The Empirical Case for 'Impractical' Majors

The unemployment myth collapses under scrutiny

I build software for a living, so I grew up inside the assumption that humanities degrees are a financial mistake. The Federal Reserve Bank of New York's data (2023-2025) says otherwise. Philosophy majors face just 3.2% unemployment, nearly half the 6.1% rate for computer science majors and well below computer engineering at 7.5%. Art history graduates do better still at 3.0%. History and English majors hover around 4.6%, competitive with most fields.

The American Academy of Arts & Sciences confirms it: humanities bachelor's holders see 5.2% unemployment versus 4.3% for all college graduates, a gap that narrows to 0.6 percentage points for those with advanced degrees. The numbers don't support the idea that studying philosophy or history condemns you to joblessness.

Salary trajectory is the more interesting story. Harvard economist David Deming's longitudinal analysis describes a "catch-up phenomenon." At ages 23-25, computer science and engineering graduates earn roughly $61,744 while history and social science majors start at $45,032, a 37% gap. By age 40, social science and history majors earn an average of $131,154 against $124,458 for CS and engineering graduates. The gap doesn't just close. It reverses.

Georgetown CEW's 2025 report finds that 14 of 19 humanities and arts majors lead to median earnings above $65,000 at prime working age, surpassing the 25th percentile of STEM earnings. PayScale data shows political science majors start at $54,000 and reach $90,000 mid-career, matching chemistry and exceeding accounting.

Most jobs don't actually require specific majors

Job requirements show another gap between perception and reality. Indeed Hiring Lab reports that 52% of U.S. job postings now mention no formal education requirement, up from 48% in 2019. Only 17.8% require a four-year degree or higher, down from 20.4% five years prior.

Federal Reserve research shows that only about one-fourth of college graduates work in jobs directly related to their major. For history and liberal arts graduates, that "non-vocational" education isn't a liability; it's the norm across every field. A Harvard Business School study found that in 9 out of 10 job postings requesting a bachelor's degree, the listed duties were identical to postings without a degree requirement.

The AAC&U employer survey (2023) of more than 1,000 executives and hiring managers gets at what employers actually want: 85% say "range of knowledge combined with specific skills" is ideal, while 15% prioritize specific skills alone. The survey states it plainly: "Field-specific knowledge and skills developed through the major are NOT what employers value most in new hires."

Technical skills depreciate faster than ever

The case for investing exclusively in technical training weakens when you look at skill obsolescence research. The half-life of technical knowledge has collapsed:

• 1920s: engineering knowledge half-life approximately 35 years
• 1960: approximately 10 years
• 1991: less than 5 years for engineers; less than 3 years for software engineers
• 2025: IBM reports 2.5 years for technical skills; AI-related skills as low as 2 years

The OECD documents technical skill lifespan dropping from 30 years in 1987 to roughly 2 years today. The World Economic Forum's 2025 Future of Jobs Report projects that 39% of core skills will become outdated by 2030, with two-thirds of workers needing retraining.

Chief Learning Officer Magazine frames it as the perishable-versus-durable divide. Technical skills (vendor-specific technologies, programming languages, organization-specific tools) have half-lives under 2.5 years. Durable skills like communication, critical thinking, creativity, and leadership keep half-lives past 7.5 years.

The programming language graveyard makes this concrete. AngularJS, once dominant, reached end-of-life in 2021. Objective-C has been largely supplanted by Swift. Perl has faded from mainstream use. Even COBOL, still running 90% of Fortune 500 backend systems, faces a skills crisis: 89% of large businesses worry about legacy staff shortages, with 63% of vacancies unfilled as the original programmers retire without replacements. The thing you learned to make yourself employable is the thing most likely to be obsolete before your loans are paid.

AI threatens routine technical work more than creative roles

The automation research generally favors humanities-adjacent skills, and the disruption to entry-level programming is already visible. Bureau of Labor Statistics data shows U.S. programmer employment fell 27.5% between 2023 and 2025. Entry-level tech postings dropped 60% between 2022 and 2024. Google, Meta, and Salesforce have cut or halted junior hiring.

GitHub Copilot now has 15 million users, a 400% increase in one year. Junior developers see the largest productivity gains, 35-39% speed increases, which is exactly what makes them replaceable. Stanford data shows employment for software developers aged 22-25 declined nearly 20% from the 2022 peak.

The skills that resist automation point the other way: empathy, ethical judgment, complex communication, creativity, cross-cultural competence, leadership. The MIT Sloan EPOCH Framework names five human capabilities AI cannot match, including ethical judgment and the one they simply call hope. McKinsey projects demand for social and emotional skills will grow 26% by 2030 while demand for basic cognitive skills declines 19-23%.

Personality and Social Psychology Bulletin research (2024-2025) rates creative skills "highly complementary to automation and unlikely to be replaced," with technical skills more vulnerable. Jobs requiring writing are 50% more likely to request liberal arts skills; problem-solving jobs are 40% more likely to value humanities backgrounds.

Steve Jobs' calligraphy story is true, with nuance

The claim that Steve Jobs credited a Reed College calligraphy class for Apple's typography is substantially verified. Professor Robert Palladino, a former Trappist monk who taught the class, confirmed teaching Jobs, noting that Jobs later "came back to Reed to tell me he was working on computers out of his parents' garage."

In his 2005 Stanford commencement speech, Jobs said: "I learned about serif and sans serif typefaces, about varying the amount of space between different letter combinations, about what makes great typography great... Ten years later, when we were designing the first Macintosh computer, it all came back to me."

The story needs context, though. The Xerox Alto (1973) already had high-resolution bitmapped displays, multiple proportionally spaced fonts, and WYSIWYG text editing, capabilities Jobs saw during his 1979 visit to Xerox PARC. As computer historian Ken Shirriff puts it, the calligraphy class likely gave Jobs "the aesthetic vocabulary to appreciate what he was seeing" rather than the sole inspiration. True about the personal influence, incomplete about where the technology came from.

Cognitive diversity drives measurable innovation gains

Research on team composition supports diverse educational backgrounds. Scott Page's "Diversity Trumps Ability" theorem (published in PNAS) shows that randomly selected cognitively diverse groups can outperform groups of the most individually skilled members on complex problems. High-ability people who think alike reach the same dead ends.

The business numbers track with this. Boston Consulting Group's 2018 survey of more than 1,700 companies found that organizations with above-average leadership diversity reported innovation revenue 19 percentage points higher than less diverse companies (45% vs. 26% of total revenue). McKinsey's 2023 analysis of 1,265 companies across 23 countries shows top-quartile diverse companies are 39% more likely to outperform financially.

The evidence at the level of a single team is thinner, but it points the same way. Alison Reynolds and David Lewis (Harvard Business Review, 2017) ran one strategy exercise with more than 100 groups of executives. Of the six teams they wrote up, the three with more cognitive diversity finished in time and the three with less were slower or never finished. Six teams is a small sample, and they called the result indicative.

Jobs put it directly: "Part of what made the Macintosh great was that the people working on it were musicians, poets and artists and zoologists and historians who also happened to be the best computer scientists in the world."

Tech ethics failures reveal the cost of narrow training

The cost of building without a humanistic perspective shows up in the wreckage. Internal Facebook research found that "64% of all extremist group joins are due to our recommendation tools." Former executives testified the platform used "Big Tobacco's playbook...to make our offering addictive" and that the system was "optimized for frictionless virality" of extreme content.

Amazon's AI recruiting tool, developed in 2014 and trained on 10 years of male-dominated resumes, was scrapped after it systematically discriminated against women, penalizing resumes containing the word "women's" and downgrading graduates of all-women's colleges. ProPublica found the COMPAS recidivism algorithm disproportionately classified Black defendants as more likely to reoffend. The Cambridge Analytica scandal harvested 87 million Facebook profiles without informed consent and drew a $5 billion FTC fine, the largest privacy penalty in U.S. history.

Safiya Noble's "Algorithms of Oppression" (2018), part of the work behind her 2021 MacArthur Fellowship, documented how Google search for "Black girls" returned pornographic results while "white girls" returned radically different content. She coined "algorithmic oppression" for the way systems "replicate the power structures" of society.

The most pointed evidence is about engineering education itself. Rice University's Erin Cech found that engineering students are less concerned about public welfare at the end of their degree than at the beginning, a decline that "does not rebound upon entering the workplace." Multiple studies confirm students' moral reasoning decreases over the course of engineering studies, even after ethics training. We are teaching people to build the machines and unteaching them the part where they ask whether they should.

Philosophy majors outperform on standardized reasoning tests

The strongest evidence that humanities education improves cognitive skill comes from standardized tests. Across 24 undergraduate majors, GRE scores put philosophy majors first in verbal reasoning and first in analytical writing. They rank 9th in quantitative reasoning, higher than business majors, every social science major, and every other humanities major. On cumulative rankings across the major tests, philosophy places first overall.

LSAT data (2022) puts philosophy second among all majors, behind only physics and math, ahead of pre-law and criminal justice. GMAT data has philosophy majors in the top five, scoring higher than management, finance, accounting, and marketing.

The question is whether high-performing students simply choose philosophy (selection) or whether philosophy training improves reasoning (treatment). Thomas Metcalf's 2021 analysis compared the incoming SAT/ACT scores of intended philosophy majors with their later GRE/LSAT scores and found the incoming scores aren't especially impressive. His conclusion: "A substantial portion of the effect is treatment." The major appears to do something to the mind that takes it.

What employers actually want contradicts the vocational narrative

America Succeeds' analysis of 82 million job postings found that 7 of the top 10 most requested skills are durable soft skills, requested 3.8 times more often than the top 5 technical skills. Leadership and communication appear in more than half of all listings.

AAC&U's 2013 employer survey was blunter: 93% of employers agreed that "a candidate's demonstrated capacity to think critically, communicate clearly, and solve complex problems is more important than their undergraduate major." Later AAC&U surveys find critical thinking near the top of what employers want, and near the top of where they think graduates fall short.

The NACE Job Outlook survey (2024-2025) confirms the order: problem-solving ability (90% of recruiters), teamwork (80%+), and communication (75%+) lead, with technical skills fourth. Use of GPA as a screening tool has dropped 35% in five years; only 38.3% of employers still use it.

Conclusion

Put the numbers next to the advice and they don't match. Humanities graduates reach competitive employment and catch up on salary by mid-career. Technical skill half-lives have fallen to 2-5 years while durable skills hold for decades. The automation that's hollowing out entry-level programming leaves intact the reasoning and communication that a humanities degree is built to train.

So the "safe" choice carries a risk nobody prices in. A narrow technical focus front-loads earnings and leaves you exposed to the next obsolescence cycle, and to whatever model is writing the code by then.

None of this is an argument against STEM. The evidence points at the person who can do both, the one who learned to build the machine and never stopped asking whether they should. I'd hire that person over the specialist who can only do the first half. The data, for once, agrees with me.