Published in


Source: Esquire Magazine.


An algorithm that assesses YOUR APPLICATION? Hans Lieman of Esquire Magazine put it to the test.

Article originally appears in Dutch within the October release of Esquire Magazine. English translation below completed by Seedlink. Link to original article appears at the end of this piece.

This is what the originally Chinese technology company, Seedlink which now has a Dutch owner, promises: an algorithmic system that makes it easier to hire or promote people. It does not look at gender, age or origin. Not even into diplomas or work experience?

All the kind of things that HR-employees usually cannot ignore. Instead, the algorithm maps out a person’s personality and calculates to what extent a person fits within a relevant department. In hip-business jargon: culture fit. To do that it does need a certain input, the language used among the talent pool.

Seedlink’s technology started development in China back in 2013. For the past one and a half years the company has been offering their services in the Netherlands. The US will be next. Large companies such as Ford, Coca Cola and L’Oreal are already clients. They use the tool to make selections from the sometimes thousands of job applicants who respond to a job offer, or to test if the right people are in the right place within the company. This is how it works: the client indicates which personality and characteristics are important. For example, a logistics company in Hong Kong thought it was important that its employees have a ‘digital mindset, to be creative and agile’, whatever that may mean.

Seedlink’s algorithm is firstly used on existing staff, especially on the best performing employees at different levels within the organization. After this process, potential employees go through the same digital interview. They all have to answer three open-ended questions. The algorithm looks at similarities in their language that indicate the personality traits most needed for success. Based on this, a ‘predictive score of success’ is calculated.

Before I visited the Seedlinks Dutch office for an interview with co-founder Rina Joosten-Rabou, I applied through their system for a fictitious job offer of Global HR Manager at Seedlink. I logged into my computer and received my first question: What would you do if you could travel backward or forward in time? ‘It does not really matter what we ask as long as it encourages you to express yourself freely and there is no right or wrong answer’, says Joosten-Rabou. One hour after answering I was sitting in front of her, reviewing my application. She opened her laptop. Next to my name there was written ‘Good news!’. ‘Rens has the potential to be a top performer.’ The algorithm gave me a score of 71 on a scale of 100. That is the chance that I will be successful as Global HR Manager and that I will thrive within Seedlinks corporate culture. Compared to Seedlinks sixty employees, I score best on creativity and worst on organizational talent. I said that I can recognize myself within this score.

Joosten-Rabou replied that she would invite me for a follow-up interview. (Whoever gets high scores from Seedlinks robot, usually goes through a regular job application process with the client — interviews, further assessments, and so on). Seedlink’s selection method is based on two theories. The first is that assessing personality traits can lead to a good cultural fit and good work performance. This is supported by American scientific research from 1998 by Schmidt and Hunter.

Although they also concluded that other selection methods, such as cognitive skill tests and tests for job knowledge, are actually much better predictors of job success than personality questionnaires. The second theory is that an algorithm can reliably discover personality traits in a person’s natural language. Anyone who has had a Freudian slip knows that.

For decades, scientific research has been conducted into what our language use says about us. The American social psychologist James Pennebaker in particular is active in this research area. In a 1999 study he speaks of a ‘linguistic fingerprint’: no two people talk or write the same. Just like with a fingerprint, you can hardly adjust your language for the occasion. Some aspects of that language use can also be linked to personality traits: Pennebaker and King see ‘small but meaningful’ correlations between the two. More correlations are found in more recent studies. For example, emotional instability is related to negative emotional words (‘terrible’, ’lazy’), extraversion with words about social events (‘drinking’, ‘dancing’, ‘crowd’) and openness with words about culture (‘poem’, ‘culture’, ‘literature’).

Extraverts generally talk in abstract and less concrete terms than introverted people (“What a fantastic match” vs. “With Ajax’s offensive game it became an exciting match”), who are usually more careful with conclusions (which you notice with words like ‘maybe’ in their language).

Despite the double meaning that words can have, advanced computer programs can find this kind of linguistic clues well in text, as is concluded in British research. However, it must be handled with care. For example, it makes a difference whether someone speaks or writes in a formal or informal way. In scientific literature, however, I can not find linguistic indications of that those vague management qualities such as being able to think digitally and to be flexible. Seedlink does not only rely on scientifically proven language-personality traits, says Joosten-Rabou. The system also measures similarities in language use — word choice, sentence structure, chosen form of person — between employees who, according to their superior or colleagues, have such a digital mindset.

The question that looms up the most: do people with a less broad vocabulary, but possess an agile way of working, miss out? ‘No’, says Joosten-Rabou. ‘Our system does not look at the beauty of language or grammar. Only the group that really has a hard time with writing a text don’t stand a chance, but that is also the case when sending out a traditional motivational letter.’ Joosten-Rabou wishes to do the ‘right thing with Seedlink’. ‘Human recruiters and HR managers will always have a prejudice’.

This is the reason that there are not that many females who have high functions within the corporate world. This is also the reason why a Dutch criminal offender will have an easier time getting a job than applicants with a migrant background who don’t have a criminal record (research from the Magazine of Criminology, 2017). An algorithm has no preconceptions. “Well, less, maybe. That algorithms are neutral is an illusion”. The algorithm is as neutral as the data which it is fed. This data is often colored. Seedlink uses data from ‘successful employees’ as a reference, while Seedlink’s clients determine who they are and thus what success means. And if there are unequal opportunities within the company because of sexism..? Joosten-Rabou understands my concerns. “That’s why,” she says, “we investigate the underlying behaviors of employees at different levels within the company”. She loves to let the numbers speak for themselves: since L’Oreal has been working with Seedlink, thirty percent of the newly hired employees have a background (diplomas, work experience) which wouldn’t have made it throughout the standard selection process. According to Seedlink’s own research, clients take forty percent more diverse candidates than they would have done beforehand. ‘The technology works’, says Joosten-Rabou. She doesn’t want to completely get rid of the human factor. Her idea is to create ‘an interplay between man and machine’. On the other hand, she would like to get rid of the resume.

— — —

This article originally appears in Esquire Magazine — Netherlands. Translations reprinted with permission.
Author: Rens Lieman (2018, 5th, September). Het Potentieel van een Top Performer. Translated on 1st of October 2018.

Original article:



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store