Signaling: Costly Proof and the Economics of Credibility

It is March 2004. You are sitting in a glass-walled office on the 14th floor of a mid-sized technology firm in Gangnam-gu, Seoul. On your desk: 200 resumes, all of them from applicants for the same junior software engineering position. Your company builds enterprise logistics software. The work is demanding, detail-oriented, and unforgiving of mediocrity.

You have read through the stack twice. The problem is not that the candidates look bad. The problem is that they all look identical. Same age range, same graduation year, same listed skills: Java, SQL, Linux, data structures. Most have done internships. Several list personal projects. A handful mention open-source contributions you cannot verify.

Then you notice something. Eleven of the resumes list degrees from what Koreans call the SKY universities: Seoul National University, Korea University, and Yonsei University. These are the three most selective universities in the country. Admission requires years of intensive preparation and performance in the grueling College Scholastic Ability Test. The acceptance rate at Seoul National University in the early 2000s hovered around 3 percent for competitive programs.

You have two options.

Option A: Invite the eleven SKY graduates directly to a final interview, skip the screening round, and extend an offer to the strongest candidate from that group. You save three weeks of assessment time and roughly 4 million Korean won in staff hours. The logic is straightforward: the Korean university admissions system already ran a brutal, multi-year filter. You are just reading its output.

Option B: Design a standardized two-day skills assessment and run all 200 candidates through it blind, with names and institutions redacted. You might find extraordinary engineers from Chungnam National University or Hanyang University who outperform every SKY graduate in the room. The logic here is also straightforward: past academic performance and current engineering ability are not the same thing.

You have 48 hours to decide before your director asks for a shortlist.

Sit with this for a moment. Both options have real costs and real benefits. Option A risks missing talent. Option B risks wasting resources on a process that produces noise rather than signal. The uncomfortable truth is that you are not really choosing between fairness and efficiency. You are choosing between two different theories of what information is worth trusting, and why.