Multiple devices, multiple points of failure

There is an inexhaustible number of problems that can occur on even simple software when you introduce the unpredictability of network interactions. A single off-by-one error in an upstream service could mean a missing the closing curly-brace in a JSON string, rendering an entire payload impossible to parse. Internet service provider (ISP) service interruptions or weak wireless signals can result in timeouts and incomplete payload delivery. Meanwhile, the stability of the remote system you're requesting a resource from is entirely out of your control. With all these factors introducing the potential for errors, we can't simply hope to avoid errors or exceptions in our software. We must assume they will occur, and design around that eventuality.

...