I argue for two modifications in Sunstein’s definition of manipulation, designed to make the definition more compatible with a welfarist/utilitarian view of what manipulation is and why we should care about it. I think we need a clearer distinction between good and bad manipulation, and a distinction between good and bad reflection. Good reflection is actively open-minded and serves to achieve the decision maker’s goals. Manipulation is designed to prevent such reflection, or prevent the choice that would be made if such reflection occurred, and this is why it is bad, even if it is consistent with reflection that merely bolsters a favored option.