Is public policy responsive to demographically and ideologically unrepresentative comments given at public meetings? I investigate this possibility using a novel data set of over 40,000 comments made at the San Francisco Planning Commission between 1998 and 2021, matched to information about proposed developments discussed in hearings and administrative data on commenters. I document four stylized facts: First, commenters at public meetings are unrepresentative of the public along racial, gender, age, and homeownership lines; second, distance to the proposed development predicts commenting behavior, but only among those in opposition; third, commission votes are correlated with commenters’ preferences; finally, the alignment of White commenters (vs. other racial groups) and neighborhood group representatives and the general public (vs. other interest groups) better predict project approvals.