Argues the post isn't about being replaced but about becoming invisible: NDA-bound accomplishments no longer signal anything when an LLM can plausibly replicate them in a week. The actionable response is shifting work into public artifacts — open source, writing, demos — that build a portable trust budget independent of any employer.
The original essayist describes a mid-career engineer's dread: tasks they were paid handsomely for now take a junior with Claude an afternoon. They're not laid off, but the economic value of their specialized labor is eroding in ways that aren't visible from the outside.
The highest-voted discussion thread argues that knowing what to build, which tradeoffs matter, and when code is actually correct is a senior skill LLMs amplify rather than replace. Taste is positioned as the moat that survives commoditization of raw code production.
The counter-argument in the second-most-upvoted thread holds that 'taste' is a comforting story senior engineers tell themselves. As models get better at evaluating their own output and proposing architectures, the judgment layer is being compressed too, just on a longer timeline than code generation.
On June 7, a bearblog post titled *"LLMs are eroding my software engineering career and I don't know what to do"* hit 1,071 points on Hacker News, which is roughly the upper quartile of front-page outcomes for a personal essay with no benchmark, no demo, and no announcement attached. The author — mid-career, employed, not laid off — describes a quieter kind of dread: the work they used to be paid handsomely for now takes a junior plus Claude an afternoon, and the parts of the job that *do* still require them are increasingly invisible to anyone outside their current employer.
The comment section behaved the way HN comment sections behave when something hits a nerve: 1,400+ replies, the top thread arguing about "taste" as a durable skill, the second-top thread arguing that taste is also being eaten, and a long tail of engineers describing the exact same career texture in different industries. The post isn't really a complaint about job loss — it's a complaint about *legibility loss*, and that's a different problem with a different fix.
We wrote about this thread twice already today, once on junior-tier compression and once on the seniority-ladder break. There's a third mechanism in that comment section that nobody on Twitter has named cleanly yet, and it's the one practitioners can actually act on this week.
Most engineers have a resume that reads, essentially: *trust me*. Eight years at three companies. Senior IC. Owned the payments service. Reduced p99 latency 40%. The artifacts are all behind an NDA. Hiring managers used to take this on faith because the supply of credible engineers was constrained and the cost of a bad hire was a quarter, not a year.
That trust budget is collapsing for a specific reason: when an LLM can plausibly write a payments service in a week, the claim "I built the payments service" stops being a signal and starts being a baseline. The market response is predictable — interviewers are getting more skeptical, take-homes are getting longer, reference checks are getting deeper, and the gap between *what a candidate claims* and *what a hiring loop can verify* is becoming the rate-limiting step in the funnel.
The HN author doesn't quite say this, but their commenters do. The replies that get upvoted are the ones from engineers with *external evidence*: a maintainer of a library you've heard of, a writer with a blog older than the model they're worried about, a speaker whose 2019 talk still gets cited. The common thread isn't seniority or domain — it's that some fraction of their career output exists in a place where an employer didn't have to vouch for it. A GitHub graph, a paper, a Stack Overflow reputation that predates the dead-internet era, a talk recording. Things a model didn't write and a manager can't claw back.
This is not a new idea — patio11 was writing about "career capital you own" in 2012 — but it has gone from a nice-to-have for ambitious people to a structural requirement for ordinary ones. The reason is just supply and demand. The supply of engineers who can plausibly *describe* having built X has gone up (because LLMs raised the floor on what "plausibly describe" means). The supply of engineers with *verifiable* X has not. Compensation flows toward the scarcer signal.
There's a counter-argument worth steelmanning: most great engineers do not maintain OSS, and many of the best ones never will. True. But the counter-argument was stronger five years ago, when employer reputation laundering was load-bearing — Google on the resume meant something specific because Google's hiring loop did filtering work the market trusted. That filtering trust is itself getting commoditized; if a model can pass a Google loop, the loop stops being a moat for the candidates who passed it.
The actionable read is not "start a side project" — that advice has been bad for a decade because most side projects die in week three and the survivors mostly serve their author's portfolio rather than any user. The read is narrower: find the 5–10% of your existing job that could plausibly exist outside your company, and move it outside your company.
Concretely: the internal tool you wrote that solves a generic problem — extract the generic part, MIT-license it, put it on GitHub with a real README and one example. The post-mortem you wrote that has no proprietary information — sanitize the company name and publish it. The talk you gave at an internal eng all-hands — give it at a meetup. The unit-test pattern your team adopted — write 800 words about it. None of this is "build in public" performance art; it's converting work you already did from a private artifact into a public one. The cost is a few hours per artifact. The payoff is that next time you're on the market, the recruiter's screen contains things they can click on, and so does the LLM that's increasingly doing the first-pass screening on the other side.
The second move is more uncomfortable: assume any closed-source skill you're proud of will be table stakes in 24 months and price your time accordingly. If your current premium comes from being unusually good at writing CRUD endpoints, that premium is on a clock. If it comes from being the person who can debug a distributed-systems failure at 2 a.m. while the rest of the team panics, that's a different curve — but only if anyone outside your current employer knows you can do it. The work hasn't changed value; the *evidence* of the work has.
The HN post will roll off the front page in 12 hours and the panic in its replies will subside, but the underlying repricing won't. We're moving from a labor market where employer prestige was the dominant trust signal to one where verifiable individual output is, and the transition is going to be uneven and ugly — especially for mid-career engineers who optimized for the previous regime and have a decade of excellent work that exists only inside a former employer's monorepo. The engineers who notice this in 2026 and start converting private work into public artifacts at a steady, boring cadence will look, in 2028, like they got lucky. They didn't. They just read the comment section correctly.
I spent most of the last two years making a rather large hobby project. A 3D renderer for cad/visualization. It's pretty hard and slow work because getting anything wrong is usually hard to debug. Get a sign wrong and you have a black image suddenly, with painstaking debugging following. I
I see many comments saying, "AI can't do X with 80-100% accuracy; therefore our professions are in good hands."While I don't want to sound overly pessimistic, the models are improving at a rapid rate. If asked ~3 years ago where the state of the models are today, it would sound l
My career path is suprisingly similar to the author's. Weirdly enough, what he takes as the first pillar to fall is the one I see most undamaged currently.LLMs routinely fail at our business specifics: Local tax regulations, particularities of the accounting process, specifics of our ledger imp
I always remember of the infamous Steve Jobs quote "Ideas are cheap". If execution is everything, and frontier LLMs solve execution, then ideas are the gateway to abundance now, but abundance alone does not guarantee "stickiness".What I think is often overlooked is the human &quo
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
Wut? I pilot LLMs all day but there's no way in hell I'd agree to be at the helm of a finance product. That first pillar is still there. Maybe the author isn't aware of the impact they have, but I know, with the evidence of reverted PRs, that when I step outside my area of deep knowle