Our basic solution to these constraints is a system of replicated vote servers which store information about the popularity of each article separately from the articles of information themselves. Using their normal news reading clients, users contribute votes for or against articles they process. These votes are sent to the nearest vote server where they are grouped together and shared with other vote servers to create a net-wide collective opinion on the relevance of each article. These aggregate opinions are then used by the news reader clients to filter the articles shown to the user. Users can also make their votes directly available to other users, which allows the custom or group moderation of articles.