A survey of 20 million LinkedIn customers over a 5-year period raised some eyebrows as they quietly analyzed people’s connection suggestions to see how it might affect their career paths.
The experiment, launched this month in the journal Science, sought to find out whether LinkedIn acquaintances known as “weak ties,” or immediate colleagues and friends known as “strong ties,” were best at discovering new people for people. alternative plan.
For many who thought the sturdy attachment could be eliminated with your follow-up gigs, the check found alternative was right.
“The authors propose that the weakest ties have the greatest impact on job mobility, while the strongest ties have the least impact on job mobility,” reads the paper’s abstract. “Taken together, these results help resolve the apparent ‘weak bond paradox’ and provide evidence for the power of the weak bond idea.”
Researchers from LinkedIn, Harvard Business School, and MIT conducted a number of large-scale randomized experiments using the “people you may know” feature of the social network for-thinkfluencer. This section of the website and app highlights profiles that people may find helpful based on their frequent pursuits and background, and suggest you join with him. Once relevant, you can message each other, see how your work is progressing, follow each other’s posts, recommend each other’s work, and more.
The experiment involved analyzing the 2 billion connections made by more than 20 million customers over a five-year period, and how those connections drove the acceptance of 600,000 new jobs.
Crucially, for this check, LinkedIn is rapidly adapting the “People You May Know” algorithm to provide certain clients with a suggested mix of strong and weak hyperlinks at random ranges, allowing researchers to examine and measure their performance How to influence future job adjustments. Think of it as an A/B check. Some clients have been shown to be scenario A, others are B, which yields a wealth of knowledge that collectively indicates whether a closed or unconnected connection is most important to the job seeker.
“Rational weak ties are the biggest,” said Shinan Aral, a professor of management at MIT and co-author of a brand new study mentioned in the opinion.
“Not the weakest, but a bit stronger than the weakest.” We learned that the constructive effect of weak ties on job search diminishes once a person has more than 10 frequent ties with another.
one way is moral
In addition to this conclusion, the examination now attracts consideration for other reasons. LinkedIn has been criticized for not informing customers about the experiment.
Michael Zimmer, associate professor of computer science and director of the Center for Knowledge, Ethics, and Society at Marquette College, said of The New York Times: “The results suggest that some clients have higher chances of getting job postings, or differences in access to jobs. Great post. These are the long-term implications that we have to consider after we consider the ethics of participating in this massive knowledge analysis.”
However, the creators of the check disagree that someone was screwed.
Karthik Rajkumar, a senior analytics scientist at LinkedIn and a co-author of the study, said: “No one is being disenfranchised when looking for a job.” MIT’s Aral added that the purpose of the analysis was to develop a More powerful algorithms to help people when looking for job help.
“Do an experiment with 20 million people and then, based on the information you study from that, come up with a stronger algorithm for everyone’s job prospects, and that’s what they’re trying to do, instead of making some people socially mobile, and no one else.”
In addition, please note LinkedIn’s Privacy Statement: “We use knowledge and public advice to analyze and improve our providers to provide you and others with better, more intuitive and personalized expertise, drive membership, and Committed to promoting “
That means you make your choice by handing all that knowledge to Microsoft’s LinkedIn.
Tech companies often conduct so-called A/B tests to find ways to expand engagement with services and products: some see model A, others see model B, after which the most-liked model is identified and used from there . Some publications – not us – publish A/B titles and determine the most clicked titles after a few minutes.
It has been determined that The New York Times conducts A/B exams of headlines and companies to investigate people’s habits without seeking explicit permission.
When a tech firm does it, it is a harmful social experimentWhen the NYT does it, it is an A/B check pic.twitter.com/l8sekLvELd
— Alec Stapp (@AlecStapp) September 25, 2022
Alex Stamos, current director of the Stanford Internet Observatory and former chief security officer at Facebook, write down This analysis was reviewed and authorized by MIT’s Institutional Review Board (IRB). ®
LinkedIn suggests job seekers have higher acquaintances • The Register