features, properties and states. This term is gaining popularity along with the development of the Big Data phenomenon.
The volume of data around the world almost doubles every two years, as does the performance of computers, so using a machine in data analysis, you can establish the relationship of almost everything that surrounds us and make predictions.
If, for example, a payment card company constantly evaluates the billions, if not trillions, of records of purchases made by its customers, then machine analysis can produce almost immediately startling predictions of this kind: this couple’s marriage is facing a serious crisis. , this man feels out of place, and that woman will become seriously ill within three months.
Selling digital footprints
The data of payment card users has always been extremely rich, and powerful computers and sophisticated algorithms allow data collectors to learn even more. Today, in the age of smartphones and tablets, sensors and cameras, digital footprints are everywhere. They can be bought, and moreover – often they are freely available. Their value lies in the ability to identify connections.
If you make the right combination of digital footprints, then the person who left them becomes an open book that data scientists can not only read, but even rewrite some chapters, imposing other people’s needs and opinions. Quite a few people believe that this is how Donald Trump won the US presidential election.
Stanford University researcher Michal Kosinski has combined big data with psychometrics, which studies the theory and methodology of psychological measurements. From 2010 to 2014 He served as Deputy Head of the Psychometric Center at the University of Cambridge. He has been studying correlations between digital footprints and personality profile since 2007, when he shared with his 150 friends on Facebook psychological questionnaires (https://discovermyprofile.com), which were recommended many times, and the author received information on thousands of people.
People simply enjoyed answering questions that resembled psychological tests in women’s magazines – it is no coincidence that two-thirds of the participants were women. Four years later, the number of survey participants reached six million, and today there are about eight million.
Testing is based on the theory of the so-called five-factor model of personality, which has been the standard in modern psychology for about thirty years. According to this theory, a person’s personality includes five common traits, the so-called “big five”: openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism – which manifest themselves in various forms. The first letters of the names of these traits in English form the acronym OCEAN, so this derivation of the model is also called the “ocean method”.
Tell us who your “like” is
Thus, the psychologist-philosopher Kosinski received hundreds of thousands of psychograms within a short period of time. Having prepared a sample of psychograms from 58,000 US test subjects, he began looking for correlations between test results and their Facebook profiles. In addition to the “big five”, to study the sample, subjects were asked or identified such characteristics as age, gender, sexual orientation, political preferences, drug use, intelligence, and, most valuablely, Facebook activities in the form of “Like” marks, to which the research team led by Kosinski was given access with the consent of the survey participants.
The links between the Big Five and the Likes proved to be extremely close. Having information about the “likes” of the subject, it was possible to draw unambiguous conclusions about the traits of his character and other signs that appeared in one way or another in the survey, and vice versa: according to certain signs, it was possible to judge “Like” marks, which was beneficial for the target audience. advertising campaign. 10 “likes” were enough to get to know a person better than usual colleagues at work. 70 is better than friends, 150 is better than close relatives. And if it was possible to analyze more than 300 “likes”, it was possible to get to know the subject better than his life partner.
Of course, the more values the Facebook activity analysis covers, the more accurate the conclusions. But there are also some particularly meaningful “likes”, the correlations of which can be difficult to trace in terms of the cause – such a typical phenomenon of big data.
This brings us back to the beginning of the article. If, for example, you liked a post about swimming, then most likely you are very satisfied with life. If you love spiral fries, then you are more than average likely to be erudite. This does not mean that you can become smarter by clicking “Like” under a post on the topic of french fries in the form of spirals. And even if they are. But if you not only “like” swimming posts, but also regularly visit the pool, then it is really quite likely that you are still satisfied with life.
Michal Kosinski is one of those scientists who warn that this kind of data can be dangerous if it falls into the wrong hands. In this regard, his theory about the big data aspects of Trump’s election seems plausible: Kosinski believes that a British company stole from him an “ocean model” that he developed for Facebook, and it was with its help that Donald Trump was able to reach to the minds of the voters. Founded in 2013, this sinister company goes by the name Cambridge Analytica (CA), a nod to its hiring from the University of Cambridge.
The Trigger of a Political Earthquake
Chip greeted Alexander Tyler, head of CA data analysis and processing, with reproaches. According to him, his company first heard about “this researcher Kosinski” in December 2016, when the Swiss magazine Das Magazin published his outrage. Tyler would later add:
“We have been in contact with some of the university staff from his faculty, but not directly with him.” The fact that Kosinski is unfamiliar with CA, despite the fact that he worked as deputy head of a small institute, whose services the company used not only in terms of personnel, but also, possibly, content, does not sound convincing to everyone. But one way or another, CA often does not clarify its position, rumors are not officially refuted, which fuels speculation.
It is not yet clear what effect the CA Brexit campaign had on the outcome of the election, but in any case, it seems that CA is enjoying the fact that it is being blamed for a political earthquake of such a class as the UK Brexit referendum and its results or Trump’s victory.
Tyler’s words that CA does not use Facebook data are also doubtful: “This is the property of Facebook.” Undeniably, but still the question is whether CA is using the user’s data – and even his unsuspecting friend on Facebook – when he opens his private profile to take a survey. In 2015, The Guardian newspaper also tried to ask CA, presenting the results of research, but did not achieve answers. In any case, CA refers to the existence of various sources: “The data we work with varies depending on the country.
In Russia, personal data can only be collected with a prior declaration of consent and for certain purposes. But in America it is different: we can license data from large aggregators – Acxiom, Experian or Aristotle. These companies collect all kinds of personal data that may be associated with, for example, loyalty programs, a payment card company, or magazine distribution. As early as 2013, Acxiom had detailed information on more than 800 million people.
European privacy regulations can be useful in fighting against the actions of companies like CA or Aristotle, which also has a history of political campaigning. But if, during a checkout, survey, or sweepstakes, a user clicks the “Agree to use my personal data for marketing purposes” button, then one should expect to be manipulated through email marketing or even political propaganda.
Psychological tricks and election victory
Without a doubt, CA contributed to the Donald Trump campaign with psychometric methods and digital tools. To do this, the company used the “ocean method” and “modeled the psychological profile of every adult American consumer,” Tyler told Chip. What exactly does this mean? “We have psychological portraits of hundreds of thousands of people that we have connected to our base dataset.”
This baseline covered all 220 million U.S. adults collected by major data mining companies such as the aggregators above and licensed by CA for future use. Thus, data and information about the lifestyle and consumer behavior of the population were obtained. This also includes the results of their own research. Then, using machine learning, the personal values of each citizen were calculated.
CA’s campaign for Trump proved incredibly effective, focusing on just 17 states where the election outcome was difficult to predict. And since the CA knew not only about the political preferences and many other sympathies of citizens, but also about their psychological essence, it could determine with amazing accuracy which messages to address to which people. A special application prompted Trump campaign volunteers the personality types of the residents of a particular building so that they could direct their conversation with the residents in the right direction. Volunteers recorded the feedback in the same application and sent it directly to CA in order to optimize the system.
The President of the United States as a tool for Big Data
To spread Trump’s messages, his team primarily used Facebook. The so-called “dark posts” were also used, when some ads only got into the feed of certain groups of people. Such micro-targeting involving dubious news, partly even of dubious reliability, is determined not at all by what the candidate stands for, but by what the potential voter wants to hear. Trump’s message was not one – there were many thousands of them, and they were aimed at completely different people. Trump has become a tool for Big Data.
The head of data analysis and processing at CA sees political advertising as nothing more than a product promotion campaign. “The issue is efficiency gains – that really takes center stage. It is necessary to convey the advertisement of the right content to the right person and thus express the message so that the person reacts to it.
As soon as you get on the Internet, you immediately find yourself under the guns of advertisers. This is very annoying, but you can approach advertising from the other side, considering it as a deal, even when it comes to free services. Most users agree to it because Orwell’s scenarios and the impact of advertising on consumers are two different things. As long as she doesn’t turn to the dark side, allowing data to be used for political manipulation, stigmatization and other forms of prejudice, little can go wrong. By this motto, in the digital universe, one could still live quite tolerably. But Trump’s victory made it clear that this conclusion is false.
Even if CA’s contribution to the blond construction mogul’s triumph is overstated, it must be admitted that the new US government has, at least indirectly, access to a database of psychograms of all citizens, which includes, among other things, confidential information such as sexual orientation and political and religious preferences.
“Alt-right” in the background
Stephen Bannon, the leading spokesman for the ultra-right movement in the US, the so-called “alt-right”, has been repeatedly accused of anti-Semitism and racism. He headed the Trump campaign, now holds the post of chief strategic adviser to the president and is even a permanent member of the US National Security Council. But that’s not all. It is he who is credited with an important control function in the board of directors of CA. Our request for confirmation from CA did not follow, neither did they deny it.
Bannon is credited with ideas such as creating a database of Muslims and developing a communications strategy for the president using big data. Who could bring them to life better than “his” company CA? The question is not entirely rhetorical: Palantir, whose main investor is the CIA, is also engaged in data analysis for political purposes.
First of all, she is connected with the special services: according to rumors, her predictive analytics helped the United States find where Osama bin Laden himself is hiding. Peter Thiel, a major German-born Palantir shareholder, has played an important role in the Trump transition team and has ties to him, not least because of his multi-million dollar campaign investment.
Thiel is Bannon’s opponent not only because of Palantir’s business interests. This billionaire at one time became the first external investor in Facebook, having received a membership on the board of directors of the company. Whether Facebook CEO Mark Zuckerberg could have become friends with Trump through Thiel is a separate question.
But Zuckerberg certainly wouldn’t want to see CA doing questionable government errands by dubiously getting Facebook data, or maybe even putting fake news into the system as conceived by Stephen Bannon, who plays the dual role of CA spokesman and Trump adviser. And no matter who wins, Trump remains under the influence of representatives of two companies who are ready to use confidential information of the population for political purposes.
The future of election campaigns
US consumer” –
Alexander Tyler,
Head of Analysis and Data Processing at Cambridge Analytica
But what about European countries? Is such microtargeting possible during election campaigns on the other side of the Atlantic? In principle, it is possible, says CA spokesman Alexander Tyler: “We would have to use different tools that were in the USA. But the rest, perhaps, in general, would not be much different. ”
However, the next US president may well do without the help of companies such as Cambridge Analytica. The largest social network Facebook stores on its servers the personal data of more than a billion users who themselves willingly share any private information. In addition, Facebook cooperates with Anxciom and other companies that collect and process data. At least from the point of view of Americans, this data does not belong to users, but to the social network Facebook, and if Mark Zuckerberg suddenly has such a desire, he can even secure the presidency of the United States in the elections in 2020.
Photo: Carlos Barria/Reuters, Tobias Hase/dpa/Global Look Press, Carlo Allegri/Reuters, Lauren Bamford/michalkosinski.com
