Puppets in the web

A study conducted by the social network Facebook showed that users empathize with each other. At the same time, now we know that Facebook is able to imperceptibly manipulate the mood of billions of people

This week it became known that the Facebook social network deliberately manipulated almost 700,000 of its users for the purpose of a scientific experiment. From the article published in the PNAS scientific journal, it turned out that for one week (from January 11 to 18) 2012, several hundred thousand English -language users of the Facebook social network, without knowing it, acted as experimental rabbits: the number of friends (News Fed) was automatically reduced by the number of messages containing words in one case with the negative, and in the other with the other with the other with the other with the other with the other with the other positive coloring. The aim of the study was to find out how this would affect the publications, and therefore, possibly, the “mood” of the tested.

It is curious that some earlier studies demonstrated the reverse dependence of user moods on the emotional coloring of what they observed in their tape friends: people whose friends wrote about good, more often published negative messages. From the point of view of philistine psychology, this phenomenon is easily explained, because it is annoying when others are too good (unfortunately, the opposite is often true). However, the large-scale experience conducted by the research department (Core Data Science Team) Facebook together with scientists from the University of California at San Francisco and Cornell University showed that users of the largest social network are more likely to empathize than envy. In order to establish this, four groups of about 150,000 users in each were randomly selected. For the first of them, an algorithm that forms a tape of friends (in a normal situation, this algorithm forms a tape from messages that are assessed as the most relevant to the current activity and interests of the user) was adjusted so that from 10 to 90 percent of messages containing words that have a positive coloring did not fall into the tape. For the second group, a negative background of messages was artificially reduced artificially. The remaining two groups became controlled - from the tapes of these users, messages were remote randomly, regardless of their emotional coloring.

The experiment showed that users from the first (sad) group began to use positive words and more often negative in their own messages, and vice versa, a good mood of virtual friends forced users more often to write about good and less often about bad. Both effects manifested themselves with high statistical reliability, however, they themselves turned out to be insignificant: the frequency of using positive/negative words changed compared to the control group only to tenths of a percent. The authors of the study declared this the first convincing confirmation of the ability of people to “infect” from each other a mood, not only with personal communication, but also on social networks on the Internet. In fact, the results of the study can be interpreted differently. Firstly, the presence of positive words in the recording does not say anything about the actual emotional coloring of the message (for example, “excellent! I am surrounded by beautiful-minded idiots”), and secondly, the observed effect may indicate not so much about empathy as social mimicry, fear of fear among funny friends with a sad white broth (and vice versa).

For sociologists, the results of the experiment can be important, first of all, thanks to an unprecedentedly huge sample, hundreds and even thousands of times the sample of a typical sociological study. But ordinary users outraged (possible) participation in a scientific experiment as experimental experiment without any warning from the social network.

In fact, the researchers from Facebook, apparently, did not make anything illegal. According to the text of the user agreement, which is inevitably accepted by any person registered in Facebook, the network can use the collected information about the user “for internal operations, including the elimination of errors, data analysis, tests, research and improvement of services”. The work of the Ranking algorithm, which forms a tape of friends, has not been regulated at all, in the article in PNAS, for example, it is only said about it that “[this] algorithm is constantly improving and tested by the Facebook command in the interests of demonstrating to the user of the content that will be the most relevant and involved for them.” Nevertheless, in American legislation (namely, Facebook submits for its jurisdiction) for experiments in public (Human Subject Research) there is a fairly tough regulation. It could be applied to an experience conducted by a social network, because the adjustment of the Ranking algorithm can be considered manipulation over the natural environment of a person , and such actions fall under the formal definition of the experiment. But the legislation regulates not all experiments in humans, but only those conducted by organizations funded by the state body or agency included in a specific list, or if the results of these experiments are the basis for contacting the sanitary supervision of the quality of food and US medicines, as in the case, for example, testing of drugs. Facebook does not test the medicine and does not receive money from state agencies, and other co -authors of the articles representing truly federally funded Cornell and California universities were just “helping to determine the general research format and prepare publication”, as directly stated in the text of the article.

Although formally Facebook was not violated by the law, the leading author of the study by Adam Craimer considered it necessary to answer the criticism that fell on a social network and apologize for the recording on his Facebook page.

“We conduct any research on Facebook only in order to understand how to improve our service. [...] I assure us that we did not want to upset anyone at all. I understand that some of this might seem doubtful. My co -authors are apologizing for the form in which the study was presented in the article and for this anxiety,” Craimer wrote.

However, in the same record there is a hint that the motivation of the network could somewhat go beyond altruism and scientific curiosity. “It seemed to us to figure out whether, watching how friends publish positive content, people experience negative emotions or feel lonely. At the same time, we were worried that the negative attitude of friends leads to the fact that people begin to use Facebook less often,” Craimer explains. Of course, as the most frequent visit by users of the social network is a direct commercial interest to Facebook. The experiment showed that not only we can affect the emotions of friends on the Internet with our statuses, but also the social network itself, adjusting the Ranking algorithm for friends, can (albeit relatively insignificant) manipulate the mood of hundreds of thousands, and potentially billions of users. But, changing the algorithm, you can regulate not only the level of abstract negative and positive records.

Whether we like it or not, social networks are more and more and more affect real social processes. A scientific understanding of how and why this happens, it is necessary to think about how and in whose interest the results of such studies can be used. The Internet is a free environment, but it is quite easy to manipulate in it. It is simple to protect against this: do not forget about reality and use social networks less.