That is very interesting. Skimming through the thesis, I find it fascinating introductory material to Nix. And you are correct, Nix started as a deployment tool. Although I'd argue in an era (1978!) when software are mostly distributed as source, reliable deployment largely implies reliable building from source.