Require that a CONFIDENCE level be included when adding a source to the Family Tree
FamilySearch should consider requiring a confidence level (high, medium, low) each time a source is added to the family tree. If implemented correctly this would help improve the accuracy of the tree. For example, if a user enters a birth date/place and cites a birth certificate as a source, that would be an example of high confidence. If they cite the person's death certificate that lists their birth year, that would be medium. Low confidence would be given to uncited sources like history books that can't be corraborated with other sources. Since everyone's definition of high, medium, and low will be different, take the guess work out of it by implementing a pop-up menu that has an extensive list & let the user choose the appropriate one. Once chosen the corresponding confidence level will be applied.
Then there are cases where websites such as find-a-grave are used as sources. That is okay provided a gravestone exists. However, it should be a medium confidence source, because it assumes that the stone mason spelled the name right and got the dates correct. In cases where a readable gravestone does not exist and/or a cemetery plot map doesn't show the person being buried there, find-a-grave should not be cited, but if it is, no higher than low confidence.
Then there is the special case when there are no direct sources exist and indirect sources combined with reasoning is used in an effort to fill in the blanks. One example of this can be seen in Memories for Sgt. Nathaniel Washburn (ID LHKN-L96) entitled "Strategy for locating Nathaniel Washburn's land and possibly his home - mill - and grave in Adams County, Ohio". In this case the pop-up menu would need to have an "Other" category and its confidence level would have to be established by someone at FamilySearch after reviewing the memory or other form of reasoning.
When viewing the list of sources attached to a record, color-coding each source (green = high confidence, yellow = medium confidence, red = low confidence, gray = undefined/tbd confidence) would be an additional enhancement.
Comments
-
I disagree (strongly) that this would be of any use whatsoever. People would just check "high" on everything and move on.
3 -
Julia, I agree that if left up to the people to choose high, medium, and low, they'd do just what you say. That's why I said "take the guess work out of it by implementing a pop-up menu that has an extensive list & let the user choose the appropriate one. Once chosen the corresponding confidence level will be applied." Obviously, if people are so inclined to pick the wrong one from the pop-up menu, then it would take another user to correct it. But at least they wouldn't be guessing if its high, medium, or low.
If you have a better solution to fool-proofing the Family Tree please offer. More brains are better than 1.
0 -
I am just another 'lowly' User/Patron ...
I understand, where you are coming from; and, what you are trying to achieve ...
But ...
That said ...
I am with 'Julie' on this one ...
As a "Programmer" once said to me ...
You CANNOT make a "System" FOOLPROOF; because, Fools are so INGENIOUS ...
And, exactly, what 'Julie' said, Users/Patrons will just mark EVERYTHING as "High", they will want that; think it is for the best; and, will also take the easiest route.
That is 'human nature' ...
Such is life ...
Just my thoughts.
Brett
ps: Great in 'Theory', just not so good in 'Practice' ...
1 -
Confidence levels are a very complex subject.
To start with, they are not a property of sources in isolation. A marriage certificate can be 100% accurate in every detail, but if it's connected to the wrong person, it's completely untrustworthy. Ditto for every other type of genealogical source, from Wikipedia to original church registers.
Confidence is more usefully framed as a property of individual conclusions, but even there, it's not straightforward. You can't determine confidence based on the number of sources that agree on it, nor on their type. For example, I have an ancestor who reported her sister's birthdate as her own at both of her marriages. I only have one source -- her baptism -- for her actual birthdate, but considering her sister's marriage and death records, I'm confident that the baptism is correct, and the birthdate from her marriages is not.
Obviously, no computer database or algorithm can usefully assign a confidence level to a conclusion that was reached based on a combination of clues, and a rating of the individual documents that provided those clues is also unlikely to be of any practical use.
Some genealogy software includes the ability to rate or classify sources, but entering and maintaining the ratings is a lot of work, and they fail to prevent mistakes, so I really, really see no point to them.
3