You might want to think about what it was that, in a highly-competitive landscape of networking options, lead to TCP/IP and BGP succeeding where other options didn't.
I've just unearthed my copy of The Matrix, John S. Quarterman's exploration of "computer networks and conferencing systems worldwide". It was published in 1990, just on the cusp of the breakout of the Internet (there's a late 1990s revised edition I've not seen).
It covers a lot of ground. Not of much practical use at this point, but a phenomenal historical document: Layers & protocols, management protocols, administration. And the networks!
You could also ask why (privately developed) Ethernet, instead of the other LAN protocols like Token Ring. One possible answer is government is an elephant and has the power to push their standards - perhaps due to algore.
That's an interesting question, but is beside my point. My point is to counter the notion that without government funding, we wouldn't have a global network, that nobody would have connected computers together.
And I'll suggest that your response above points strongly at a probable explanation: the U.S. government as a single "consumer" in this case, was able to force standards adoption.
Hardly the first time that's happened. Look to the establishment of the US Bureau of Standards (by Herbert Hoover, as Secretary of Commerce), the standardisation of fasteners, plumbing, railroads, electrical components, and transmissions, among other elements. The US Navy's transport needs in Vietnam resulted in the marine shipping industry standardising on the "20 foot equivalent" standard cargo container (the concept itself had been kicked around for decades).
There are numerous other document, data, signage, fabrication, materials, commodity, etc., standards established by the US Federal (and at times, state) governments.
It's a very powerful role. One that seems lacking of the past two decades or so.
I've just unearthed my copy of The Matrix, John S. Quarterman's exploration of "computer networks and conferencing systems worldwide". It was published in 1990, just on the cusp of the breakout of the Internet (there's a late 1990s revised edition I've not seen).
It covers a lot of ground. Not of much practical use at this point, but a phenomenal historical document: Layers & protocols, management protocols, administration. And the networks!
AQ, HEPnet, PHYSNET, BITNET, USENET, UUCP, FidoNet, Ean, VNETT, XEROX Internet, EASYnet, Tandem, HP Internet, UUNET, DASnet, CUNYVM, CERN, EUnet, FNET, ARISTOTE, SMARTIX, REUNIR, Minitel, Dnet, DFN, AGFNET, BELWU, ARIADNE, HEANET, INFNET, SURFnet, Enet, JANET, UKnet, GreeNet, COSMOS, NORDUnet, DENet, DKnet, ....
(Nothing about WWW or Gopher though. Al Gore is mentioned.)
On the one hand, one network would have had to emerge. But on the other, why the Internet?