Gossip protocols usually create a complete view of the participating members to randomly choose a subset of the members to share information.
This paper suggests these algorithms suffer from scalability issues, citing that building the complete view is time and memory consuming. This paper introduces the concept of a local view, which is a subset of all members reachable via a node, stored on each node. Each node can (usually) have a different local view throughout the node life. The algorithm presented here details how the local view of a node changes over time due to node failure, new node addition and network partition.