i can agree at some extent why it could be at 10bits/sec.
the brain is known to do some shortcuts when parsing/speed reading but slows down when we try to extract details from written works. it is also more tiring to scrutinize details than to just read articles.
The Caltech release says they derived it from "a vast amount of scientific literature" including studies of how people read and write. I think the key is going to be how they derived that number from existing studies.