Conrad Dunkerson wrote:
Indeed, that text still appears in our 'No original research' policy. However, efforts to stamp out use of 'primary sources' to spread information that no other national (or international) 'news' / 'reporting' entity has deemed worthy of commenting on have led to a wide-spread view that 'primary sources' in general are bad. They aren't. Once something has been verified as notable we should often take primary sources OVER secondary ones.
I strongly disagree with that, and think this comes out of an unfortunately widespread view that non-scientific research isn't "really" research. Gathering, interpreting, cross-referencing, and checking the validity of the primary sources on an individual like, say, Thomas Jefferson, in order to write a biography about him, is original historical research, and best left to reputable historians. At Wikipedia, we should prefer secondary sources on his life---published biographies of Thomas Jefferson written by reputable historians. If you discover some new primary sources relating to his life that have not been mentioned in the existing secondary sources, that constitutes original historical research, and you should publish it in a history journal or book, or at the very least convince someone to write a newspaper article about it, before it should go into Wikipedia.
The same actually goes for too-close-to-primary secondary sources. We should not write our article on World War II by referring to contemporary newspaper reports, which were often wrong and require expertise to properly use, but instead should write it by referring to existing, published histories of World War II written by reputable historians.
-Mark