For background: The We're Related mobile app compares persons in my Ancestry Member Tree with ancestors of famous persons (and Facebook friends) and presents then as "cousins" with a relationship based on a common ancestor.
1) Initially, I thought that "Ancestry is using my Ancestry Member Tree, and someone else's Ancestry Member Tree with the famous person included, finding common ancestors, and presenting them to me."
One of the things that Ancestry does in the We're Related app is to extend some of my own ancestral lines that are not in my Ancestry Member Tree. Others have noticed this also in their app cousin relationships.
For instance, here is one of the recent cousin relationships I was given on the We're Related mobile app:
The app says that Mitt Romney is my 9th cousin, and the common ancestor is John Mead (1634-1699).
Here is my relationship path from John Mead to my great-grandfather, Charles Auble (two screens):
According to the mobile app, my line from John Mead to Charles Auble is:
* John Mead (1634-1699)
* Benjamin Mead (1667-1746)
* Elizabeth Mead (1705-1783)
* Eunice Peck (1735-1828)
* Shubel Knapp (1757-1831)
* William Knapp (1775-1856)
* Sarah G. Knapp (1818-1900)
* Charles Auble (1849-1916)
It is logical that, since this is my ancestral line back from Charles Auble, that this line should show up in my Ancestry Member Tree used in them obile app. It doesn't. Here is a screen shot of the ancestors of Charles Auble:
As you can see, my Ancestry Member Tree does not have ancestors for William Knapp (1775-1856).
My conclusion is that Ancestry.com is NOT using my complete Ancestry Member Tree as a basis for my line in the We're Related mobile app. The mobile app has five more generations of ancestors for William Knapp (1775-1856) as shown in the app screen shots above.
2) So where does the mobile app get it's information for MY line? I have hypotheses:
a) Perhaps Ancestry.com does this:
* It uses the four generations of my ancestors that appear in the mobile app under Tree (it only shows four generations, me back to my great-grandparents, it won't show any more).
* Perhaps it takes those 8 great-grandparents from my connected Ancestry Member Tree and constructs an ancestor list based on information from many other Ancestry Member Trees (say, back to the 1500s time frame if possible - perhaps 10-12 generations back from me. 12 generations would be 4,096 potential ancestors, but in reality there are a lot fewer because very few researchers have anything near a complete tree).
* Then it compares those potential ancestors with the ancestors of the famous person (perhaps created the same way using a number of Ancestry Member Trees).
That would explain how the additional generations were added to my tree. I looked at the "Member Connect" entries for William Knapp (1775-1856) in my Ancestry Member Tree. There were seven of them, and six had his parents as Shubel Knapp. Only one got back as far as Benjamin Mead. Their search must have found other trees to go the next generation(s).
b) Or perhaps Ancestry.com does something a little different:
* Start with my full Ancestry Member Tree, and note the end-of-line ancestors.
* Do a search to find other Ancestry Member Trees with my end-of-line ancestors and add them if found.
* Compile a list of my ancestors back, say, 12 generations from my own AMT and any end-of-line search results.
* Compare my augmented ancestor list with the ancestor list of the famous person.
Either way, they have added potential ancestors to my end-of-line ancestors.
3) There is always the possibility that they are right, but there's also the possibility that they are wrong.
For instance, I've worked for 29 years looking for William Knapp's (1775-1856) parents. A book says they are Shubel Knapp (1757-1831) and Rebecca Mead (1767-1830). There are also at least six AMTs that say that. Shubel Knapp would have been age 18 at William's birth in 1775, and Rebecca would have been age 8. There is no marriage record for Shubel Knapp and Rebecca Mead, but some trees say 1782. The year range for their other children is 1786 to 1812, including another William Knapp (1786-1830).
My critical evaluation of the available information says that William Knapp (1775-1856) is not the son of Shubel and Rebecca (Mead) Knapp. Therefore, my line back to John Mead on the mobile app is wrong.
4) So this leaves me with the quandary of what to do next. I clicked on the link on the mobile app that says "Does this path look correct to you?" for my line and said "No." I don't know if that feedback will help or not. Will the Romney cousin relationship disappear from the mobile app on my phone?
5) Why doesn't the mobile app find my documented cousin relationship with Mitt Romney? I wrote about it in I'm Mitt Romney's (Distant) Cousin (posted 28 December 2011) - we are 9th cousins once removed according to my research and that of William Reitwiesner.
6) It's still a mystery how the mobile app works. I'm wondering if Ancestry.com will share their methodology with us at some point in time? It's one of my questions to ask at RootsTech!