Another finding on YouTube: a lecture at the Computer Museum History Center, given by some of the programmers of the first COBOL Compilers.
This lecture was given in 1997 and – even back then – COBOL was already seen as antiquated. This was also around the time when CODASYL (the organisation behind the COBOL-Standard) ceased its operations.
I find it notable how relentless these programmers talk about the quality of their work. They often discuss reasons for the bad design of the language: (admittedly) tight hardware-specifications, budgets, deadlines, frequent hardware-failures, ignorant managers, pressure from the government, tough competition, involved companies pursuing conflicting agendas, missing quality-standards („the damn thing has to run“), bad communication („the engineers did not talk to us“), lack of competence („We were no Computer Scientists – Computer Scientists worked at universities – we were Computer Practitioners“, „we didn’t know anything about elegance“),…
Harold Lawson even admits that they didn’t really understand what a compiler even is: they thought of a source code more as a spreadsheed – and I have to say: this matches my impression of COBOL that I have had already before watching this video: it does look somewhat like a table-description-language. Also he says he would have designed COBOL differently, if – at the time – he had known about compilers what he knows today.