Digital Fortunes iPad, iPad on the Wall

LinkedIn, the professional networking site, is testing an algorithm that will predict what job a person will have in five years' time. The head of Handelsblatt's companies and markets desk, currently based in New York, went to see what his future held.
But will I still be a photographic model in five years' time?

My future self sat in front of me and smiled. I smiled back, searching for words and something in common with this stranger sitting opposite me. I felt a strange connection – because he was, in some way, a version of me.

My awkward recent encounter with my future self in New York was part of an experiment by LinkedIn, the world’s largest professional networking site. I was meeting my “Future Me” – the person who has the kind of job I would very probably have in five years.

The person, or rather LinkedIn profile of a stranger, was selected after a data analysis of the social network site’s more than 332 million members. According to the data, his career path up until five years ago most closely matched my own.

I was persuaded to sign up for the experiment by Lutz Finger, a native German who heads LinkedIn’s data analysis team in California. When Mr. Finger talks about data, his eyes sparkle, and this helped to convince me to take part in the testing of “Future Me.”

It is not a ready-made tool at LinkedIn, so Mr. Finger and his team had to delve deeply into the data. The trick is to manipulate the data as little as possible, and not make too many constricting assumptions.

The job title doesn’t say anything. It is about the abilities needed for the job. Lutz Finger, Head of data analysis, LinkedIn

You have to “ask the right questions,” Mr. Finger said. And, above all, you have to know what you really want to know.

The first task was to filter a much smaller frame of reference from the vast number of profiles according to my professional past. That sounded logical because I’m a journalist and so am unlikely to be a doctor, pilot or lawyer in five years’ time.

But the analysis was not simply about discarding all LinkedIn members who are not journalists. Mr. Finger made more specific demands of the algorithm, such as: “The current position of the candidate is similar to a position which other members had five years ago.” And also: “The candidate and the members are all in the same place (New York).”

That troubled me. Wasn’t Mr. Finger assuming too much here?  Who says I want to stay in New York?

Well, the data says that. “Most members change jobs very locally,” Mr. Finger explained. Location is important, unlike skin color and gender, for example, which are not.

The next step of analysis was more difficult: Skills. At LinkedIn, members can enter their skills and other members can endorse them. “The job title doesn’t say anything,” said Mr. Finger. “It is about the abilities needed for the job.”

So the team analyzed my abilities and compared them with those of the 1,000 members left after the first filter. Not as easy as it sounds. “Just because you claim to be a great karaoke singer, doesn’t mean we would recommend you to open a karaoke bar in Japan,” said Mr. Finger.

This narrowed the field down to 500 members, whose profiles all matched mine “with regard to work history, training and abilities.”

Next Mr. Finger and his team applied specially developed criteria to filter out a “Top 50.” One of them was my “Future Me” – supposedly already further up the career ladder.

But other candidates were also selected who had changed industries. That was not so important for me, because the data showed that nearly half of journalists registered with LinkedIn change profession in the first three years. Someone like me, with 17 years in the job, is highly likely to stay in the industry.

Mr. Finger and his team also compared “fame,” by measuring how popular the profile of a person is. They also selected rather more unusual profiles, but which had still attracted a lot of attention.

At this point the importance of asking the right questions became clear – a computer algorithm can only analyze what it is told. Now humans had to take over.

How helpful would it be for businesses, for instance, to check the suitability of potential employees by using such an analysis?

Of the remaining 51 members, LinkedIn staff selected 10 final candidates. They sifted human characteristics to see which members might fit and which might not. Finally I received three possible variations of “Future Me” to choose from.

I didn’t find it difficult to decide on one – Raju Narisetti.

His career has been similar to mine, even if his was in the United States and mine mostly in Germany. He worked for years as a journalist at the Wall Street Journal. He is now with the business newspaper’s parent company, Rupert Murdoch’s News Corp, in charge of strategy and acquisitions. So while he didn’t change industries, he certainly changed his job.

Could I imagine such a step in five years? Why not? I am currently in charge of the companies and markets section at Handelsblatt, which is like the Wall Street Journal in Germany. Mr. Narisetti held a similar position at the WSJ years ago.

While talking with Mr. Narisetti, I noticed that I was looking for things in common with this American of Indian descent, who is a good bit shorter than me with less gray hair. The “Future Me” experiment subconsciously produced the feeling that there has to be a connection to him, because the data said so.

That is a kind of data credence that could be dangerous – because it’s based on the assumption that analyzing big data produces something real. Even Mr. Finger complains that we believe too much in the results of big data analyses.

But I do have a connection to Mr. Narisetti, if not a close one. Clearly we have similar qualities and preferences – for example, to create something over and above pure journalism. Mr. Narisetti set up an online newsroom when he was at the Wall Street Journal. I was in charge of the print newsroom at Handelsblatt. He likes working with a team, and so do I.

I don’t know if I will ever meet Mr. Narisetti again. For both of us, it was an interesting experience, but no more than that. And yet “Future Me” – a LinkedIn plaything for now – could be much more.

How helpful would it be for businesses, for instance, to check the suitability of potential employees by using such an analysis?

“Future Me” is a sign of just how quickly digitalization is changing the recruitment industry. It’s no longer just human beings who select candidates – computers, algorithms and data are increasingly involved.

LinkedIn’s experiment showed me just how carefully and responsibly we have to deal with that.


Grischa Brower-Rabinowitsch is the head of Handelsblatt's companies and markets section. To contact the author: [email protected]