it seems Yariv has unpublished some posts, including the one you mention. But I had read it earlier, and I also raised an eyebrow when he illustrated the "correct" way to write code, which to me just seemed to use slightly longer expression chains with identifier names that i guess were supposed to be more descriptive (they were talking about chess pieces or something). but there was nothing clearer about it at all, if you didnt know those particular names...it did not illustrate any advantage inherent in erlang that couldnt be true for python or most any other language as well.
anyway ++1 on every point you made. when writing code that is to be consumed by others, you have to question every little trick and shortcut you think is cool and/or convenient and insure it has a proper and consistent place in the overall concept. I also try to focus on "narrative" style documentation...since API docs dont say much by themselves. though I do receive some complaints that illustration through example is "inaccurate".