The Proven-Better-New Framework
Pincus built Farmville not by inventing a new genre but by isolating what an existing game had already proven, keeping everything that worked, and changing only one or two specific things. He calls this proven-better-new, and he applies it as a sequential test, not a brainstorm. Start by taking the proven product and copying it as closely as possible: onboarding, display logic, pricing structure. Assume the incumbents have already done the work of figuring out what users want most. Then ask whether you have anything that 10 out of 10 users would call genuinely better. If the answer is no, and it usually is, you probably have new, meaning an untested twist. That twist needs to be isolated and tested by hand before any software gets built. Farmville's edge over the game it competed with was not a new mechanic. It was better art, cleaner math, and the removal of a stranger-danger social feature that Pincus believed their audience did not want. That was it. They turned it on without marketing and hit 171,000 installs on day one. "If they do like it better, then we can start to use AI and say, is there a way to use AI agents to automate this. But that's a very very secondary. The first is can we get to a clearly better product experience." The framework keeps founders from falling in love with the new idea before proving the better part. Most things that feel like innovations are just additions to an unproven base, and building them out is how teams waste six months discovering users did not actually want the thing.