Karthik Rajkumar

Who they are

The only verified material on this page is the paper itself and a MIT Sloan summary of it. That summary describes Rajkumar and Guillaume Saint-Jacques as formerly of LinkedIn. The paper's own disclosure statement says that two of its authors are or were LinkedIn employees. The site does not hold his current position, training or earlier work, and this page does not guess at them.

The paper has five authors in this order: Karthik Rajkumar, Guillaume Saint-Jacques, Iavor Bojinov, Erik Brynjolfsson and Sinan Aral. It was published in Science, volume 377, issue 6612, 1304-1310, in September 2022.

Key papers

Rajkumar, K., Saint-Jacques, G., Bojinov, I., Brynjolfsson, E., & Aral, S. (2022). A causal test of the strength of weak ties. Science, 377(6612), 1304-1310.

The authors ran randomised experiments on LinkedIn's People You May Know algorithm to test whether weak ties increase job mobility. They report experimental causal evidence supporting the strength of weak ties, and they propose three revisions: the effect of tie strength is nonlinear, the measure of tie strength matters, and the effect varies by industry.

The study covered over 20 million members across five years, 2015 to 2019. The 2015 wave was a global experiment with over 4 million subjects that created over 19 million new connections. The 2019 wave covered over 16 million subjects, about 2 billion new connections and over 70 million job applications.

Key ideas

Moderately weak ties work best. Adding ties with more mutual connections raised the probability of job transmission up to a point, and ties with more than about ten friends in common reduced it.

The measure matters. When strength was counted by messaging interaction instead, the weakest ties raised job transmission the most and the strongest raised it the least, with a roughly flat relationship across the middle quartiles. This reverses the correlational results in the same data, in which stronger ties were associated with more transmission.

Industry matters. Weak ties produced more applications in digital sectors, measured by IT intensity, software use, suitability for machine learning and remote work. Strong ties produced more in less digital, less automated sectors.

Limits the authors state. Members could not be forced to act on suggestions, so the analysis is intent-to-treat and the exposed members were slightly younger and more active job seekers. LinkedIn is a professional network that may differ from other platforms. The 2019 wave was too short to measure long-term job transmission, so the industry analysis covers job applications, not hires. Individual-level data are not public.

---