Monday, May 31, 2010

“Evidence: There’s No Better Rule”

See “Evidence Management” for an overview of this series and for links to other articles.

“You [are not] responsible for my mistakes and wrong conclusions,” [said Pip.]

“Not a particle of evidence, Pip,” said Mr. Jaggers, shaking his head and gathering up his skirts. “Take nothing on its looks; take every thing on evidence. There’s no better rule.”1


I’ve discussed the immaturity of the source summaries of and Today, I will examine the embryonic nature of their evidence management and their total lack of evidence summaries (the green boxes in the diagram, below).

The Evidence Management Diagram
Evidence can be managed when
separated from sources and conclusions

Current genealogy programs lack separate treatment of evidence. Source summaries and records of individuals are left struggling to fill the gap. Some aspects are picked up by each, as shown in the diagram below. Contrast this with the evidence management diagram (above).

Diagram of genealogy programs that lack evidence management
Current genealogy programs suffer because
evidence is not managed separately

One may well ask if it matters that users follow established research processes and proven techniques for evidence analysis. Consider the following:

  • User retention goes up when success goes up. Conversely, research errors are demotivating at best and—when publicly and embarrassingly discovered—can be humiliating.
  • Genealogy practitioners have invested tens of thousands of hours distilling successes and failures into a research process that repeatably yields positive results.
  • Subscribers whose retention is most at risk are the very users least likely to follow successful research practices without guidance.
  • Software that reflects and enables successful research practices will lead to greater research success and greater subscriber retention.

Think back to the example in “Evidence Management Explained” where the month and day for the conclusion came from one source and the year came from several other sources. Compare the two diagrams above, while you think about these questions:

  • When evidence and conclusion are one and the same, what do you do when making a conclusion that is different from the evidence in any one source?
    • Do you enter a primary conclusion that is not linked to any source?
    • Do you link the primary conclusion to a source that states something different from the conclusion?
  • When evidence and sources are one and the same, what do you do when the source contains both primary and secondary information?
  • When evidence and conclusion are one and the same, what do you do with contradictory evidence?
    • Do you enter an erroneous, alternate conclusion?
    • Do you leave the contradictory evidence unconnected to the conclusion?
    • Do you enter source notes explaining the contradictory evidence?
    • What if the source also contains supportive evidence?

Further, when evidence is not managed separately from sources and conclusions:

  • Reports can not be generated about the evidences.
  • Evidence lists can’t be generated.
  • Evidence can’t be flagged as primary/secondary, direct/indirect, or supporting/contradicting.
  • Lists can’t be sorted by characteristic.
  • Lists can’t be grouped by geography or by time period.
  • Lists can’t be printed showing evidence that requires additional work or special handling.
  • Should you discover you misidentified a person, it takes a lot of manual work to move evidence from one individual to another.

Evidence. There’s no better rule.


     1.  Charles Dickens, Great Expectations (Boston: Estes and Lauriat, 1881), 366; digital images, Google Books ( : digitized 19 March 2008).

1 comment:

  1. AI, this is your best yet.

    The emphasis on evidence-handling skips over the problem with all trees, including those on and New Familysearch. For it is also the problem with a large number of the non-tree databases, and shares this problem with the majority of IGI entries at (which are the principal basis for the Nw Familysearch tree).

    This is that the ~source~ is a ~conclusion~ made on the basis of some information known only to the submitter or compiler of, say, a genealogical publication or a family group sheet -- or a tree.

    The trees, of course, are vastly complicated because so many are just gedcoms or other trees' pieces lifted from who-knows-where. It is quite apparent that very often the tree-cobbler/uploader does not bother to review the material even for sense.

    For this stuff, the idea of unpacking the 'source' to evaluate 'evidence' is impossible because there is no intrinsic pointer to there having been any evidence to begin with.

    It is true that if one is familiar with existing literature and records concerning a given family group, one can detect tampering and distortions as well as mere repetition of Widely Held Mistaken Beliefs perpetrated by "believers in" sourceless conclusions from a genealogy that may have been recopied many times in other non-researched genealogies (including, of course, numerous trees).

    One of the bizarre tactics I have seen is omission of such data as dates and places from specific records, where they conflict with the objective of the tree-uploader. One example is omitting a date of marriage for people already dead, but nonetheless attributing the maiden name of the bride from the marriage record to the already-dead wife of same-named husband. Another is omitting other children and location of a Rhode Island couple in order to make one child and the name of the father pseudo-plausibly fit a family in Kent County, Delaware wherein the father died testate.

    Of course, evidence-handling software techniques cannot deal with this fundamental dishonesty. They also will not affect the triviality and untruthiness of the vast majority of trees.


Note: Only a member of this blog may post a comment.