I'd previously been under the impression that the details of the #ARPANET IMP-to-IMP protocol (as opposed to the well documented #1822protocol for Host-to-IMP communications) had been lost, but it turns out that's not the case. In fact, it appears that in the last 15 years copies of multiple versions of the full source code for the #IMP software have been recovered by using OCR on hard copies. Much more information at https://walden-family.com/impcode/. I don't know the extent that the protocol itself has been reverse engineered from the code.
I love that this was written using #Midas, a programming language which was developed by a hospital. I think I did actually already know that, maybe from https://katiehafner.com/books-new/where-wizards-stay-up-late/ - presumably #BBN just used code they already had from an earlier consulting project.
Anyway - preserve your software projects. I'm sure many of the people doing this in the 60s and 70s didn't realise that what they were doing would be historically significant at the time.





