Loose coupling vs tight coupling, or, direct process dependencies vs resource flows

NRP does not link processes together directly, as in previous process -> this process -> next process.

Instead, the output of one process is a resource, which may become an input to another process. The processes might not know about each other beforehand. They might live in different networks. One process might create outputs that are inputs to many other processes, and likewise one process might get inputs from many other processes.

These indirect links sometimes confuse people who are used to conventional project management systems with direct dependencies. But the indirect links (loose coupling) are lots more flexible, and in many situations (like the ones where the next process is unknown) allow things to happen that cannot happen with tight coupling.

Here's somebody writing about the same tradeoffs in workflow systems: http://bionics.it/posts/workflows-dataflow-not-task-deps