Why Software Eventually Rots and How to Beat the Clock
Software Doesn’t Rust—But It Definitely Rots
In this week’s Commit & Push, we sit down with economist with a background in software engineering Robin Hanson to probe a question most devs feel in their bones but rarely name out loud: Why do big codebases seem to decay no matter how many tickets we close? From the hidden cost of interdependencies to the looming impact of shrinking world populations, Hanson pushes us to zoom out and see software as just one example of a much larger economic story.
Software Rot Is Real—And It’s Not About Bits
Unlike houses eaten by termites, our code doesn’t lose electrons. The rot Hanson describes is structural: every new feature breeds more interdependent parts, making future changes exponentially harder. Eventually, the cost of another patch outweighs the cost of a rewrite—and we scrap the whole thing.
“Large systems that have been modified end up harder to modify,” Hanson reminds us. “Slowly over time, it gets harder to usefully change them, and we throw them away.”
Refactor, Replace, Or Let It Die
We developers love a heroic rewrite. Hanson is skeptical. Total replacement often fails because the live system keeps evolving while the green-field build lags behind. We counter with the “strangler” pattern—incrementally engulfing legacy modules until nothing old remains. Hanson concedes the tactic works, but only delays the inevitable. Very few systems, he argues, survive 40–50 years without a full reboot.
Key takeaway: Whether we nuke the monolith or strangle it, we’re paying the rot tax. The smart play is to time those payments to business reality, not engineering idealism.
Technical Debt vs Rot
Technical debt is the interest we owe for cutting corners today. Rot is the entropy that accrues even when we code clean. Debt can be paid down; rot can only be managed or outlived. That distinction matters, especially for startups racing to product-market fit. Sometimes it’s rational to pile on debt—and accept future rot—because survival today is worth more than perfection tomorrow.
The AI Surge: Threat Or 10x Multiplier?
Hanson’s economic lens reframes the “Will AI take my job?” panic:
- Elastic demand: Software isn’t a fixed pie like orchestral violin seats. Lower dev costs unlock millions of new automation opportunities.
- Rising wages, at first: As AI tools 10× individual output, the total value of software balloons, and experienced engineers capture a slice of that growth.
- Skill bar shift: Entry-level coding tasks will vanish first. To stay valuable, we’ll orchestrate AI agents, own architecture, and steward product strategy.
Short-term, we’re looking at a boom, not a bust. Long-term? That depends on forces outside our repos.
When Software Bites The World
Hanson flips Marc Andreessen’s famous line—“Software is eating the world”—on its head. Because software carries heavy fixed maintenance costs, a shrinking global economy could see decaying stacks “bite” back. As customers disappear and revenue shrinks, the funding required to keep elaborate dependency trees alive may evaporate, collapsing layers of tooling we now take for granted.
Population Decline And The Innovation Squeeze
Demographers expect the world population to peak roughly 30 years from now and then fall. Hanson’s model links fewer people to slower idea generation; fewer ideas mean slower economic growth—and less appetite for sprawling codebases that demand constant care. In that world, hardware gets built to last, languages stabilize, and only the most essential software survives.
Final Thoughts
Software rot is more than a refactoring chore; it’s an economic phenomenon tied to incentives, demographics, and the limits of complexity. To thrive, we need to:
- Recognize rot early: Monitor interdependency creep, not just code smells.
- Refactor with intent: Target modules that unlock modularity, and accept that buys time—not immortality.
- Leverage AI smartly: Use it to amplify senior judgment, not replace it.
- Align with business cycles: Invest in code health when revenue supports it; ship scrappy when survival demands it.
- Prepare for macro shifts: Population decline and slowing innovation could upend our current growth assumptions.
Great engineering isn’t just about elegant code; it’s about navigating the economics of change. Hanson’s warning is clear: ignore rot—or the larger forces that accelerate it—and the systems we build today may strangle us tomorrow.
About Robin Hanson
Robin Hanson is an associate professor of economics at George Mason University, with an early career spent developing systems for Lockheed and NASA. Author of The Age of Em and pioneer of “futarchy” (governance by prediction markets), Hanson explores how technology, incentives, and demography shape civilization’s trajectory.