Abstract: Many content systems (e.g., social networks) present to a user a set of content items posted by other individuals. The user may selectively view content items that reinforce and are consistent with the user's perspective, creating an “echo chamber” effect. Conversely, content systems that selectively expose users to content items exhibiting contrary perspectives, and from individuals with no connection with the user, may alienate the user. Presented herein are techniques for recommending content items that present a different perspective from that of the user, and from individuals who share a similar profile to the user (e.g., alternative opinions from other individuals within the user's social circle or community). Optionally, opinions may be selected that do not directly oppose the user's perspective, but that are orthogonal with it. Such selective recommendations may persuade the user to consider contrary viewpoints that may alter the user's perspective while reducing user alienation.