LinkedIn secretly experimented on 20 million users in job-mobility test

Some people might have missed out on job opportunities while the tech giant honed its recommendation features, in experiments that spanned five years
The LinkedIn app on iOS
Carl Court/Getty Images
Zi Wang4 October 2022

Do you feel like job hunting on LinkedIn got harder in recent years? It might not completely be your fault, because you might have been part of social experiments run by the company.

According to a recently published study, LinkedIn ran experiments on more than 20 million user accounts worldwide from 2015 to 2019, with the aim to improve people’s experience on the platform and help more people get more jobs.

However, that might have backfired for some users, according to the study’s results.

The experiments saw LinkedIn randomly vary the number of weak and strong connections it suggested to people via its People You May Know algorithm, which pushes new connections to users.

LinkedIn ran the tests based on the “strength of weak ties” sociology theory, which posits that people are more likely to get jobs through weak ties, or acquaintances, than through strong ones.

That means some people might have missed out on job opportunities because they weren’t recommended as many weak connections as others, while they were part of the experiments.

The effects, while seemingly minor, generated big differences for those on the winning end: after a year from their first connection, people who had received more recommendations to connect with “weak” contacts were twice as likely to score jobs at those contacts’ companies, compared to those that connected more with “strong” contacts.

In fact, researchers from LinkedIn, M.I.T, Stanford, and Harvard Business School agreed that “the weakest ties had the greatest impact on job mobility, whereas the strongest ties had the least”, based on their findings from analysing the data.

LinkedIn as an engine of job mobility

While LinkedIn didn’t specify where the accounts experimented on were based, according to its user base, the platform counts 33 million users in the UK as of April 2022, out of a global total of 850 million. That means 3.8 per cent, or 760,000, of users in the study might have been British.

The researchers further found that moderately weak ties (as measured by mutual connections) and the weakest ties (as measured by interaction intensity) “created the most job mobility”.

However, researchers also found that this varied by industry. Workers in digital industries, such as software engineering, saw more job mobility through weak ties, but for less digital industries, strong ties actually helped people get more jobs.

This makes sense, as workers in less digital industries are less likely to be on LinkedIn in the first place. Creative industries are also infamous for their nepotism problem, with the Arts Council writing that “if you’re not from a privileged background, you’re even less likely to succeed”.

Good intentions, mixed results

Sinan Aral, lead author of the study, told the New York Times that LinkedIn’s intention was to refine its algorithm to improve everyone’s job prospects, “rather than anointing some people to have social mobility and others to not”.

The experiments don’t appear to have violated any of LinkedIn’s policies, even though most users probably were not aware they were being experimented on.

These experiments, called A/B tests in tech lingo, are common among tech firms, which use them to try out different versions of their products to find those that resonate best with users.

But some experts have questioned whether the scale and impact of LinkedIn’s study goes beyond simple A/B testing.

Michael Zimmer, an associate professor of computer science and the director of the Center for Data, Ethics, and Society at Marquette University, told the New York Times that the findings pointed to some users receiving a notable leg-up in the job-hunting process.

“These are the kind of long-term consequences that need to be contemplated when we think of the ethics of engaging in this kind of big-data research,” he added.

LinkedIn has contacted The Standard since publication of this article to argue that its A/B testing procedures are not fairly characterised as experiments, albeit this is how they were described in the original study.

It also asserts that the word secret is misleading in its view because even though it did not seek explicit opt-in consent from individuals to participate, the firm believes it acted consistently with its own user agreement, privacy policy and member settings.

In particular, LinkedIn says that it was “transparent” with members through the research section of this user agreement. It has also clarified that 20 million members were involved in the study.

Create a FREE account to continue reading

eros

Registration is a free and easy way to support our journalism.

Join our community where you can: comment on stories; sign up to newsletters; enter competitions and access content on our app.

Your email address

Must be at least 6 characters, include an upper and lower case character and a number

You must be at least 18 years old to create an account

* Required fields

Already have an account? SIGN IN

By clicking Create Account you confirm that your data has been entered correctly and you have read and agree to our Terms of use , Cookie policy and Privacy policy .

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged in