I think it's the "most reliable" way to get them to work. When you have a centralized server, both devices know exactly how to connect to the centralized server and communicate through it.

Local connectivity seems to present all sort of silly issues that make the user experience a bit inconsistent. Heck, look at Chromecast, it works great about 90-95% of the time. That 5-10% of the time it doesn't work, it becomes a complete pain.