I found your analogy funny.
But in this case they’re more related than you think. I was enumerating types of software that connect to the P2P network and would be affected by block size changes. In terms of an upgrade problem, if you want to do a hard fork, it becomes obvious that upgrading is far simpler if you don’t require updating those 15million nodes.
This is not to say that the other nodes lack importance. They are of critical importance, as you need those nodes for the SPV nodes to connect to!
This article describes why the Segwit2x approach was taken, and I stand by the claim — if you want to upgrade the network, you’re better off not changing everything — thats just a hard thing to do. It’s the same reason why segwit use is so anemic today.