I get what you mean. After all the basics of a microprocessor relies on logic gates and they can be implemented even with vacuum tubes. What I mean with "start with the first processor ever" is "to know the basics from one of the popular first commercially available encapsulated microprocessors fabricated by a known brand still alive" but that was too long, sorry for not being too specific.
The only low level programming I'm familiar with is x86. I never bothered learning Z80. If that were the case I could had started with Z80 even if it isn't the first. While a 4004 won't look like 8086, a 4 bit processor will be enough to know the roots.