Elizabeth Sivak studied sociology for almost ten years. But then, in the middle of a research project, she suddenly realized that it is time once again to sit at the Desk.
Sivak is studying family and childhood at the National research University Higher school of Economics in Moscow. In 2015, she studied the movement of young people — in a series of interviews, respondents were asked to list the ten places they have visited in the past five days. A year later, she reviewed the data and felt the disappointment so limited it seemed to her method of personal interviews. The fact that a colleague told her about “the Copenhagen networks” (Copenhagen Networks Study), the innovative project, which tracked contacts in social networks, the demographics and the location of the about 1 000 students with five-minute granularity and for five months. Then she realized that the overdue cool change. “I realized that these new types of data will make a real revolution in the social Sciences, she says. — And thought: that’s great.”
After that Sivak decided to learn to program and join the revolution. Now they are with colleagues in computer science are studying huge and unmanageable amounts of data, understanding digital print companies. They monitor the activity of people on the Internet; study of digitized books and historical documents; interpret data from wearable sensors, which record every move and every contact; conduct online surveys and experiments, gathering millions of pieces of information; and research databases that are so large that companies reveal the secrets only after painstaking analysis.
The last decade researchers have taken the methods adopted to get to the bottom of things, and sociologists scratching their heads for more than a century: from the psychological foundations of human morality to the influence of misinformation and to the factors that make some artists more successful than others. For example, one study found widespread racism in the algorithms of decision-making in health; in the other data from the mobile phones were used for mapping the poor regions of Rwanda.
“The biggest achievement is the shift in the understanding of the digital behavioural data as an interesting and useful source,” says Marcus Stromeyer (Markus Strohmaier), a specialist in computational sociology from Institute of social Sciences name of Leibniz in Cologne, Germany.
But not everyone accepted this shift with alacrity. Some sociologists worry that the “registrars”, whose ambitions are as high as their databases are not familiar with previous studies. Another complaint: some calculators only look at the patterns do not take into account the reasons and make a valid conclusions from incomplete and dirty data — often obtained from social networks and other sources where hygiene is of the data leaves much to be desired.
But the exchange of “pleasantries” are mutual. So, the sociologists, and mathematicians who came from such areas as physics and engineering, I believe that many social theories are too nebulous, vaguely defined and are not tested by experience.
In the camp of sociologists sparked a kind of struggle for power, says mark Keuscnigg (Marc Keuschnigg), analytical sociologist Lincheping University in Swedish norrköping. “Who will win, and decide what will be the social Sciences.”
And yet the warring parties gradually reconciled. “Points of intersection with the computational sociology with more traditional, says Keuscnigg, noting a boom in joint journals, conferences and curricula. — Mutual respect is also increased”.
Computing revolution
In 2007, a small group of scientists with big ambitions convened a conference to discuss a new method of data analysis in the social Sciences. Their knowledge they wanted to change the world. The political scientist Gary king (Gary King) of Harvard University in Cambridge, Massachusetts, said that the flow of digital information “will enable much more to learn about our society and eventually begin to solve — to solve the main problems on the path to prosperity of mankind.”
By the time it was published several computational case studies. In the 2006 study examined the role of social influence on the popularity of the music for this was created an artificial online music market with 14 341 by the user. Participants downloaded a different song — sometimes knowing how popular they are, sometimes blindly. It turned out that popularity of a song to predict the harder it is, the more listeners influence each other. This partly explains why guess the sensation rarely obtained.
Two years later, another study analysed the displacement of 100 thousand mobile phone users over six months. It turned out that people in General follow a simple and repeatable route. The authors were able to calculate the probability of finding a particular person in a particular place and suggested that the detection of patterns will facilitate urban planning, the understanding of epidemics or to prepare for emergencies.
In the same year, technology magazine Wired published an article which argued that the era of “big data” will put an end to theory in all the Sciences. Although it was a barrage of criticism for the sketchiness and oversimplification, the article was spot on: it took more than ten years, and sociologists continue to refer to her as an example of what the relevance of the theory in question.
But data have continued its triumphant March. Duncan watts (Duncan Watts), sociologist at the University of Pennsylvania in Philadelphia, changes in the sociology recalled events in biology 1990-ies, when high methods generated a huge amount of data on DNA sequences and gene expression. “It was an avalanche of new data, and it required a completely different approach,” he says.
But many traditional sociologists first fruits of the revolution not impressed, and the methods it considered questionable. Skeptics believe the study of social networks kind of experiment on thousands of uninformed participants, not giving to the same his consent. In 2018 there is evidence that the British consulting firm “Cambridge Analytica” (Cambridge Analytica) collected data of millions of users of Facebook without their consent. This scandal and its reverberations gave rise to a certain skepticism to the study of social networks. Some scientists even had to turn research projects because of the platform implement new privacy policy.
Socially immature
At first it even was blamed for “toy” problems — some you can answer with data, but pressing problems of sociology, they do not solve. For example, how to deal with the inequality, or how to influence public opinion. “In the beginning was a lot of research on Twitter, which, I think, the sociologists received a hostile reception,” says Claudia Wagner (Claudia Wagner) sociologist-calculator from the Institute of social Sciences name of Leibniz.
Some argue that the fascination with “toy” problems were just a phase. With the increasing complexity of the analysis and the diversity of sources, these issues become increasingly important, for example, — the origins of discrimination, inequality and radicalization, says Stromeyer. “We have only recently begun to receive evidence by which can address important issues,” he says.
For example, last year the researchers of health and behavioral Economics analyzed medical records of more than 50 million Americans to appreciate a frequently used algorithm that determines which complex patients with special health needs receive the extra care or additional intervention. The team used simulations to show that the algorithm systematically discriminare black, and this affects the health of millions of people. The researchers then turned to the established differences in health to trace the sources of this bias and suggest ways to fix it. So, algorithms should not consider the amount spent on medical care of a patient is a good criterion of the necessary assistance: due to unequal access to medical care for black Americans typically spent less money than white Americans, even with the same needs.
But access to good data isn’t the only problem: scientists moving from physics or computer science, are often blamed for the neglect of sociological theories that explain human behavior. “They’re just looking for patterns,” says Julie Andrighetto (Giulia Andrighetto). She studied philosophy, but now a sociologist by the evaluator at the Institute of cognitive Sciences and technologies National research Council in Rome. “But the mechanisms that provoke this behavior, they are not looking for.”
This requires a solid understanding of sociological theory. Specialist in computational sociology from the University of Hamad bin Khalifa in Doha Czisny an ‘ he defended his doctoral thesis in 2010, when this area was just starting to gain momentum. Then she studied the exchange of news in social networks. At first she worked only with other scientists as”geeks” and understand in a strictly sociological questions was hard. Now she collaborates with scientists, studying the influence of the media on public opinion and Vice versa, and at the same time as encourage people to diversify their news sources. “Over time, each party understands the language and methods of the other,” says Anh.
Today there is concrete evidence of convergence. The first major conference that brings together both approaches, is scheduled for 2021. To bridge the gap, universities are creating a Department with representatives from different directions. So, at the George Mason University in Fairfax, Virginia, there is a special Department. Summer camps in computational sociology are held in more than 30 locations around the world, and many young students-enthusiasts (along with the increase of jobs) some hope that the struggle for power will be replaced by productive cooperation.
More communication
The Union of the two approaches can bear fruit. Data researcher Joshua Blumenstok (Joshua Blumenstock) from the University of Washington in Seattle and his colleagues concluded — according to the mobile phones of millions of Rwandans about their socio-economic status, and then confirmed the results by comparing them with data from conventional surveys. This method can use a policy to determine which areas of the country need help, and better track the results of those actions.
But lack of communication is still affected. A sociologist from Harvard, Joan Donovan (Joan Donovan) cites the example of last year’s study, where the authors outlined a network of “haters” on platforms such as “Facebook” and “Vkontakte” and showed how its structure has varied over time. She says the study authors, physicists and scientists computer scientists do not refer to the key works in the field of sociology, and the evaluation results came out poorer than I could. They examined too few social networks, even past studies have shown that the “haters” that follow a charismatic leader in many areas. She believes the team has come to a dangerous conclusion: social media platforms can guide the discussion in the “hate” groups, creating, in particular, fake profiles or pretending to internal strife. It is fraught with unpleasant consequences: so increase the volume of discussions within the group and increases its ranking in the search algorithms, she says. She believes that the best strategy is to contain the spread of hateful messages, limiting their visibility in search engine.
Physicist Neil Johnson (Neil Johnson) from George Washington University in Washington, D.C., and lead author of the study hatred have become accustomed to criticism from sociologists. He believes that led only the most relevant links. As for the search algorithms, the social networks can manipulate them, he said, — “exactly the same as they do now, drowning out the popularity of pages and groups with misinformation about vaccines and covid-19”. He studied misinformation, conflict and extremism and says he gets complaints whenever it publishes an article loud. However, his work resonated with politicians: he is frequently approached for advice organizations who like the quantitative nature of its work and the possibility to simulate the impact of certain interventions. “We can really look at specific issues in collaboration with other scientists, as had not been done,” he says. For his part, Johnson worried that sociologists too often seek to computational methods without having the proper training.
Sociology to solve the world’s problems
Some theories still more specific. Andrighetto studying social norms — shared rules defining what is acceptable and what is not. She says that researchers have spent over a decade on the drawing up of clear definitions and the development of theory. So, the theory assumes that social changes are changing people’s reactions to the situation. It is believed that social norms change slowly and by intense social interactions. Thanks to these verifiable claims Andrighetto combines computing experience with sociological theory: she conducts online experiments to test how the simulated changes in social norms affect behavior.
She is not alone in his desire to change the world through sociology. Watts said that he and other scientists are too often chasing publications, forgetting about the real solutions. “It seemed to me that the work is done when the article is coming out, he says. My objective is to spread ideas to the public, and someone else comes in and figures out how to implement them for the sake of change in the real world”.
So this shift happened, said watts, researchers from both camps to develop a thirst for collaboration. Some believe that this is already happening. “Traditional sociology and the sociology of computing are getting closer, says Wagner. — After 20 years, the gap not be at all”.
