A Viral Essay Insists Coding Careers Aren’t Over — and the Hiring Data Agrees With Half of It
The most-discussed technology essay of the week wasn’t a model release or a funding round — it was a blog post arguing that coding has not been “solved”. Alex Ewerlöf, a veteran developer with two engineering degrees who describes himself as an early AI adopter, published “Coding Is Not Solved” on 26 September; two days later it was the top story on Hacker News, clearing 337 points. Its argument lands directly on one of the biggest career questions of the AI era: whether software engineering remains a career worth entering at all.
🔍 THE BOTTOM LINE
The essay is right that the demand for working software — and people accountable for it — hasn’t vanished; Indeed’s pay data and the stubborn senior-hiring numbers back that up. But the strongest evidence of the year says the strain is landing almost entirely on the first rung of the ladder, so “coding is not solved” is a true statement that doesn’t guarantee coding careers for the people who most need to hear it.
What the Essay Actually Claims
Ewerlöf’s target is the “coding is solved” narrative — the claim, associated with people like Anthropic’s Boris Cherny, that AI can now write production code and engineering is down to “taste.” His rebuttal rests on three arguments. First, creation is cheap but maintenance, reliability, security and scalability are “the majority of the cost” — the non-functional requirements that consume most real budgets. Second, most software that requires hiring has low risk tolerance: healthcare, finance, aviation, “wherever a mistake can cost money, lives or legal consequences you need accountability.” Third — his sharpest point — AI cannot be held accountable: “The worst thing you can do to AI is to unplug it… You cannot punish AI, therefore it can never be held accountable.”
He also documents his own AI-fluency credentials to pre-empt the “luddite” response — he built his own LLM harness and taught these topics — and his advice to developers cuts both ways: “AI is a bar raiser: if the quality of your output is equal or subpar to AI, upskill.” And to managers forcing AI into every workflow, he warns that AI overuse “directly puts an expiration date on your skill set.”
Where the Hiring Data Backs Him Up
If coding were truly solved, the market’s response would show up as collapsing demand and falling pay. Neither is happening. Indeed Hiring Lab’s September analysis found advertised pay in the most AI-exposed occupations up 46 percent since 2021, against 25 percent in the least-exposed — with software development among the fastest risers. Senior postings in those fields surged 14.7 percent year on year, and overall software development postings have rebounded 15 percent from their post-2022 lows. Employers are not behaving like an industry that believes the job is over.
That matches what employers tell universities. In AP’s reporting on the class of 2026, Tulane professor Nicholas Mattei described companies demanding graduates “jump into a middle management role as opposed to an introductory or junior developer role” — demand that exists only because the work still needs humans, a pattern we tracked when AI labs flattened their own job titles.
Where the Data Complicates Him
The essay’s weakest spot is distribution. “Coding is not solved” is accurate in aggregate, but the same Indeed dataset shows the entry-level share of salaried postings in AI-exposed occupations fell from 29 percent to 10 percent between 2021 and 2026, while entry-level ads dropped 7.5 percent year on year. Stanford’s Digital Economy Lab put young workers aged 22–25 in the most AI-exposed occupations roughly 19 percent behind their peers — and, in the AP analysis, recent computer science graduates face an unemployment rate around 7.1 percent even as philosophy and art history graduates do comparatively better. A separate Census working paper found the most AI-exposed majors facing a 13 percent earnings drop in their first quarter of work.
So the honest synthesis is uncomfortable for both camps. The optimists are right that software careers survive — the pessimists are right that the bottom rung is thinning. The Dallas Fed reached the same shape a different way: entry-level roles shrinking while experienced workers’ wages rise. And the counter-evidence keeps arriving too — a null-result study found the class of 2026 graduated into statistically normal conditions, a reminder that this debate is genuinely unsettled.
What This Means in New Zealand
The NZ software market is small enough that both effects show up at once. Hays’ FY26/27 salary guide, covered here on 18 September, found 82 percent of NZ employers reporting tech capability gaps — meaning senior engineers remain hard to hire — while only 22 percent of NZ tech professionals have had formal AI training. SEEK’s August data shows AI skills in 4.0 percent of job ads, up 93.3 percent annually, concentrated in ICT where 17.6 percent of ads mention AI. The senior squeeze and the junior squeeze are the same story seen from two ends.
For students — where New Zealand’s computer science enrolment data is thinner than the US’s — the practical read from Ewerlöf’s essay plus the hiring data is the one his own career illustrates: the market pays for demonstrated accountability, not tool use. The Computing Research Association put it bluntly in the AP story: if companies don’t invest in entry-level workers, “they’re not going to have mid-level career people in a few years.”
FAQ
Is coding really “not solved”? The essay argues no: maintenance, security and accountability — the majority of real software cost — remain human work. Hiring data supports the demand side; the debate is about the entry level, not the top.
Are software engineering jobs still worth training for? Yes at senior and mid level, where pay is rising fastest and postings are growing. At entry level the data shows postings down and exposure concentrated — new graduates are carrying most of the adjustment cost.
What did Hacker News make of the essay? It reached 337 points on 28 September 2026, with discussion focused on whether management mandates to “use AI” measure output or just token spend.
Does this apply to New Zealand? Partly. NZ employers report tech capability gaps and rising AI skill requirements in job ads, though local graduate-outcome data is thinner than the US research base.
🔍 THE BOTTOM LINE
“Coding is not solved” is the correct correction to a hype cycle that mistook code generation for software engineering. The catch the essay doesn’t quite land: a job market can value software engineers enormously and still have almost nothing to offer a 22-year-old holding a first degree — and that’s the combination the 2026 data actually shows.
📰 Sources
- Alex Ewerlöf — Coding Is Not Solved (26 September 2026); Hacker News discussion 337 points (28 September 2026)
- Indeed Hiring Lab — AI Exposure Isn’t Squeezing Advertised Pay (17 September 2026)
- Stanford Digital Economy Lab — Canaries in the Coal Mine (August 2026)
- AP News — Computer science grads face a brutal job market (24 September 2026)
- Hays Salary Guide FY26/27, New Zealand technology snapshot
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
