It's absolutely not a miracle anything gets through at all, if you know how TCP/IP protocol works. And on top of that there are whole lot of other redundancies on application level. A situation where some parts of communication seemingly disappear into another dimension (and the other side is not taking that into account) are not something that should be considered normal in this day and age. It's something that should be looked at.
I do not believe it's Apple censoring messages (given that they have no ability to do so), but comparing Internet to a sinking ship or runner not arriving is very, very wrong.
I do know how TCP/IP works. Of course it's highly reliable, otherwise the Internet (and local networks) would not be practical.
But you do get hyperbole, don't you? (The TCP joke you posted suggests you do get it.)
Ah... that's the nature of human communication - among other things, whether meaning and nuance are being adequately conveyed.
What we have here is a garden variety end-user complaint. Something malfunctioned, and they imagine that the cause must be exotic, rather than one of the myriad garden variety failures that can occur.
I've worked on both the technical and creative end of a wide range of communications media for nearly 50 years - not only am I fairly well-steeped in it's history, but I've been living with its realities. Theory is one thing, practice is another. Murphy's Law still rules.
I've stood in hallways outside operations centers with vice-presidents demanding, "How can we ensure that this never happens again?" There are times you have to reply, "First, let us get the system running again. We can analyze the failure later." Considering how many times those failures turn out to be one-offs that are beyond economic justification to prevent...
Of course failures should be looked at and analyzed. That's a key part of improving reliability. But a lot of failures are built into the system. Let's say the local cell tower has reached its maximum channel capacity, and the next available tower is too distant to establish a reliable signal... How often should the messaging application re-try before timing-out? Infinite re-tries are generally not a good idea (as the joke you posted points out).
In my current line of work a fair amount of what I do relates to communications that are not getting through - poor cellular voice/data and Wi-Fi connections, failure to connect to servers even when those digital data paths appear to be working, malfunctioning microphones and speakers, incorrect passwords, failing to properly understand the nature of the user's problem, etc., etc., etc., etc., etc....
I live in a world dominated by the imperfections, failures, and exceptions. 99.9999% reliability may be meaningless to the person in the remaining 0.0001%. They don't care that it's working well for everyone else. And the higher the expectation of reliability, the more disappointing and/or distressing the imperfections and failures may seem. It's human nature.
From my perspective my life would be easier if more people appreciated just how complex these systems are, and remembered (or understood) just how marvelous and yes, miraculous, it is that we can communicate so quickly and reliably in so many ways.