That's not how it works.
Nodes relay information to other nodes but one node doesn't need to broadcast to the entire network!
The burden of 'massive redundancy' you speak of is distributed across the thousands of nodes,
each responsible for communication with other nodes.
Wow. Now I understand why you're so confused about this blocksize issue. That _is_ how it works.
Every node in the network must receive and process every transaction in every block.
... from just the nodes its connected to. It does not have to broadcast blocks/tx's to every node in existence.
There is no distribution or sharing of that load.
The 'distribution' he was referring to was the fact that I only need to send to my connected nodes, and they send to their pool and so forth.