On 2 July 1965, the day Title VII of the Civil Rights Act took effect, Duke Power made a change at its Dan River Steam Station in Draper, North Carolina. It abandoned the policy that had confined its Black employees to the Labor Department, where the highest-paying job paid less than the lowest-paying job in any of the four other departments at the plant.1 On the same day it added a new condition for anyone seeking placement outside Labor. Satisfactory scores on two aptitude tests, the Wonderlic Personnel Test and the Bennett Mechanical Comprehension Test, now sat alongside the high school diploma the company had begun requiring back in 1955. Neither test was directed at measuring the ability to learn any particular job, and the passing scores were set near the national median for high school graduates, stringent enough to screen out roughly half of all graduates.1

Six years later the Supreme Court took the case and ruled against the company, and Chief Justice Burger wrote a line that has outlived almost everything else about the dispute. Diplomas and tests are useful servants, he said, but Congress had mandated the common sense proposition that they are not to become masters of reality.1 The holding is the part everyone remembers. What I keep coming back to is the calendar. A written rule went away on the exact day the law required it to go away, and the sorting it had produced carried on producing itself, because the sorting had never lived in the written rule.

The line came out of the posting. Almost nothing else moved.

The degree had been standing in for capability long before any of that. It settled into management hiring decades ago and stayed there without anyone having to argue for it. What changed more recently is where it spread. Joseph Fuller and Manjari Raman went through more than 26 million job postings for a 2017 study and gave the pattern a name, degree inflation, meaning a credential demanded for work that had never called for one. They measured it as a gap between what postings asked for and what the people already doing the job actually held. In 2015, 67 percent of production supervisor postings asked for a college degree while 16 percent of working production supervisors had one. More than six million jobs sat inside that gap.2

Then came the correction, with nobody forcing it. Employers decided the degree was a lazy proxy, announcements followed, and the number of roles shedding degree requirements rose almost fourfold between 2014 and 2023. In February 2024 Fuller returned to the question. He and colleagues at the Burning Glass Institute and Harvard Business School's Project on Managing the Future of Work published what happened after the announcements. They matched more than eleven thousand roles where a requirement had been dropped against the career histories of the people actually hired into them. Removing the requirement raised the share of hires without a bachelor's degree by about 3.5 percentage points in the affected roles. Spread across the whole hiring market, that came to roughly 97,000 workers in a year out of something like 77 million hires, or, as the authors put it, not even one in seven hundred.3

The average conceals the interesting part. About 37 percent of the firms followed through, hiring roughly 18 percent more non-degreed people into the roles they had opened up. Around 45 percent removed the language and carried on hiring the same mix as before. The remaining fifth made early gains and then handed them back, finishing with a smaller share of non-degreed hires than when they started.3

A policy describes a decision. The decision itself happens somewhere else, with a person holding two resumes and a seat that has been open for months. Rewriting the description leaves that person untouched.

The group that gave it back

The researchers offer a plain read on that last cluster. Their sense is that initial executive enthusiasm never turned into changes in the underlying systems and practices, and whatever progress had been made came apart once leadership's attention moved elsewhere.3 I would put it more starkly. Executive attention was doing structural work in those companies. It was the one thing holding a new behavior against a screening habit nobody had touched, and when it wandered off, the habit was still sitting there intact.

You can watch the habit persist in the same people who removed the requirement. TestGorilla, which sells skills-assessment software and has an obvious interest in the answer, surveyed just over a thousand people involved in hiring decisions across the United States and the United Kingdom in April 2025. Fifty-three percent said their organization had eliminated degree requirements, up from 30 percent the year before. In that same survey, 41 percent said a degree matters more now than it did five years ago, and 67 percent said they still hire from resumes.4 A resume announces a degree whether the posting asked for one or not.

Find the room, not the document

I have sat on enough hiring panels across delivery and QA leadership to recognize the pressure that manufactures this. A req stays open while the work keeps arriving, and by the time a manager is choosing, the job description has stopped being the reference point. He is weighing one candidate against the others in the stack, and against how bad it will feel to be wrong. Under that pressure people reach for whatever lowers the felt risk fastest, and a familiar credential does that better than an unfamiliar one. The Burning Glass authors put it more mildly, that even when a degree is no longer required it remains a heuristic a hiring manager is likely to use to weigh a candidate's strengths.3

Which makes the useful question a physical one. Where does the selection actually occur, and what is in front of the person making it? In hiring that is the screen and the shortlist, and while resumes remain the input, the degree is in the room regardless of what the posting says. The general version shows up anywhere a leader announces something and treats the announcement as the change, the failure I wrote about when a company's AI program arrived as an email nobody's actual work changed to accommodate. It sits next to the habit of describing a dysfunction so fluently that describing it starts to feel like fixing it.

There is a reason to care beyond fairness. Among the firms that followed through, workers hired without degrees into roles that used to demand them stayed longer, with a two-year retention rate ten percentage points above their degree-holding colleagues, and they came in on an average salary increase of about 25 percent.3 Whatever that is worth, the companies that changed only the document never collected any of it.

Pick one policy your organization changed in the last eighteen months and go find the moment where the matching decision actually gets made. Who is in that moment, and what are they holding? If nothing in their hands has changed, you have a document, and it would be better to know that now than when somebody asks you how it is going.