Detangling profiles?
At what point do you give up trying to return profiles to their originally intended persona and just force-fit them into a currently accurate form?
I've spent the last week trying to sort out four profiles (before I make any changes) which I believe are trying to represent three different individuals. However these four profiles are the result of merges of (at least) 24 profiles which I think belong to (at least) 8 different people. Each of the profiles has details / sources / spouses / children that really belong to one, or more, of the others. And every one of them started out as someone different before they morphed, or were hijacked, into who they appear to be now.
Archibald MacArthur G381-W4H
Archibald MacArthur GPXH-WBS
Archibald MacArthur 9S82-SWP
Archibald MacArthur L2D8-4WL
Their wives are in a similar state with merges resulting in one of them giving birth at ten years old, and a couple of others well into their sixties. Having two children, three months apart, on two different continents, not a problem!
Answers
-
The only time I give up is when it becomes clear that some of the merges took place in New Family Search so that it is impossible to completely separate all the profiles. But then I also find various sort of logic problems really fun and so enjoy this kind of challenge.
However, when I run into the kind of tangle you are describing, I turn to pen and paper. A large piece of paper. Or even a white board. There is no computer screen, or even sets of screens, large enough to handle this.
First I go through the entire chain of merges with all its side branches and write down very single ID that was ever merged in so I don't lose any of them and leave plenty of space between each one.
Then I restore every ID. Sometimes I can tell in the change log that some old merges were clearly correct and I don't restore the IDs involved in those.
Next I go to the very first entries in the Change Logs of everyone and note on my paper the basic information that shows who each ID was intended to be. This can be the original birth information or who his spouse or children are. Using this information, I decide how many people are actually involved here, whether the eight you have found so far or more.
The next step might be a bit controversial. Usually I find it extremely helpful to create a new profile for each of the individuals I have identified. Yes, this creates a new change log that might bury old work on the profiles but most of that work generally turns out to be badly done. I give these profiles easily identifiable names by putting in a temporary Title and/or Suffix such as "First Archibald MacArthur of New York."
Now the work starts. I pick one of the original IDs and go to the person's youngest child. I have found that this is the whole key to making this work. At this point the child will likely have multiple sets of parents. I go to the very first entries of the child's change log to find the parents this profile was originally created with. Sometimes this gives me more ID to add to my list if a parent-child relationship was deleted rather than parents having been merged.
I evaluate the child's sources to determine if those very first original parents were correct. I work to make the child's profile the best I can and delete all the parent-child relationships that were created through incorrect merges and add back any parent child relationships that were just deleted. Then I will add that child as a child of the new, clean copy duplicate of that child's real father. This leaves the child with multiple sets of parents: 1) the ones that went through all the incorrect merges which were reversed and still have a bunch of incorrect spouses and children and 2) the one that is the new profile I created which at this point will just have basic identifying information. I don't worry about incorrectly attached source at this point and I definitely ignore all Research Helps.
Then I go back to that first ID I had picked and start on the next youngest child. As I finish a group of children, I then evaluate the mother that was associated with that group and get that mother linked to only her correct copies of her husband and her correct children.
Generally, after working through all the children it becomes very clear which groups of the unmerged IDs should now be remerged based on their now correctly assigned children and spouses. Before remerging, I check the information on each ID and remove incorrect information that had been placed there through bad merges. I still at this point ignore any suggested possible duplicates under Research Helps since those are probably what caused all the trouble in the first place and only use Merge By ID. With the new profiles I created, I make a careful judgement call as to whether it is better keep it as the final profile or to merge it into one of the corresponding existing profiles. With each new merge I note on my paper where each ID goes.
After correctly remerging the 24 or more profiles you have found into the 8 people you think you are dealing with, or whatever turns out to be the correct situation, I check that each child now has only one set of parents and that each correct couple are fully combined.
Only now do I turn to source attachments. I go through each source using the Review Attachments link and make sure the source is attached to correct individual and only to them.
Finally, I make sure all Research Helps are appropriately address or dismissed.
This type of organized, regimented, step-by-step procedure usually makes it possible to repair even the worst these bad tangles.
Good luck!
4 -
A point I have expressed here previously is my feeling that conflated profiles are far more serious an issue (within Family Tree) than duplicates. Fortunately, I have had time on my hands over the last ten years, so have been able to spend several days at a time dealing with matters like this - albeit nowhere near as tangled as the example provided!
So, my answer to your first question would be to give as much time to the issue that you can afford. I have had to walk away from such an exercise on one or two occasions, but (if this has become the case) I add a note in the Collaboration section explaining what work I have carried out and what is / appears to be outstanding. For example, I would note that I had separated individuals who lived at the same time, but who it was obvious had never lived within a hundred miles of each other. At the same time, I would explain that a couple of conflated profiles (identities very similar) still remain under the same ID, so there was still further work to be carried out before separating them.
In general terms, I do find myself completely exasperated by careless users (who even claim to have a lot of experience, when I have contacted them!) who confuse a John Smith who never left British shores with one who lived their whole life in the U.S.
But I do have some empathy here - having spent all week trying to establish the separate identities (through census, marriage records, etc.) of eight individuals named Jane Fletcher, who were all born within a few years of each other and did live their whole lives within a ten mile radius in north Yorkshire! Fortunately, none of their IDs had been conflated, so (after so much work) I have added a "Not to be confused with…" note to each profile and added them to my Following list! I'll still be very surprised if another user doesn't try to merge at least two or three of them, though!
2