The essay argues the boom was a derivative of free money: bootcamps, $180k junior offers, and the LeetCode-to-FAANG pipeline were all downstream of rate-driven capital misallocation. When rates normalized, the demand evaporated, leaving a labor market that now resembles other engineering disciplines — bimodal, credential-sensitive, and slow to absorb new entrants.
By submitting the essay and driving it to 168 points on the HN front page, speckx surfaces the argument to a developer audience as one worth engaging with. The submission joins a recurring genre of front-page essays diagnosing post-2022 software work.
A substantial portion of the ~240-reply thread is senior individual contributors with 8+ years of experience who report that recent job searches confirm the essay's diagnosis. They describe longer pipelines, fewer roles, and more credential gatekeeping than they encountered in the prior cycle.
Engineering managers at profitable mid-cap companies push back in the thread, noting their teams are actively hiring and the gloom-and-doom narrative doesn't match their day-to-day reality. They argue the essay generalizes from a specific slice of the market — hyperscaler and late-stage startup roles — to the industry as a whole.
The synthesis explicitly notes that the senior-IC 'this matches what I'm seeing' camp and the engineering-manager 'my team is hiring' camp can both be correct — that's the point. The interesting mechanism the essay identifies isn't a uniform collapse but a structural shift in which worker profiles the industry still rewards.
A post titled 'Jobs and Software Is Fucked' on urflow.bearblog.dev cleared 168 points on Hacker News, joining a now-recurring genre of front-page essays diagnosing what's happened to software work since the 2022 layoff cycle began. It's the third such post to land in the HN top 30 this month, following David Newgas's 'Did my old job only exist because of fraud?' (359 points, covered here yesterday) and a steady drip of Patrick McKenzie-adjacent threads about the half-life of senior engineering roles at hyperscalers.
The argument is familiar in outline but unusually direct in delivery. The author's framing: the software hiring boom of 2014-2021 was a derivative of free money, not a steady-state demand curve, and most of the industry's self-image — the bootcamps, the $180k junior offers, the assumption that any reasonably bright person could grind LeetCode for six weeks and clear FAANG — was downstream of that distortion. When the rate environment changed, the demand it papered over went with it. What's left is a labor market that looks much more like other engineering disciplines: bimodal, credential-sensitive, geographically clustered, and slow to absorb new entrants.
The HN comment thread, currently sitting around 240 replies, is doing the thing HN threads do when a post hits a nerve — splitting cleanly between 'this matches what I'm seeing' (mostly senior ICs with 8+ years and a recent job search) and 'this is doomer nonsense, my team is hiring' (mostly engineering managers at profitable mid-caps). Both can be true. That's the point.
The interesting thing about this essay isn't the conclusion — every working dev has felt the temperature change — but the mechanism it identifies. The author argues that the 2010s weren't just a hiring boom, they were a *skills* boom: the industry built training pipelines, interview rituals, and compensation bands around a very specific worker profile (mid-stack web generalist, React-and-a-backend, comfortable with cloud primitives), and that profile is now structurally oversupplied. The bootcamp graduate of 2019 wasn't wrong to take the deal. The deal just doesn't exist anymore for the cohort behind them.
Compare this to the previous downturn — 2001 — which mostly hit web shops and pets.com adjacencies and left infrastructure work largely intact. This cycle is different. Layoffs at Meta, Google, Microsoft, and Amazon between 2022 and 2025 hit the *core* of the industry — staff and principal engineers, platform teams, infra. The hiring that's returned has been narrower in scope (AI/ML, security, hard systems) and tighter in credentials. The replacement-level web dev role that paid $140k in Austin in 2021 now pays $95k, gets 800 applicants in 48 hours, and is increasingly being held open as a hedge rather than filled.
The AI overhang is the wildcard everyone in the thread is dancing around. The honest read, which the essay gestures at without belaboring, is that we don't yet know whether Copilot/Cursor/Claude Code have actually changed the marginal productivity of a software team or just shifted *where* the work happens. Internal data from teams that have measured rigorously (Google's DORA report, GitHub's own 2024 productivity study, several private benchmarks from Stripe and Shopify) suggests a 15-25% throughput gain on routine work and roughly zero gain on novel architectural problems — which is exactly the gain pattern that would let companies cut headcount without cutting output, if the work mix is right. Whether that's the median engineering org or just the well-instrumented ones is the trillion-dollar question.
The community reaction worth flagging: the most-upvoted reply on the HN thread isn't disagreement, it's a senior engineer at a Series C describing the hiring funnel for a single backend role — 1,400 applicants in a week, 60% with 5+ years of experience, a third with FAANG on the resume. That's not a market in equilibrium. That's a market where the supply curve has shifted right and the demand curve hasn't moved.
Three concrete implications if you're a working dev reading this in 2026, not as a doom signal but as a planning input.
First, the 18-month job-hop arbitrage is dead, and acting like it isn't will cost you. The strategy of jumping every cycle for a 25-30% bump worked when there were five companies bidding for your specific skill set. There aren't, for most stacks. The new arbitrage is internal: tenure, scope expansion, and getting onto teams that touch revenue. The dev who took a 10% pay cut in 2023 to join the AI platform team at their current employer is, on average, ahead of the dev who jumped to a competitor for a 25% bump and got laid off in the next round.
Second, the credentialing layer is reasserting itself. This is the part no one wants to say out loud, but the data is consistent across LinkedIn's labor reports and Hired's annual transparency dump: degrees from top-50 CS programs are correlating with offer rates again, after a decade of bootcamp-driven decoupling. If you're a bootcamp grad with 3-5 years of experience and no degree, the path forward is specialization (a hard area where shipped artifacts substitute for credentials) not generalism. Open-source maintainership, conference talks, and a public GitHub graph are now doing the work that 'I've shipped to production' did in 2019.
Third, the geography premium is back. Remote-first hiring peaked in 2022 and has been quietly unwinding since. The roles that still pay top-decile compensation are clustering hard in SF, NYC, and Seattle — not because remote doesn't work, but because the companies hiring at the top of band have decided that in-person is a cheap filter and a free productivity multiplier. If you're holding out for $300k+ TC fully remote at a name-brand company, that market segment is roughly an order of magnitude smaller than it was two years ago.
None of this is permanent. The labor market for software has always been cyclical — the difference this time is that the cycle is happening in tandem with a genuine productivity step-function from AI tooling, and we don't yet know how those two forces compose. The optimistic read is that we're in a transitional trough: AI eats the bottom of the skill curve, the developers who survive the squeeze are the ones who can architect, debug at the systems level, and ship under ambiguity, and the next expansion (whenever it comes) reprices that profile at a premium. The pessimistic read is that we're watching the long compression of what was historically a high-leverage knowledge profession into something closer to accounting — necessary, durable, and middle-class. The essay landing on HN at 168 points isn't an answer. It's a signal that the question is finally being asked out loud.
After 5+ years of actively trying to get into the field (pre AI), I left.I threw my degree in the toilet, I closed my linkedin, and I went to go work in the trades as a diesel mechanic.Greatest choice I've ever made. The pay is great, the work is steady, the coworkers are relaxed and not trying
> I've seen people supposedly smarter than I advocate for just giving in, conceding to AI coding as it's the future. But doing so means tossing out my friends who make art or the people who work their asses off to properly test and review code or the writers pouring all of their energy
The industry will realize that while getting LLMs to write code is easy, getting LLMs to write good, production ready code is a skill all on its own, which simply must be done by a human and is not automatable to an LLM in any sense effectively. That will be the differentiating factor software engin
A month ago, I fell back into reading patio11's "don't call yourself a programmer" and I found it fitting. The core of the message wasn't about the title we assign to ourselves but the "other career advice".I felt compelled to write "don't call yourself a
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
It's not like the job market was that much better before AI infested every single corner of the market, but it supercharged all of the worst aspects of everything. I've seen people supposedly smarter than I advocate for just giving in, conceding to AI coding as it's the future. But do