Argues that junior engineering work — CRUD endpoints, glue code, tests — is the exact surface area where Claude, Copilot, and Cursor compress hours into minutes. A senior engineer paired with an AI agent now does the work of 1.5-2 juniors, and hiring managers have adjusted, hollowing out the entry-level rung first.
Pulled the NY Fed's Labor Market for Recent College Graduates series and found that for the first time in the dataset's history, recent grads (5.8%) are unemployed at a higher rate than the overall workforce (4.0%). The lines crossed in mid-2024 and the gap is widening through 2026, breaking a multi-decade pattern where the college premium showed up clearly in employment data.
Surfaced the Olson analysis to Hacker News where it drew 187 points and 203 comments, signaling that developers find the structural reversal credible and alarming rather than dismissing it as a one-off statistical artifact.
Olson's breakdown by major shows CS and computer engineering grads at 6-7% unemployment — higher than philosophy, history, and art history. Five years ago those same majors anchored the bottom of the unemployment table, making this the sharpest and fastest reversal in the dataset.
Randal Olson — the data scientist best known for plotting every dataset that crosses his desk — pulled the New York Fed's Labor Market for Recent College Graduates series and noticed something that hadn't happened in the dataset's history: the unemployment rate for recent US college graduates (ages 22-27) is now higher than the unemployment rate for the overall workforce.
The gap isn't a rounding error. Recent grads are sitting around 5.8% unemployment against an overall rate near 4.0%. For the entire post-2000 history of this series, the lines moved together in recessions and pulled apart in recoveries — but the grad line was always *below* the overall line. That's the whole point of going to college, statistically speaking. Olson's chart shows the lines crossing in mid-2024 and the gap *widening* through 2025 and into 2026. This is not a blip.
The breakdown by major is where it gets uncomfortable for this audience. Computer science and computer engineering grads now post some of the highest unemployment rates of any field — roughly 6-7%, ahead of philosophy, history, and yes, art history. Five years ago those same majors were at the bottom of the unemployment table. The reversal is sharper and faster than anything else in the dataset.
The usual response to a chart like this is to argue about whether college is "worth it." Skip that. The interesting question is: *what specifically broke for entry-level technical labor between 2022 and 2026?*
Three forces are stacking, and they're all visible in the hiring data if you look. First, the entry-level rung is the most automatable. Junior engineers spend the bulk of their first year writing CRUD endpoints, glue code, and tests — the exact surface area where Claude, Copilot, and Cursor compress hours into minutes. A senior engineer with an agent is genuinely doing the work of 1.5-2 juniors in many shops. Hiring managers have noticed. The job postings haven't disappeared; the *junior* postings have. LinkedIn's own data shows entry-level software roles down ~30% from 2022 peaks while senior roles are roughly flat.
Second, the 2021-2022 over-hire is still being digested. The big tech layoffs of 2023-2024 didn't reset the cohort — they pushed thousands of mid-level engineers into a market that was simultaneously freezing new-grad reqs. When a Stripe-trained L4 with three years of production experience will take an entry-level offer, the bootcamp grad doesn't get a callback. That dynamic doesn't reverse until either the senior supply dries up or growth restarts. Neither is happening yet.
Third, the offshore stack got good. The default playbook in 2018 was "hire two juniors in SF." In 2026, the same dollar buys a senior in Warsaw plus a Claude Pro seat. Companies that used to absorb green grads as a long-term bet are now buying competence on day one from a global market.
What's striking in Olson's chart is *which* majors held up. Nursing, education, civil engineering — fields with non-substitutable physical or licensed labor — are doing fine. Finance and economics grads are mid-pack. The fields hit hardest are the ones where the work product is text or code that a model can plausibly generate. That's not a coincidence; it's the entire thesis of the last three years of AI capex being validated in the labor market.
The community reaction on Hacker News (187 points, top of the front page) split roughly three ways. The first camp says this is cyclical — CS hiring always overcorrects, and 2027 will look fine. The second says it's structural and the four-year CS degree is now a worse ROI than a trade. The third, smaller camp pointed out the most defensible read of the data: the unemployment rate measures people *looking* for work, and a lot of recent CS grads are not unemployed — they're underemployed, working retail or gig work while they wait out the freeze. If you fold in the underemployment numbers (also tracked by the Fed), the picture for CS specifically gets meaningfully worse, not better.
If you're hiring, the math has flipped. A year ago, paying up for a senior was a defensive move. Now it's the only move that pencils, because the junior you would've hired in 2021 is being asked to deliver senior output on day one and most of them can't. Either restructure the role around an agent-augmented senior, or build an actual apprenticeship program — because the implicit "we'll train them" model is the thing that broke. Companies that figure out how to onboard new grads into agent-native workflows in 2026-2027 will own the next senior cohort cheaply.
If you're a senior engineer reading this: your leverage is up, but the floor under you isn't as solid as it looks. The same dynamic that's eating juniors works its way up. The defense is the same one that's always worked — own a system end-to-end, build judgment that doesn't fit in a prompt, and stay close to revenue. Pure coding throughput is not where the moat lives anymore.
If you're a recent grad or a student: the degree is necessary but no longer sufficient, and "GPA + leetcode" is a losing strategy in this market. The grads getting hired in 2026 have three things on their resume that didn't matter as much in 2019: a real internship (not a Big Tech name-brand one — *any* shipped production work), a public portfolio with commits people can read, and demonstrable fluency with the agentic tooling your future manager already uses. Hiring managers are screening for "can you ship with Claude" the way they used to screen for "can you use git."
The inversion in Olson's chart isn't a prediction — it's already a fact for the 2024 and 2025 graduating cohorts. The question for the next 18 months is whether it persists into 2027, at which point it stops being a labor-market blip and starts being the new equilibrium. The leading indicator to watch isn't tech hiring; it's CS enrollment. If incoming freshmen pull back, supply rebalances by 2030. If they don't — and current application data suggests they aren't — we're looking at a structural oversupply of entry-level software labor that the industry hasn't priced in yet.
This is the easiest niche to pick on but I am mid career for cybersecurity. I spend a decent amount of time trying to advise people away from this career field for college. So so so so so many people are going to college for cyber not realizing when they graduate, they are in totality unemployable.
I think the article is correct to point out remote work as a big culprit, but for the wrong reasons. The article says "Employers, the Fed argues, are wary of hiring inexperienced people into remote roles, where the on-the-job mentorship that turns a new grad into a productive worker is hard to
I wonder what the impact of the rising base rate of employees with college degrees is. In 1992, a fresh college graduate had better educational attainment than 42% of the labor force. In 2016 (latest date I found numbers for), that was down to 32%. https://www.bls.gov/spotlight/2
To me the real question is why a new grad would have lower unemployment to begin with compared to the average worker. Presumably the average worker has more on-the-job experience, so it seems like maybe people are weighting that more heavily now compared to before.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
The story here isn't college grads as much as young people in general. We are eating our young.We stopped building new housing, which turns housing into a transfer of wealth from those who don't have it (the young) to those who have been holding it (the not young).We have eliminated entry