I wrote all of the original code for TYMNET from 1968 through 1971 in assembly for the SPC-12 (which lasted for only a year or so), the SDS-940(which ran the network supervisor, access control, and accounting package), and the Varian 620-I (the network nodes). Norman Hardy contributed ideas, a few of which were taken from the Octopus network (1966-1968) at Lawrence Radiation Laboratory, Livermore, which networked the IBM 7030 Stretch, the CDC 6600, and some other machines and peripherals. Ann Hardy and David Gardner made helpful modifications to the SDS-940 operating system.
After 1972 significant contributors to TYMNET included Joe Rinde (Supervisor on the TYMNET engine), Carl Holmberg (further development on Supervisor after Rinde), John Kopf (operating system for the TYMNET engine), Art Caisse (interfacing TYMNET to various mainframes), and June Nishimoto (PDP-10 base enhancements). Guy Blood and Verne Van Vlear did not work on the network itself, but took care of accounting and access control databases.
On the hardware side, I designed and microcoded the CPU of the TYMNET engine. Other hardware engineers on the engine project were Barry Burnsides, Ron Graves, and Larry Pizzella.
The last major contributors to the TYMNET technology made the conversion to SPARC workstations in 1996. They included Romulo Raffo, Bill Soley, and myself.
Bill Combs was the first and major sales person for TYMNET.
In a project of such a size, many other people were involved, but the above list is a fairly complete list of the major innovators and contributors.
Incidentally, I booted the network in its complete form in November of 1971. It ran without a single system crash or reboot until March of 2003, when it was shut down. It survived many hardware and software upgrades during that time without a single system wide failure. Also, I wrote three versions of the routing function that determined the optimum path through the network, one in assembly for the SDS-940 in 1971, one in assembly and microcode for the TYMNET engine in 1974, and the last one in C for the SPARC in 1996. The one I wrote in 1974 ran 24/7 for 22 years without being touched or "maintained". I think that might be some sort of industry record for mission critical software.