The horizontal scaling section ties everything together well, Pub/Sub as the answer to "user A on server 1, user B on server 2" is the exact moment this stuff stops being theoretical and becomes a real architecture decision.

Good reminder too that Pub/Sub isn't a client-server replacement, easy to conflate it with WebSockets/SSE when skimming, but it's solving a completely different problem (backend fan-out vs client delivery).