Software Drift
Software drift is the growing gap between the dependencies, runtimes, and frameworks a codebase actually uses and their current, supported versions.
Software drift is the distance between what your codebase runs — packages, frameworks, runtimes — and the current, supported versions of those same components. Every release you skip widens the gap.
How It Accumulates
Drift grows quietly. A project pins its dependencies to pass a release, priorities shift, and months later the stack is several major versions behind. Nothing is visibly broken, so nothing gets fixed — until an end-of-life runtime or a vulnerability in an old version turns the backlog into an emergency.
AI-assisted coding accelerates the problem: assistants trained on "latest" or outdated APIs write code against versions you do not run, so the gap between assumed and actual versions widens from both directions.
Why It Matters
Drift is measurable maintenance risk. Old versions eventually lose support and security patches, upgrades get harder the longer they wait, and the fix window narrows. Measuring drift — for example as a 0–100 DriftScore built on libyears — turns an invisible backlog into a number a team can track, budget, and gate on.
Related Terms
Drift is measured with metrics like DriftScore and libyear, recorded against a lockfile, and becomes urgent as components approach end of life.