But long options are not so complicated an idea that “invent” might be too high falutin’.
Just for fun I’ll split hairs though: TOPS-20 (and VMS) basically only had “long” options (they could be a single character of course). CP/M and DOS got their arg syntax from TOPS-10 & RSTS but we never had TOPS-10 at MIT AI.
What RMS came up with was to have both long and short, to use two dashes to disambiguate from concatenated short options, and used the = syntax (which dd, at least, already used). Then bfox put command completion into readline, which definitely was inspired by TOPS-20 command completion (itself from TENEX).
Before that we had our own OS, ITS, and most programs didnt take options because the system didn’t really have a command line at all: instead of what is called a ‘shell’ in Multics (and copied by Unix) we just used the debugger. And of course lispms, where the whole concept of command line doesn’t make sense.
By 1984 there was one Unix machine, a VAX, in the lab but nobody used it (small word size, didn’t have much software, primitive development environment compared to what we were used to) which is why rms started using it to develop GNU.
Whew. More than I thought one could write on the topic!
Now ELF is ubiquitous I wish the tab command completion were built into a section of the binary rather than being in sidecar files elsewhere in the filesystem, which is fragile.