Probability of a Relationship Error
I just read again the 2016 article Green Temple Hunting by Mark M. Shepherd, viewable at FSFamilyTreeUserGroup, http://fsfamilytreeusergroup.com/training_by_topic.php (search the website for “Green Temple Hunting”). I understood Mark to say that starting with anyone born before 1800, there was a 96% probability of a relationship error between that person and any third+ cousins.
I tell people that come to the FamilySearch Center to be very cautious of relationships beyond fourth and fifth cousins that they have not proven themselves. Should I? Has anyone published an update to Mark’s analysis?
I have not read that article - but from the link you posted found Terry Mason's Principles of Documentation on assuring the records we add to FamilySearch are accurate
Terry added this quote: In his October 2018 Saturday morning talk about "Truth and the Plan" By President Dallin H. Oaks, First Counselor in the First Presidency said, "We need to be cautious as we seek truth and choose sources for that search. ... If the source is anonymous or unknown, the information may also be suspect. Our personal decisions should be based on information from sources that are qualified on the subject. The methods of science lead us to what we call scientific truth."
We see too often in FamilySearch the anguish and angst that is created when the records we have researched and added, are changed, or added to .. or entire families obliterated by another FS user, who provides no documentation for the changes made!
Yes - we should be cautious of relationships that are not proven. If only everyone did that - then FamilySearch would be an even better gem that it is now.1
I tell people that come to the FamilySearch Center to be very cautious of relationships beyond fourth and fifth cousins that they have not proven themselves. Should I?
Is this generalization true? Yes. Is it true in every part of Family Tree? No. There is a world of difference between a tree with no sources and a tree with 20 sources per person that are images of historical records.
Scientists have a label for projects like Family Tree: citizen science. Unlike most citizen science, however, the work on Family Tree includes analysis and synthesis. As the tree matures, analysis and synthesis will be more and more dominant. My hope is that the global community of FT contributors will also mature enough that we can engage in content disputes like scientists do, with civility, respect, and data.
Even after years of work by many thousands of contributors, it may be that the majority of PIDs on FT still are not supported by reliable sources. FT staff could offer some statistics, perhaps. Personally, I see no point in getting into content disputes where sources are absent. That is just "Is too!" and "Is not!" Instead, I focus on finding sources that are high quality historical records.
Something I do, that you could help library patrons do, is use RelativeFinder to assess the quality of a tree. RelativeFinder has a tool to connect two deceased persons. Choose any two distant descendants in a family, print off the resulting chart, and then for each PID on the chart note (a) the generation time between parent and child and (b) the number of sources attached to each PID. That exercise will reveal some important problems with the tree. This is a small and often very powerful exercise in data analysis.1