For the engineers, particularly the Original Engineers, the internet is a logical architecture that allows us to connect virtually any kind of networking machine together.  That architecture can change. 

Here's the 1995 Federal Networking Council definition of "the internet":

There are two key things to notice about the Engineers' approach.  First, this definition of the “internet” emphasizes globally unique addressing (supporting interconnectivity) and the use of TCP/IP, but makes clear that these elements can change.

IP can have “extensions/follow-ons,” TCP/IP can be subsumed by “other IP-compatible protocols,” and services using communications infrastructure can be made available privately or publicly, depending on what makes sense. Any logical architecture that provides for interconnection between networks and a set of agreed-on protocols (with some connection to the historical TCP/IP suite) will be “the internet” to the Engineers.

Second, the FNC/Engineer definition does not recognize the role of the transport pipes, because the Engineers are indifferent to transport.  What they care about is the logical architecture, the overarching set of protocols that brought different networks together.  And could change.

The telcos have a different way of looking at things -- they see "the internet" as the pipes, and don't care much about the religion of the original protocols.

But the two groups, the Engineers and the Telcos, have common ground:  the protocols are expected to change.  They can change and we'll still have "the internet". 

There's a third group -- let's call it the internet visionaries group -- who focus on the social and cultural effects of the internet, and use those effects to shape their own definition of "the internet."  What they care about are standards that allow end-to-end communication and relationships that the pipes don't control.  They're not so sure the protocols should change.  If they change, it won't be "the internet" any more.