Need more here.

SuperPeer uses the concept of clustering. There are two kinds of entities in a SuperPeer architecture, endpoints and joints. Sometimes we don't care which one they are; at that point, we call them nodes. An endpoint can connect to only one joint. A joint can have many endpoints connected to it. A joint can have many connections to other joints. These connections have the concept of parent and child. When an endpoint is connected to a joint the joint is the parent and the endpoint is the child. A joint can only have only one parent, but it can have many children. This description was for a hierarchical SuperPeer network, I think there is at least one other type, that uses gateways instead of hierarchies.

''You wrote "A joint can have many connections to other joints. These connections have the concept of parent and child." Which connections?''

''Also, you wrote "An endpoint can connect to only one joint." Does it have to connect to a joint?''