I was inclined to disagree due to the author's strong arguments and precise use of language, but after re-reading the part around Dr. Richard's comments, I do think the author made a bit of a logical leap.
It went from "I'm Dr. Richard, and BCPL supports pointer arithmetic starting from zero" to "I'm the author, and the IBM 7094 handles all the pointer arithmetic during compile time, and time is limited; therefore, Dr Richard added zero-based indexing to speed up compile time." When really, his intentions are assumed by the author. Dr. Richard sounds like a very pure computer scientist. He may have added that feature without consideration of when pointer arithmetic is calculated simply because he wanted it in his BCPL language.
Really, Dr. Richard seems to sum up the issue nicely, "I can see no sensible reason why the first element of a BCPL array should have subscript one."
Also, wasn't the original question why zero vs one, not, why we started using pointer arithmetic at all?
It went from "I'm Dr. Richard, and BCPL supports pointer arithmetic starting from zero" to "I'm the author, and the IBM 7094 handles all the pointer arithmetic during compile time, and time is limited; therefore, Dr Richard added zero-based indexing to speed up compile time." When really, his intentions are assumed by the author. Dr. Richard sounds like a very pure computer scientist. He may have added that feature without consideration of when pointer arithmetic is calculated simply because he wanted it in his BCPL language.
Really, Dr. Richard seems to sum up the issue nicely, "I can see no sensible reason why the first element of a BCPL array should have subscript one."
Also, wasn't the original question why zero vs one, not, why we started using pointer arithmetic at all?