How Developers Lose Time Rebuilding Context After Every Break

Published July 15, 2026

The hidden tax of context switching

The cost of switching projects isn't the work itself — it's the time spent getting back to the mental state you were already in before you left. That state includes things that never make it into code: which approaches you already tried and rejected, why a particular abstraction exists, what the actual blocker was versus what it looked like on the surface. None of that is visible in the codebase. It only existed in working memory, and working memory doesn't survive a two-week gap.

What re-reading commits and PRs actually recovers (and what it doesn't)

Commits and PRs are a reliable record of what changed. They're a poor record of why. A diff shows you added a retry loop around an API call; it doesn't show you that you tried three other approaches first and this was the one that didn't flake under load. Reading the code back tells you the current state, but reconstructing the reasoning that got you there means re-deriving it from scratch — which, ironically, can take almost as long as solving the problem the first time.

A rough cost estimate per switch

Anecdotally, the reconstruction phase on a project you haven't touched in one to three weeks runs somewhere between twenty and forty-five minutes before you're back to writing meaningful code, and that's assuming nothing's actively broken. Across a handful of active projects rotated weekly, that adds up to hours a month spent purely on remembering, not building. It's not a dramatic cost in any single instance, which is exactly why it's easy to ignore and expensive in aggregate.

A cheaper way to preserve context between sessions

The fix isn't writing more documentation — it's writing down the small set of things that don't survive in code: what you did this session, what you decided and why, what's blocking you, and what you'd do next. Kept in one place per project and summarized back to you on return, that's the difference between spending five minutes reorienting and spending forty-five reconstructing something you already figured out once.