Philosophy Majors Beat CS Majors on One Labor-Market Metric

Summary of an article first published in Techstrong IT on September 9, 2026. Read the full article.

The New York Federal Reserve’s latest data on recent graduates, aged 22 to 27, inverts a ranking that once made computer science the safe credential. Computer science now shows 7.0% unemployment and computer engineering 7.8%; philosophy, 5.1%; all majors, 4.2%. The previous release pointed the same way: computer science at 6.1%, philosophy at 3.2%.

The claim is narrow, and that is what makes it useful. Computer science still pays far more ($87,000 against $52,000 at the start of a career) and places far more graduates in jobs that use their degree. On a single metric, finding work at all, a major with almost no vocational pipeline now outperforms the most direct one. The shift arrived as generative coding tools entered daily practice.

What got cheap

Writing code to a specification got cheap. Writing the right specification did not. Checking the result grew harder, because there is far more output to check. Marco Argenti, Goldman Sachs’s chief information officer, put the risk plainly: generated code “can work well, but not do what you want it to do.” Code can be correct and still solve the wrong problem. Cheap implementation keeps that risk and multiplies its reach.

What stayed scarce is knowing what to build, stating it precisely enough to test, and judging whether the result meets it. A philosophy curriculum trains exactly that: defining terms, exposing hidden premises, testing arguments. The labor market is sending a signal nobody designed.

Three shifts for engineering leaders

The junior role. Entry-level work used to mean producing volume under senior review. Volume is now cheap and review heavier. A better junior role hands someone a bounded problem to own end to end, including confirming that it is solved.

The evaluation budget. Evaluation is usually budgeted as a share of build cost. When building gets cheaper and output rises, that share falls short, and the risk moves downstream, where fixes cost more.

Ownership of the goal. Engineers now make product decisions inside prompts; product managers make architecture decisions inside acceptance criteria, often informally and without review. Someone has to answer for what “correct” means, and someone has to hold the authority to declare a shipped feature a failure.

Where the numbers stop

The data does not show that philosophy is the better degree. Graduate students count as not employed, some computer science graduates hold out for the right role, and the period also saw a tech correction and a surge of graduates. The Fed’s own researchers attribute most of the recent rise in young-graduate unemployment to remote work, not to generative tools.

As a theory of education, the inversion proves little. As a description of the entry-level engineering job, it holds: the work changed, and hiring still describes the job it replaced.