That sure is a lot of words to say 'because indices are memory offsets'.
The author also addresses the importance of history. Voodoo knowledge _does_ pervade modern computer science, and I for one am happy to see something different.
But is it really? I'm inclined towards anonymouz's and adamnemecek's idea here: Dr Richard clearly says indices are offsets, no? It's nice the author gets all sentimental about it, but that doesn't make him right.