It might be noted that both NASA and ESnet were originally primarily DECnet networks as they pre-date IP development by a little bit. They also, in there early forms (SPAN, HEPnet, and MFEnet) predate the NSFnet.
Yes, just like users of smoke signals and telegraphs eventually converted to using the telephone. Only a question of how far you want to go back. Not to discredit SPAN, HEPnet, and MFEnet (or Bitnet or UUnet for that matter), but at that time when actually GOSIP (TP4/CLNP) was the officially mandated way to go, it was actually pretty gutsy of Dennis Jennings at NSF and then Steve Wolff to go for and continue to use IP. If it weren't for them, chances are that CLNP would have won (with us having as much lesser control as we would have lesser addressing issues now). Even DECNET Phase V was heading towards a CLNP solution.