Granted, the only time the Open Systems Interconnection model, which is being made fun of here, has come up was on job interviews—have you memorized it?—or when someone was trying to avoid work—"oh no, our group only handles layer 3, and you've got a layer 4 problem, see?". Or in metadiscussions, such as this one. Some of the layers might be seen as somewhat arbitrary slices of a cake; there is dispute whether the Address Resolution Protocol (ARP) is layer 2, layer 3, or whether it needs the new fractal layer 2.5. Once you have pigeonholes, everything must naturally (or otherwise…) fit into a particular box. Some will instead do violence to the model and invent new boxes to stuff things into. Procrustes is a patron saint here. Some things (security, for instance) may need to be thought about and applied and maintained throughout the cake. Or not.
More productively, the lower levels do make sense, and may help you when debugging a problem—is it plugged in?—though with the increasing abstraction things get increasingly vague, or even irrelevant. Is the load balancer issue a transport, session, or presentation layer problem? Who cares!
I'm pretty sure the U.S. army had a shorter list of layers that made a lot more sense to me, maybe because it bundled layers 4 through 7 of the OSI into one "software-y stuff" layer.
(The OSI model, for reference, involves the physical, data link, network, transport, session, presentation, and application layers.)