G. Floyd Steele's Smithsonian interview (PDF) is fascinating, especially as it relates to CADAC. To give a quick overview:
Steele was contracted by John Marquetti (sic[?], his name is spelled Marchetti in other sources so I'll use that spelling for the rest of this post) of Cambridge Air Force Research Lab, on the basis that he give up on the Snark missile project and work on the DEW Line instead.
Marchetti wanted a computer to control DEW Line radar in a decentralized approach (one computer per radar, not networked together).
Marchetti wanted a simple & reliable computer, contrary to the contemporary push for a larger, centralized design: Whirlwind.
Steele fund-raised to create the CRC, and developed a small, drum-based general purpose computer for this purpose (CADAC).
Steele was kicked out of CRC in the ensuing scuffle to build the computer, and Gordon Turnbull took over.
CRC finally delivered the CADAC, but it was too late as Whirlwind had won by default.
CADAC design was released to Steele, who went on to found DICO.
Steele claims that a decentralized early warning system was possible, but he never got the chance to prove it.
Marchetti lost face, Lincoln Labs was established and started work on Whirlwind II a.k.a. SAGE.
I ran this past spouse and he speculates that the CADAC concept might have inspired the creation of BUIC, which was a smaller scale, less-centralized backup system to SAGE. There's not a lot written about BUIC project history, so I don't know where I would even start with investigating a possible link.
Still, it's making me want to learn more about early computerized air defense/EWS, non-SAGE ideas that were around at the time, as well as early minicomputers (of which I think CADAC, MADDIDA, G15, LGP-30, etc. can reasonably be classed).
