Technical debt is not a moral failing. It is financing. Shipping the simpler version now to learn what customers actually want is often the correct trade, exactly like a loan that gets a business open before it could pay cash. The failure mode is not borrowing. It is borrowing without tracking the balance or the rate.
Know your interest rate
Debt service in software looks like this: features that used to take a week now take three, because every change fights the shortcut. Deploys that need the one engineer who remembers the landmines. Bug fixes that break something adjacent, because the quick version wired two things together that should never have met. Add up the hours your team spends working around the past instead of building the future; that number is your monthly interest payment, and most shops have never calculated it.
High-interest versus low-interest debt
A hardcoded config value is low-interest debt: ugly, cheap, fine to carry for years. A data model that does not match how the business actually works is a payday loan; it compounds on every feature that touches it. Triage accordingly. The skill is not avoiding all debt, it is refusing the predatory kind, and knowing which loans to pay off first when the interest starts eating the roadmap.
The ledger habit
We keep an explicit debt ledger with clients: what we owe, why we borrowed it, what it costs per sprint, and the planned payoff date. It turns an invisible drag into a line item an owner can reason about. Some entries we pay down next quarter. Some we consciously carry forever. Both are fine, because both are decisions.
The one warning sign
When engineers start prefacing estimates with it depends how deep the rot goes, the rate has gone predatory and the ledger is overdue. Loans with a plan are strategy. Loans without one are how codebases die, quietly, at three weeks per feature.
Paying down technical debt on purpose
The point is not zero technical debt; it is zero surprise debt. Martin Fowler’s classic write-up frames the metaphor; our ledger habit makes it operational. Every Build engagement we run carries one: what we owe, why we borrowed, what it costs per sprint, when we pay it down. Owners can reason about a line item. Nobody can reason about rot.
If your estimates now start with ‘it depends how deep it goes,’ the ledger is overdue.
