Does this mean that we need to connect every pair of threads separately? Or is there a central server that helps data exchange?

If they are threads one separate machines entirely, then we will need some sort of network/communications protocol to transfer data between them (e.g. Sockets).

On the same machine, we can use things like MPI (https://en.wikipedia.org/wiki/Message_Passing_Interface) for message passing.

Will this class cover the message passing model in much more detail than this abstraction? And can coordination and information sharing techniques between 'threads' be neatly broken down into message passing vs. shared memory and synchronization, or is there a gray area somewhere in the middl?

This is building on what we learned in CS 110 with disparate processes in different address spaces passing messages and control back to each other using signals and pipes.

Do most operating systems support both the shared memory model and the message passing model? Do some support just one or the other? As far as I can tell, it seems that the shared memory model is more widespread. Is this true, or do I just think that due to my limited experience?