For about five years, LinkedIn has conducted a survey involving more than 20 million users on its platform. The aim was to find out if people got more or less job opportunities when they connected with different individuals. However, the social network had to constantly fiddle with its algorithms, affecting contact recommendations. This may have hampered the chance of those involved to find a new job.
Using the A/B test format, the survey used each user’s circle of friends to determine their weight when looking for new jobs. If the person had more familiar or strange names on their list, would that affect their chances of securing the next job?
As such, the “People You May Know” algorithm has been changed for those surveyed, offering suggestions from either relevant contacts or complete strangers in every way. The tests were published in science.
However, none of the users was aware that they were participating in the studies.
Some experts were concerned about the way LinkedIn handled the process. This is because, since individuals were not aware of the changes in the algorithm, many may have associated themselves with people who were of little help in terms of the possibility of employment.
Michael Zimmer, director of the Center for Data, Ethics and Society at Marquette University, told the NYT:
Findings suggest that some users had better access to job opportunities or a significant difference in access to them.
Other experts expressed concern, as they believe this “raised questions about industry transparency and research oversight.”
In addition, the report indicated that it was possible that this affected financial and family support issues, as it impacted the chances of someone getting a spot in the job market during the tests.
LinkedIn disagrees with the experts
After the publication of the New York Times report, a spokesperson for the social network told the ARS Technica that “the company disputes this characterization of its study, no one was disadvantaged by the experiments”.
Karthik Rajkumar, one of the authors of the essay, said:
Our ‘People You May Know’ A/B test was for the purpose of improving the relevance of connection recommendations, not studying job results. Reports like the one from the NYT confuse the projection and observational nature of the data, making it look more like an experiment on people, which is inaccurate.
The platform stressed that it did not exclusively study just two options, but 7 treatments of variations of the algorithm. In the end, LinkedIn understood that if ties to third parties are weak, job mobility will be greater.