Kieran here. Can't view the main article for it, so this may have already been discussed there.
For what it's worth, most of the passwords I use are truly random. But yes, of course using any weak password is a poor choice.
The applied cryptography sources you mention are probably using the theory published by NIST, which has been shown to be fundamentally flawed for real-world passwords . According to NIST, the first character is worth about 4 bits, the next seven characters around 2 bits and it drops after that. Fortunately, in the real-world things are a bit better (pun intended).