A mid-career software engineer posted to Hacker News this week describing something harder to name than a layoff: their work is still there, but there is less of it, it pays less per unit, and the companies that used to fight over their skills are now staffing the same output with fewer people. No pink slip. No severance. Just a slow, structural thinning of the market they built their household's financial life around.
That post lit up the comments because it named a pattern a lot of families in tech are living but haven't quite articulated yet.
What's actually changing
The mechanism matters here. LLMs are not replacing software engineers the way factory automation replaced assembly workers — one robot, one fewer job, visible on a spreadsheet. What they are doing is collapsing the unit economics of certain coding tasks. A mid-level engineer who once spent two days writing boilerplate integration code can now do it in two hours. That sounds like a productivity gain, and it is — for the company. What it means for headcount is that a team that needed six people to ship a product now needs four. The two who remain are doing the same work. The two who were laid off didn't lose to a machine in a dramatic confrontation; the position simply wasn't backfilled.
Recent BLS data on software and IT occupations shows job posting volume declining even as productivity metrics at the firms posting those jobs hold flat or rise. That gap is the story. The industry is not shrinking. It is doing the same amount with fewer people.
For a household that runs on one or two tech salaries, the risk is not sudden unemployment. It is a labor market that is slowly but structurally less able to absorb mid-level workers at the rates that market had priced in for the past decade. Mortgage decisions, school choices, and retirement timelines that were built on those assumptions deserve a second look.
What we'd actually do
Map your income concentration. Sit down this week with a single piece of paper and write down what percentage of your household's gross income comes from one employer, one skill set, or one industry. If one number is above 70 percent, that is a concentration risk worth naming — not panicking about, but naming. A family with one software salary and a partner who does freelance work in a completely different sector is structurally more resilient than two software salaries, even if the total dollar amount is higher.
Identify the skills adjacent to your current work that LLMs cannot yet do cheaply. This is not about learning prompt engineering and calling it a pivot. It is about honest self-assessment. Tasks that require physical presence, regulatory accountability, long-term client relationships, or domain knowledge that is hard to codify in training data are holding their value better than tasks that are fundamentally text transformation. A software engineer who deeply understands a regulated industry — healthcare billing, aerospace certification, financial compliance — has a surface area that is harder to erode than one whose work is pure greenfield web development.
Build a three-month expense floor in cash, not in theory. The scenario to prepare for in a slowly eroding market is not a sudden catastrophic job loss — it is a six-month period of reduced contract rates, a longer-than-expected job search, or a forced transition into a lower-compensation role while retraining. Three months of actual expenses in a high-yield savings account is the buffer that lets you make those moves without selling investments at bad prices or carrying credit card debt. If you don't have it, set an automatic transfer today for whatever amount is not laughable.
Have one income-generating skill that does not depend on a software labor market. This does not need to be a second career. It can be a neighbor who pays you to manage their rental listing, a weekend skill that generates occasional cash, or a professional credential in a related but distinct field. The value is optionality, not income replacement.
The bigger picture
Every major shift in the labor market has produced a version of this post — a skilled worker in a field that felt permanent watching the floor shift under them. The response that has consistently worked is not denial and not catastrophism. It is building a household that can absorb a 20 to 30 percent income hit for a year without structural damage, and then making deliberate moves to diversify the skills and income streams that household runs on.
Software engineering is not going away. But the number of seats, the compensation curve, and the career ladder that a generation of families built their financial plans around is changing shape. The families who will be fine are the ones who see that clearly and adjust incrementally — not the ones who wait for a clean break that may never come.
