You missed the fact that the real Word does open this file just fine, its just the toy web version that has issues (and maybe Mac too but eh)
But this really goes more into the facet of Office files that allowed embedding pretty much anything into them, and relying on this "filter" system (I guess OLE) to handle embedded objects. So while the DOC file itself is getting parsed and rendered pretty much perfectly, the embedded objects are another story.
In the same sense I'd say browser might open some HTML page "fine" even if it doesn't know how to handle some image format that is used on the page; it'd still handles the HTML correctly.
You can recover text but the result is horrible. No graphics and all formatting lost.