FYI, the IP protocol does specify that if a packet that's larger then the MTU size limit assigned by any routers/switches on the way, it can "fragment" the packet (i.e.: break it down into multiple pieces).
If you use TCP as transportation protocol, it'll handle the part to put things together.
If you use UDP, the packet in most cases will simply by dropped on the receiving side.
Therefore, I think it's fair to say dropping packet is a "by design" behavior of network.
Btw, before the use of CAT5 cables, you can have medium packet loss even on intranet. That's why in the beginning days when TCP/IP was announced, people still prefer IPX/SPX over it since it has a smaller header size (of course, most system was already using IPX/SPX is the major factor).
You just have to agree that the quality of network infrastructure has been significantly improved over time that what once was normal (high packet drop rate) has been seen as weird now.