I realize you're trying to elucidate but the context of that author's sentence was "text editors" therefore things like vi/emacs/TextPad/Notepad do not understand a binary format such as a zip file. (And docx is a zip file as you already noted.)
Yes, inside the binary zip file is a text file called "document.xml" but none of those plain text editors will parse a zip file to get to that xml file. From the viewpoint of plain text editors, a docx file is an opaque binary format.
Yeah, i see but my line of thinking was more along: "if I have this docx in 100 years time will I be able to decode its contents?"
I think the answer is probably yes (pending possible " "premature" apocalyptic scenarios =)
Regarding the zip compression, docx is a "Document Container File" standardized under ISO/IEC, so provided we still have computers and the standard documentation is still know, it should in theory be possible to DEFLATE the file.
Inside you would find all the text content as xml, and as some have pointed out here, binary blobs as well. Those would be mostly font files and image files (or am I missing some others that would be important?). In my opinion the fonts would not be critical to the content meaning.
The images would of course depend on what format they were encoded in. But that's a whole other discussion!
.docx is not a binary file format unlike the .doc files from the past.
.docx is a plain text xml based format that has an Open published standard.
Most people don't know this, but you can simply unzip the contents of a docx or xlsx file and inspect its plain text contents.
Archival and long term accessibility of the document contents should not be the deciding factor in opting out of .docx