Limit merges - Merge alert for Macrae/Mcrae

LegacyUser

ClareLM said: Hello,

I'd like to suggest that there is a limit of say "3" for the number of times a record can be merged within say 6 months.

Once again a well-meaning user, this time definitely a member of the church, so probably a volunteer, has decided to merge all the records of Macraes/Macraes with the same Christian name.

I'm not sure how many members of the community will be familiar with the genealogy of the Clans of the Highlands of Scotland and the associated complexities so here is a link to a book History of Clan Macrae - https://archive.org/embed/historyofcl... which shows of the issues.

Let me highlight just a few:
- Christian names are used over and over. For example, there were probably 100,000's of Christopher Macraes in the Highlands during the period 1500 to 1900.
- there are very few sources available for births, deaths and marriages, to distinguish for example one Christopher Macrae from another.
- Marriages in the Clans were frequently between first and second cousins eg Alexander Macrae may have a wife Mary Macrae where Macrae is her MAIDEN name as well as her MARRIED name, so she is not the same person as a Mary Matheson or a Mary McDonald etc.

In FamilySearch there are lots of record for MacRaes with the same Christian name that have little or no information in them apart from the name of the spouse and the names of the children. These records are mostly NOT duplicates and where they MIGHT be duplicates it is near IMPOSSIBLE to determine if this is the case because of the lack of sources.

So in the past week or so, a well-meaning person has been working very hard to merge all the Janet Macraes, Alexander Macraes and Christopher Macraes and possibly others. I know of the Janets, Alexanders and Christophers because these are the names of my ancestors whose records had been merged with multiple other records of the same name.

I would like to share this particular record as an example Christopher ( Baine) Mc Rae K4G1-9LH. If you look in the Other information in this record you will even see
Alternate Name •
Birth Name
Christopher Mac Rae Or Christopher Colinson
Last Changed: July 24, 2020
This particular Christopher has been merged something like FIFTY times. (The list is so long that it was difficult for me to count them accurately when scrolling.) And this Christopher also has a number of different wives with different names as well as many unknown.

This same user merged my ancestor Janet Macrae, a wife of a Christopher Ban MacRae, with a Janet Reidford and in the process change Janet Reidford's surname to Macrae, also filled in the blanks birth and deaths in Janet Reidford's record with those of my ancestor. What can you say? I have been messaging this user, and they informed me that they have 14 years of experience!

In addition to all of the above, once this kind of thing occurs, it is incredibly difficult to UNDO. A restore does NOT restore everything to it's original state. And the more times a record is merged it becomes exponentially more difficult to restore. When you restore two merged records one thing that can happen is the the children are not restored to their original parents. Sometimes they are restored to the wrong family, sometimes they stay with both. Also mostly the the items under Other Information end up in the wrong record etc. Sources, if there are any, can also end up attached to the two records which increases the likelihood of them being incorrectly merged again in the future.

So surely, merging can be tracked and when a record has been merged more than a certain number of times - 3 would be a good number - then the record is locked so that no more merges can be done for say, 6 months or before contacting some other people who have worked on this record, or contacting people who are watching the record.

Come on, surely FamilySearch can better manage this ongoing problem of merges. At the very least, they should not allow volunteers to do this work without supervision unless they have proven they know what they are doing.

If you've reach here, I thank you for listening. I spent three hours last night just restoring the records of 3 of my ancestors. I will takes weeks to undo what has been done and the person has responded that "it is a work in progress" so they obviously plan to continue merging. I've requested that they stop and I've explained all of the above but the person remains blissfully ignorant of the damage that is being done.

Regards
Clare
PS I follow all the records in my extended family tree. I feel deeply sadden by what has been done to the trees of the other users many of whom will not be following their records and may never find out what has happen.

LegacyUser

Jordi Kloosterboer said: Sometimes there are legitimate many duplicates for one person, So I say not to put a limit on it.

LegacyUser

Juli said: Uh, yeah, no. Absolutely not.

If a set of parents in 19th century Hungary or Slovakia had five daughters baptised, and FS has both the original church register and the bishop's copy, and both were indexed, then there are a minimum of ten profiles for each parent floating in Family Tree, waiting to be properly merged together. And that's a relatively small group compared to some of the possibilities in England.

LegacyUser

ClareLM said: Jordi and Juli,

Thanks for your replies. It is good to hear of others experience. So if there could be 10 legitimate profiles in the system then maybe a limit of 12 would be good.

But something has got to be done to stop the merging of a large number of records.

This is the second time in as many years that someone has attempted to merge all the Macraes with the same Christian name. Last time when I sent messages to the person, they just ignored my messages and continued merging. Even after I had un-merged and noted reasons why eg spouses maiden names differ, they went back and remerged records. This went on for weeks and I repaired a lot of other people's trees where I could. Where I had difficulties I sent messages to other contributors.

This time, I do not have the time to do any repairs other than on my ancestors.

Regards

LegacyUser

Paul said: I'm dubious as to whether the idea suggested is practicable. However, once again it highlights a problem many of us experience on a regular basis.

There are so many clueless users who are causing serious damage to the whole Family Tree project. Part of this is down to overemphasis on the need to merge duplicates. In fact, I believe the major aim is for there to be one record (ID) for as many individuals as possible who ever lived on the planet.

As a result, there are plenty of overzealous individuals who carry out exactly the work described above. The negative consequence is that thousands of persons are being lost - merged into another ID of different identity. But for another, conscientious user coming along and putting right these errors, it is as if these individuals never lived on Earth.

This is where the use of Family Tree as a basis for creating ordinances can cause serious conflict with its use as a "general" genealogical database. Whilst like LDS Church members I have worked hard at reducing the number of duplicate IDs, I really think the determined drive to do this is having too much of a negative effect and eventually will have to be addressed in some way. And, no, not by writing polite notes to the erring users who are creating the problem - they either ignore you completely or respond with a polite "thank you for dealing with this", completely oblivious (or are they?) of how much time they have wasted and/or the potential / real damage they have done.

For those who would reject Clare's idea out of hand - any other positive ideas? I don't think I've seen any serious solutions offered in my 8-9 years in "dealing" with the problem. I know what I would prefer if it is a case of my ancestor being wiped-out completely from Family Tree or having a number of duplicates floating around for now.

LegacyUser

Barbara Nelson said: I actually like the idea of that kind of low number limit. It would be more system overhead, but how about asking if they are merging from an extracted record? If the answer is yes, the merge continues.

If not, the user is asked to specify what matches. Yes, they may put in gibberish. But, if they do, or if they falsely state is it an extracted record, then there is a conversation that needs to take place. Probably first a message from the person who catches the bad merge, but if necessary, by LDS personnel.

LegacyUser

Christina Sachs Wagner said: Sorry, I can't get on board with this idea. Last night, I found a couple with nine children. Each child was created using a birth/christening record, so no other information for the parents except their names was available and the mother only had her first name visible. Consequently, both parents had 9 duplicates and possibly a 10th, but I can't yet confirm she belongs to this couple, since the mother's name is only listed as Mary. I was cleaning up all the duplicates, very late last night.

The good news is there is a warning posted when a person has been merged several times and you need to be sure you are merging them correctly when you are on the merging page for yet another.

LegacyUser

Juli said: As I wrote at the end of my post, 10 is a small number compared to the possibilities in England, where many areas have profiles for all baptisms (not just the girls), plus marriages, sometimes in duplicate or even triplicate (because both the bishop's and archbishop's copies were indexed). This means that a parent can end up with 30 profiles.

LegacyUser

Juli said: The sort of user who merges profiles just because the names match is likely to be totally clueless about what "extracted" means. He would cheerfully answer "yes" to such a question, believing himself to be truthful.

LegacyUser

Christina Sachs Wagner said: Agreed! A warning is posted in the merging process, if a PID has had a lot of merges, so there is already a warning in place for users to be careful.

LegacyUser

Christina Sachs Wagner said: Have you marked the ones that should not be merged as not a match, with a detailed explanation? Detailed being the operative word. It doesn't always eliminate bad merges, but does give subsequent merge attempts a warning to check further.

LegacyUser

Adrian Bruce said: Agreed, Juli.

LegacyUser

Adrian Bruce said: It has been suggested many times that there should be some sort of impedance on merges where there is a genealogically significant difference between the two profiles.

For instance, the system detects that while the proposed merge is for two profiles named "Christopher Macrae", the two lived a century apart or have differently named wives, and says: "This is unlikely to be correct" or, "Are you sure about this?", etc.

We need to be able to merge profiles with different wives because people could remarry after the death of one spouse, but the great advantage is that many of these people follow the data blindly on the basis that "The computer says yes" (i.e. it's a hint) and if the computer actually says "No", then they are quite likely to equally blindly accept that and go away.

Incidentally I suspect that many of the Christopher Macrae profiles have been created from extraction programmes working on parish registers. As a result, they have no dates or places on them but those dates and places ought to be on the children whose baptism generated the parent's profile. As a result, I suspect that the "Are you sure?" impedance needs to review the dates and places on the child, because a Christopher Macrae with a child baptised in 1750 isn't going to be the same Christopher Macrae who has a child baptised in 1850.

In my view, the only way forward is (a) improving the quality of hints in the first place and (b) adding impedance checks on merges (because people can go looking for unhinted merges). Everything else will just brass off people who are trying to merge the 20, 30 or 40 profiles that genuinely make up one real life person.

LegacyUser

ClareLM said: Hi Christina, Clearly people do not pay attention to these merges. Regards Clare

LegacyUser

ClareLM said: Hello Christina,
Are you aware of the naming conventions of the Scottish Highlanders and hence of the number of possible matches that the system will display? It would take a very long time to perform this task. So for example, my ancestor Janet Macrae (that's her maiden name) married a Christopher Macrae. The system matches Janet Macrae with every other Janet who married a Christoper Macrae ie Janet Matheson, Janet McDonald, Janet Gordon, Janet McLeod etc. Note also that Christopher is a very common name. In the world of Highlanders its the Christian name version of Smith. Christopher, Janet, Mary, Dougal, Donald, Duncan, Alexander, John, Isabella, Christina, Ellen, Finlay are all like this and many of the other clan also used all these same names. This is an exponential problem.

Add this to the fact that the dates of birth, death and marriage are likely to be unknown for both parties in all these situations even if parentage is known. The system has no way to distinguish whether the records for these people are the same. Similarly, neither does the a user.

Please go and have a look at the book https://archive.org/stream/historyofc... History of Clan Macrae that is freely available on archive. org. This will help you to understand the complexity of the problem.

Regards
Clare

LegacyUser

ClareLM said: Sorry everyone for some reason the link to the History of Clan Macrae was not contracted. Here it is again https://archive.org/stream/historyofc.... The book is freely available on archive.org. Just put the title in the search bar on the home page.

LegacyUser

Christina Sachs Wagner said: But limiting merges is not the answer.

LegacyUser

Christina Sachs Wagner said: I'm sure it's very frustrating, but limiting merges is not the answer to this dilemma.

LegacyUser

Robert Ellison said: Question, What if you don’t want FS Computer merging repeated names?
The owner of the tree in this condition need to add data that can be managed,
By them; like numbering each candy in the jar each will be different and the
Owner can put them in the right order as they know it? This I believe would stop
Lot of data damage. The owner should make each sur-name whether info is
Known or not a specific identifier in the Sur-name for identifications, who they
Are in reference to others. Add to the sur-name; ex (John Doe) I add (Doe, John 01) To each name that’s in my generational line, (GF-01, GGf-02, GGGf-03 etc.) (Doe, John 03 - thus when searching index names and number or modified name; Like Doe, John 01-Peter, after the sur-name and (Doe, Peter 08 Mom b1890), Doe, John comes up on my index, depending where you put in comma in the sur-name. Different combination can be done to each sur-name; ex: my GF, Doe, John bill Brother or Doe, Alice Dau father, add to comments explain? Owner can make an Index list as they change the sur-name entries as who they are for fast access, see What person you want to look up, by seeing the sur-name entered, Local search file Command on your computer in Word, excel? Will give you quick access to see Sur-name to search, you have control, the tree will have few extra numbers-, Sur-name index is the key? r072720 You control the name,
What do you think:)

LegacyUser

Paul said: Clare

Christina makes a good point about needing to mark the Possible Duplicates as "Not a Match". As quickly as you detach a person they will reappear on the page as a possible duplicate. Unless you immediately mark this as "Not a Match" another user is likely to come along soon after and (as many careless users do) accept "FamilySearch's" suggestion at face value - and you'll be back to "square 1".

I'm sure the last thing you want is still extra work, but I believe this action to be very important in preventing further merges, but is easily overlooked - especially when you are anxious to get on and separate the next goodness-knows-how-many incorrect merges involving other IDs.

In general, the FamilySearch algorithm will be performing well in finding these very similar, possible matches. Unfortunately, you cannot expect it to be tweaked to address the very specific problem that applies to these Scottish Highland families.

The other "solution" (often suggested on this forum) is for sanctions to be applied to those who seem to be deliberately carrying out damaging work - i.e. continuing this even after several warnings / advice about their work, made in our personal messages to them. Unfortunately, FamilySearch appears unwilling to provide the resources for investigating such matters and possibly applying sanctions to such individuals - even when their reckless efforts create huge amounts of unplanned work for the likes of you and me.

I share your exasperation (many times throughout each year!), but see ways of tackling the problem to be very limited - given that, in other cases, genuine duplicates could be involved.

LegacyUser

Adrian Bruce said: A thought - do we actually know if these same-name-same-person-NOT merges are the result of hints of possible duplicates that have been accepted without effective checking? Or have the users actively gone looking for same name pairs? Right now I can't think if there is any sign that a hint was the origin of a merge....

LegacyUser

Adrian Bruce said: I'm curious about the extent to which this is a Scottish Highlands issue. Although the name pool might indeed be reduced if the population follows the "Scottish naming scheme", there is, so far as I know, a corresponding increase in differences because the mother's maiden name is given as a matter of course - whereas in England, MMNs are a rarity in parish registers.

Mind you - I don't know if baptism registers always give the mother's name in the Highlands. That would be a real **** if they didn't, especially since every 3rd son is named after his father if the Pattern is followed.

Maybe it's not just a reduction in the number of first names that drives these issues, but a reduction in the number of surnames in a parish??? This isn't just idle curiositcy but an analytical view of where the confusion comes from might help FS to review their algorithms.

LegacyUser

Paul said: From my viewpoint, Clare's problems do seem to be compounded by the issue of marriages of 1st/2nd cousins of the same surname - this seemingly being prevalent in that part of the world.

My personal experiences have led to me only rarely encountering that specific issue: there being at least two instances of my LATHAN relatives, of Sunderland, having married 1st cousins of that name.

There have long been jokes about a certain county (which shall remain nameless, but is in the East Anglia region of England) where relationships between close relatives is particularly common, but I have not come across this in my personal, genealogical research. Hence my feeling that the problem with Scottish Highland families was indeed probably worse than for other areas, particularly within the UK.

LegacyUser

Juli said: Um, this is a communal and open-edit Family Tree. There are no "owners".

LegacyUser

Gordon Collett said: Corrupting names with additional non-name reference systems is generally a really bad idea, particularly in a public database. Not only will this wreck the search system but your Doe, Jon 01 is going to be somebody else's Doe, John 5 or even someone else's Doe, John R17K. Names have to remain just names.

Even in a private database it is best to leave the names alone and come up with an extra data field to add whatever type of reference system you want to use.

LegacyUser

ClareLM said: Hello everyone,

I have an update on my merging friend. I had unmerged my ancestor for all the other Alexander Mcraes . I had also sent a message to the merger to please not merge the record. This is what the record now looks like. My ancestor now has 5 wives and I know for sure that he only married once.

What can be done to stop this person? When this happened a couple of years ago, I tried several times to get assistance from support. They suggested that I negotiate. Negotiation is what you do with your cousins who have differing information. And I've tried to negotiate with this person and they steadfastly refuse. Previously, when this happened I also tried reporting this as abuse but support does not consider this kind of behaviour as abuse.

I will restore my ancestor's record but I fear it will only be a matter of day before once again I am faced with this same situation.

Ayya Esika

LegacyUser

ClareLM said: And now that I've restored this record, Alexander Macrae LYDQ-VBC there are zero possible duplicates.

What is not shown in the above image is that my ancestor married a Mary Murchinson and the only other information I have is the names of his parents, where he died and one of his children.

LegacyUser

Paul said: Clare

You have done as much as you can to prevent any more damaging work being carried out on this branch. If this persists, I believe you are entitled to feel truly exasperated if nobody at FamilySearch is willing to intervene in the matter.

If you break the code of conduct relating to this (GetSat) website, a moderator can remove your post and even suspend your participation in the forum. I do not understand why it is not possible for sanctions to be taken against those who continually abuse their use of the Family Tree program.

LegacyUser

Adrian Bruce said: I understand that FS want to maximise participation - however, surely FS can direct persistent offenders towards some sort of training courses about the need to think before merging, add sources to justify your values, etc, etc,?

After all, they wouldn't let me, as a non-Church member, in to carry out ordinances but appear to place no requirements on anyone (member or otherwise) entering the data that goes towards those ordinances... Or even triggering repeated ordinances by entering repeated dupes...

Rant over...

LegacyUser

crhansen said: I agree! Sometimes we need to be bulldozers. -crh

LegacyUser

ClareLM said: Hello Everyone,

I have been doing some more thinking about how this problem might be solved. The limit could be based on the relationship that the record has to the user.

FamilySearch Tree has the functionality to determine the relationship between a user and a record ie using "View my relationship". The kind of merge that I've been referring to above is being performed by someone who is in NO way related to the person in the record.

So the system could first ascertain if the user intending to perform the merge is related, and then only if not related would the limit on the number of merges be applied.

I think such a condition would allow users in all those situations mentioned above, eg by Christina, Juli and Jordi, to proceed to merge more than the limit as the user is related to the record. If at FamilySearch Centres volunteers also which to do this work, and they are not related to the record, then perhaps a supervisor could have a password that allows the merge to proceed.

Regards
Clare