This did not become law until 1989. https://www.sunsteinlaw.com/copyright-flowchart (Stallman started work with Steele on EMACS in 1976, and the dispute happened in 1985.) Until that point, US copyright law protected the public from copyright abuse with strict notice, registration, and renewal requirements, which were eliminated in order to join the Berne Convention. [This paragraph has been edited to correct egregious errors]
Moreover, until 1980, it wasn't clear that software was copyrightable at all — the analogy to printed books is rather weak. Congress updated the law in 1980, although unfortunately did not require the publication of source code to gain the monopolies granted by copyright, an omission which is likely to cost us much of the intellectual history of the late 20th and early 21st centuries, as source code is lost long before it passes into the public domain. Passage into the public domain is the essential step that fulfills the copyright bargain with the public: a limited-time monopoly in exchange for the promotion of progress in "science" (theoretical knowledge) and the "useful arts" (practical knowledge), as the US Constitution explains. So, much of the current practice of software copyright is a fraud — the public suffers the cost of the copyright holders' limited-time monopoly, but then are cheated out of the benefit of an expanded public domain.
Consider, too, that we're talking about events that happened in 1981–5, in communities that were struggling to accommodate the impact of these radical new intellectual enclosures that destroyed the rights they had traditionally enjoyed, comparable to the enclosure laws that eliminated common grazing grounds in England before the Industrial Revolution.
> Gosling was perfectly within his rights to sell gosemacs to Unipress
This is not clear; quite aside from the question of whether Gosmacs was a derivative work of PDP-10 EMACS (it was not, under current caselaw, but that caselaw didn't exist in 1985), Gosmacs of course included contributions from other people, who presumably did not assign copyright to Gosling; they may not even have been aware there was anything to assign.
> I really don't get how you're coming at Gosling so hard...he's clearly uncomfortable talking about the topic
Entirely independent from the question of copyright notices, Gosling's account is false and defamatory, and omits the crucial background fact that the program he sold to Unipress was an imitation of a program Stallman had devoted the last seven years of his life to, and the subsequent fact that three months after the initial release of GNU Emacs, Stallman had removed the disputed code. It's unsurprising that he would be uncomfortable lying. Most people are.