In Taiwan, ten years ago, we used Pol.is, a pro-social conversation platform using machine learning to map opinion clusters and identify statements that bridge divides. These “uncommon grounds,” instead of the loudest voices, were amplified. This informs the Community Notes on X (formerly Twitter), YouTube, and Facebook today, which decentralize fact-checking by requiring those holding opposing viewpoints to agree on context before a Community Note is published. These pro-social technologies demonstrate that when we design for collaborative diversity, even polarized societies can move from outrage to overlap.