We recently worked on a greenfield project, building an idea quickly and optimizing for information rather than production-readiness. We used Hasura as our data layer to speed up development, but we soon realized that having a new tool in our stack was causing us to lose autonomy and speed. To balance the trade-offs between shipping quickly and avoiding unnecessary technical debt, we identified a strategy called "optimizing for deletion." This involves creating guard rails that prevent prototypes from leaking into other parts of the codebase, ensuring that code is completely isolated from production code and interactions are limited and easy to find. By optimizing for deletion, we can reduce the risk of building with something new and make it easier to refactor or replace tools in the future, resulting in faster development, better production code, and reduced maintenance nightmares.