Comprehension debt
The growing gap between how much code a system has and how much of it anyone on the team actually understands, a typical cost of fast AI-written code.
Addy Osmani defines it as “the growing gap between how much code exists in your system and how much of it any human being genuinely understands”. Code that nobody wrote by hand and nobody read closely can work fine, until it has to be changed, debugged at night or explained to an auditor. Then the team pays: slow fixes, fear of touching modules and a growing bus factor problem.
It differs from technical debt: the code can be clean and still not understood. Vibe coding produces it fastest; agentic engineering keeps it in check with specs, reviewed plans, architecture notes and asking the agent to explain changes, not only to make them.
Sources
Related terms
- Technical debt: Shortcuts in code and design that speed up delivery now and make every later change slower until they are paid back.
- Vibe coding: Building by prompting and accepting the result without reading the code closely.
- Bus factor: How many people would have to disappear before a project stalls. A bus factor of one means one person holds the keys.
- Agentic engineering: Professional software work in which coding agents write most of the code while engineers own the specs, the checks, the review and the result.
- Parallel agents (git worktrees): Running several coding agents at the same time on one repository, each in its own git worktree and branch, so their edits do not collide.
