Employment Strategy: Researchers propose models to maximize the success of professional recruitment

When it comes to hiring new staff, large companies often have to choose from hundreds of candidates. This is a time-consuming and resource-intensive process. Does mathematics help streamline these steps? Perhaps yes, at least in the broadest sense.

A paper published in the Journal of Statistical Mechanics: Theory and Experiment by Boston University statistical physicist Pavel Krapivsky proposes an algorithm that identifies three employment strategies, each of which corresponds to the different objectives a company may have.

Krapivsky was inspired by the famous “secretary problem” or “optimal marriage problem”. In one of its many versions, the princess must choose his future husband from a pool of 100 candidates at the epic reception. However, strict rules apply. She can only meet one suitor at a time, and has limited time to get to know him.

At the end of each encounter, she must immediately decide whether to accept or reject the suitor. She cannot revisit the previous candidates or ask them to wait for one of them while considering others. How does the princess want to make the best choice?

The secret lies in the numbers: 37, to be precise (42, raise your hand). “Dividing 100 by 2.718 gives you around 37 if this is the number of Euler, one of the most famous things in mathematical history,” explains Krapibsky.

In reality, this means that the princess must evaluate and reject the first 37 candidates while tracking the quality. Starting with candidate number 38, she must choose the first person who is better than everyone she has met before. According to Krapivsky, this strategy ensures the best possible outcome under given constraints.

This method is very reliable and it is rumoured that even Johannes Kepler used it to choose his second wife. “He studied the issues in great detail and spent a year doing this rather than his own great study and then made a choice,” says Krapibsky.

Krapivsky reformulated the problem in a more modern context and applied it to the employment practices of large corporations. The basic ideas remain the same. The company has a single parameter to assess the quality of candidates and must decide whether to hire them immediately or reject them without reconsidering them. Furthermore, this model does not allow you to reject newly hired employees.

“I don’t like firing people,” joked Krapivsky. Unlike the secretary issue, here the flow of candidates is continuous and potentially endless, making the model more realistic in modern workplaces where recruitment decisions are made based on immediate business needs.

This study explores three different employment strategies.

The Maximum Improvement Strategy (MIS) determines that a candidate will be hired only if it is higher than the score of a previously employed employee. Average Improvement Strategy (AIS) allows candidates to be hired if their score exceeds the average score of all current employees. Local Improvement Strategy (LIS) on the other hand is assessed by randomly selected employees or small employment committees, and involves each candidate who is hired only if the score exceeds the score of the interviewer or all committee members.

Unlike the best marriage issue, there is no one best strategy. Rather, choices depend on the company’s goals. If the goal is to maximize long-term quality, MIS is the best approach, but hiring will be slower. If the priority is to balance quality and employment speed, AIS is a reasonable compromise. When rapid employment is more important than quality, LIS is the most effective strategy.

“Of course, these are simplifications,” pointed out Krapibsky. For example, models such as those presented in the paper could serve as the basis for algorithms used in social networks and digital platforms.

These include platforms designed for job offers such as LinkedIn, as well as dating apps such as Tinder that coordinate future match proposals based on past “swipes” as well as those that manage content selection, resource management and artificial intelligence.

“Many of these are actually based on very simple algorithms, similar to those that suggest what you’re watching on YouTube,” concludes Krapivsky.