In her book, Evidence Explained, Elizabeth Shown Mills wrote that the most important reason to assess the reliability of sources is so that one can reach “the most reliable conclusions.” Critical analysis of sources is the basis of good genealogical research. It is extraordinarily rare that one’s pedigree is already well established, documented as” proven” in any substantial way. Therefore, research of historical records is the primary path to discovery of one’s ancestry.
Devising a simple methodology for critically analyzing sources is fraught with problems. It is impossible to fully assess reliability of any given source by examining it in isolation. History and experience teach us that no source is perfectly reliable, if one means by that complete and certain knowledge of facts related to past events. Sources utilized by genealogists—indeed, by historians generally—are inherently fallible allowing only a fragmentary, often biased, view of past events. From these fragments, we identify the most likely factoids among the falsities, the relevant among the irrelevant. Then, we superimpose our interpretation of those privileged records to create a reasoned construction of our genealogy. Therefore, the conclusions from our research are actually theories of origins and family history. For those of us who yearn for certainty in our research, this constructivist view can be disconcerting.
Continuing my meditations on the five-point steps of GPS, I have to address what I believe is a misstep in parsing or separating the search in reliable sources (step 1) from the evaluation of information found in the sources we privilege or collect (step 3). Whether this parsing of source from information found within it is deliberate or a poor choice of phrasing, I do not know. However, it can lead some researchers to eliminating sources that hold particular facts because in general the source is deemed unreliable. You simply cannot identify superficially which sources are reliable and which are not without examining the information found within them first. In a recent article Arlene Eakle expressed this best.
[U]nless you include the careful examination, in your own genealogical work, of printed sources and books, including those found on the Internet, as well as the original manuscript sources from which those printed works come, your ill-informed research will spread inaccuracies and false claims that will continue to confuse and discourage us. 
Eakle goes on to explain that every source needs to be examined for its facts and traditions, whether primary or hearsay, and every source factoid found must be documented for its fitness in relation to other facts. The researcher cannot do this if she has superficially excluded sources that were deemed unreliable in an earlier step.
Of course, none of this is says that generalizations cannot be made about various types of source. I will discuss a few generalizations about the strengths and weaknesses of various source types used by family historians in a later essay.
 Elizabeth Shown Mills, Evidence Explained: Citing History Sources from Artifacts to Cyberspace (Baltimore: Genealogical Pub. Co, 2007).
 Arlene H. Eakle, “The Davinci Code and Your Genealogy: Multiple Streams of Evidence,” NGS News Magazine 34 (Jan-Mar 2008).