The Honest Roadmap — Shape Up, Cycles, and the Graveyard

Build the execution layer: Ryan Singer's appetite-driven cycles as the operating framework, opportunity-cost reasoning as the decision logic, and the graveyard — the explicit list of things you decided not to build — as the artifact that proves the discipline happened. By the end, you have a shaped pitch and a running cycle with hard scope boundaries on a surviving product, plus a roadmap whose unit is the cycle, not the quarter-long fiction. Appetites are time-boxed spikes: I'll spend two days investigating; if no answer by then, we change approach — the same discipline applied to features instead of investigations. The graveyard is the DEPRECATED.md: the artifact that makes absence intentional and revisitable instead of implicit and forgettable. Killing features is deleting code — you've deleted days of your own work because requirements changed, and you learned that code-not-deleted accrues interest faster than code-deleted hurts. Features-not-killed do the same to a product. And Cagan's outcome-based roadmap is TDD: commit to the observable behavior, let the implementation vary, accept that it's hard for exactly the reason TDD is hard — it trades the satisfying activity for the meta-activity that makes the satisfying activity count.