My father wrote a BBC Basic interpreter for MS-DOS and lectured at Coventry University for a while. I remember discussing the Dijksra article with him. He agreed that goto should generally be avoided, but said he’d used them for very deep loops where a for loop would have run out of memory. That basic interpreter was written in assembler, so he was very used to managing program structure manually with no high level constructs. On early computers, and even today, sometimes there are constraints other than clarity of the code that need to be considered. Having said that, I've not used a goto statement for at least 30 years.