Leave room for the good accidents
The best idea in a project is usually the one nobody scheduled. Over-planning locks the doors it would have walked through.
Look back at the projects you’re proudest of and find the best single idea in each one. Odds are it wasn’t in the plan. It was a throwaway comment in a review that everyone went quiet after. A bug that looked better than the feature. Or a placeholder that outlived three rounds of intentional copy because the placeholder, written by someone who wasn’t trying, was the only line in the deck that sounded like a person.
We’ve watched this happen enough times to stop calling it luck and start calling it a resource. The good accident is a renewable one, and like most resources, you can manage it badly.
The way you manage it badly is over-planning. A project where every hour is assigned, every decision is pre-made, and every meeting has an airtight agenda is a building with no doors. Nothing unexpected can get in, which sounds like discipline and functions like insulation. The plan doesn’t just schedule the work. It schedules what kind of ideas are allowed to occur, and it only allows the ones somebody predicted in week zero, which are, by definition, the ideas everyone already had.
Luck needs a doorway, and the doorway is slack.
Not chaos. Slack. There’s a version of looseness that’s just disorganization, and clients can smell it, and they’re right to. What we mean is controlled looseness, which is a method, not a mood: the plan is rigid about the things that must be true (the deadline, the budget, the parts of scope with consequences) and deliberately soft everywhere surprise is affordable. The skeleton is fixed; the cartilage flexes.
In practice it looks unglamorous.
- An afternoon in the schedule that belongs to nobody.
- A decision held open a week longer than efficiency would like, because early certainty is often just impatience with a title.
- Reviews run loose enough that a tangent can survive its first thirty seconds, since the tangent’s first thirty seconds are always indistinguishable from a waste of time.
- Somebody in the room whose job, unofficially, is to say “wait, go back to that.”
None of it bills cleanly. All of it pays.
The uncomfortable part is that slack looks like waste right up until the moment it doesn’t, and a spreadsheet will always vote to remove it. Efficiency is legible; serendipity isn’t. So the doors get optimized shut, one scheduling pass at a time, by people acting entirely reasonably, and the projects come out exactly as good as the plan and never once better.
Exactly as good as the plan sounds fine until you sit with it. It means the ceiling was set on day one, by the version of the team that knew the least.
Plans are how the work gets done. Accidents are how it gets good. Build the schedule tight where it has to hold, and leave a few doors on the latch, and when something wanders in mid-project looking like a distraction, give it thirty seconds before you show it out.
Some of our favorite work came in through those doors. Not one piece of it knocked.
