by Matt Wright, March 2021
tl;dr Say "OSC sender" and "OSC receiver" instead of "OSC client" and "OSC server"
The OSC 1.0 spec unfortunately defines the terms "OSC Client" and "OSC Server" in terms of sending and receiving (respectively) OSC Packets. This somewhat ridiculously implies that any receiver of OSC packets is somehow providing a "service" simply by accepting OSC packets, no matter what, if anything, occurs in response.
In reality, the Client-Server model may or may not apply to any particular use(s) of OSC. There may indeed be one or more actual "services" being accessed somehow via OSC. Any designation of "clients" and "servers" should be in terms of a specific service (e.g., "tempo server", "joystick readings server", "stage lighting server", etc.), not in terms of OSC communication itself. In 2009 Freed and Schmeder wrote in these terms: "OSC is specified as a content format: OSC doesn’t specify service content or behavior, it is just a format for clients and servers to exchange data in to implement custom service behavior." Fundamentally, it doesn't matter whether or not there is a service or servers or clients; OSC is an encoding that can be used to format messages sent from somewhere to somewhere else, regardless of context.
As of 2021, Adrian Freed and Matt Wright recommend the terminology "OSC Sender" for anything that sends OSC packets and "OSC Receiver" for anything that can receive OSC packets.
The terminology "sources" (senders) and "sinks" (receivers) could apply to OSC (e.g., by analogy to how these terms can refer to electrical current).