Two modes of controlling motion. A rocket: every assumption about the environment is built into the pre-launch calculation; once launched, there is no intervention; a small error in assumptions becomes a catastrophe at the far end of the trajectory. A car: it stays on course through continuous small corrections based on what is visible through the windshield; an erroneous assumption is not a catastrophe, but the next turn of the wheel. A business plan is a genre of pre-launch calculation; Ries argues that a startup is run like a car.
Why pre-launch calculation fails under uncertainty: a calculation is only as good as the assumptions built into it—and a startup is defined by a condition in which correct assumptions do not exist in advance (A startup is defined by uncertainty, not size). Precision of execution does not compensate for false assumptions: it makes them worse, letting you travel farther in the wrong direction.
The cost of rocket mode has a name: achieved failure—the successful, precise, disciplined execution of a mistaken plan. It is the same waste as in Waste in innovation is work that produces no learning (“flawless execution of an unnecessary plan is the most insidious waste”), raised from the level of tasks to the fate of the company: a failure that looks like a series of successes right up to the end.
The everyday reality of steering mode is unheroic: a startup spends most of its time tuning its engine of growth—improving its product, marketing, and processes. Not plan execution and not dramatic pivots, but routine tuning by the instruments; a pivot is a rare special case of the same loop, not its essence.
The consequence: when the engine works and steering has been mastered, you can press the accelerator—scale at maximum acceleration. The order matters: accelerating before there is a working engine is like speeding up a rocket with no calculation at all, the worst of both modes. A subtle point: knowledge of steering cannot be expressed in advance—it is acquired only by driving. The connection to A working practice without a shared language cannot be transferred: what can be codified and transferred is the discipline of the loop (when to look at the instruments, how to decide “pivot or persevere”), but not steering itself—it remains a skill cultivated anew in every startup.
How this supports A startup abolishes not management, but management of the predictable
That note argues that managing uncertainty is a different discipline, not an absence of discipline. This one gives it a dynamic portrait: not execution of a pre-launch calculation, but continuous steering by feedback. It also makes the corporate cost of the myth concrete: demanding “rocket-style” reporting from a startup—business-plan milestones—means forcing it to simulate pre-launch calculation and move toward achieved failure with exemplary discipline.