Many of the source documents being posted to the Internet in great number consist entirely of transcribed document or indexes. For example, the recent posting to FamilySearch Record Search Pilot was four indexes of the Canadian Censuses. The indexed entries are not accompanied by the corresponding images of the original documents. I have to acknowledge that an index is a very useful tool, but all indexes are, at least, one entire step removed from the original source. Even if I find a useful entry, I will still feel obligated to look at the original source record if it is available.
One recent example, in a Census record the birthplace was listed in the index as "New York." However, an examination of the original image from the Census record showed the place as "New York City" crossed out, thus giving a valuable clue that was left out of the transcribed document. It is obvious that original information on any document be inaccurate or incorrect but it also obvious that any time the document is transcribed or indexed there may be additional mistakes.
In addition to the example above concerning Record Search Pilot, many other online resources advertise access to millions or even billions of records. However, upon examination, it is clear that in some cases, the so-called "records" are really transcribed indexes or lists rather than images of the original documents. This is not altogether bad, especially if the original record was a list or compilation. But each time someone looks at a record and makes a decision about what it says, there is a danger that mistakes will be made. It is only logical that the fewer people are in this line of transcription, the greater chance that the record is accurate.
The Rules of Evidence used in our court system recognize the lack of reliability of secondary or derivative sources through the rules on hearsay, that is, secondary testimony about what someone else has said or written. The exceptions to the rule against hearsay testimony are allowed only when there are other reasons to believe the validity of the record or conversation. In almost every case, a transcript or index of original records would be considered hearsay and only allowed into court under one of the various exceptions.
A great deal of the recent discussion about genealogical research centers on the need for sources. Elizabeth Shown Mills' major work, Evidence Explained, discusses the different levels of legal proof and also the difference between original and derivative sources. In paragraph 1.23, Mills concludes that "Neither databases nor indexes are records per se." See
Mills, Elizabeth Shown. Evidence Explained: Citing History Sources from Artifacts to Cyberspace. Baltimore, Md: Genealogical Pub. Co, 2007.
It is indeed fortunate the digitized copies of original source documents are becoming more and more common on the Internet. But, it is likewise important to always remember the distinction between the original record and any derivative record. This issue has come forcibly to the forefront for me on a daily basis in giving support to people using New FamilySearch. Among those who view the information online, there is a general tendency to accept what is there as "fact" rather than recognizing how far the information is removed from any actual source. I frequently get requests from individuals asking me to help them "copy their genealogy down from New FamilySearch." In some cases, I have had to turn over the question to someone else because I did not feel that I could answer the person without offending them. I see the same tendency in information obtained from the large online databases. Once the desired information is found, in whatever format, it is merely accepted as a fact.
As recently as today, I was talking with a researcher who had totally contradictory evidence from two different sources. However, rather than evaluate the sources, she accepted both as true and was asking me to support her conclusion.
If I could not rely on information to get passed the hearsay rule of evidence in court, I certainly would not accept the information as probative, without sufficient collaboration.