Reoccurring Person Data Problems warning

LegacyUser

Paul said: I have been confronted by a problem with which I am not familiar. The screenshot below (ref. https://www.familysearch.org/tree/ped...) illustrates.

I am probably making a fool of myself here, as the solution is probably quite straightforward. Nevertheless, I would be grateful for any advice.

As you can see, the branch includes the marriage of cousins - Stephen Wrightson and Mary Cayley - the children of sisters Mary Ann & Jane Sedgewick (daughters of Isaac), respectively.

(BTW - I am in the process of sorting out the general mess attached to this branch, so please don't look too closely at any other work that is required!)

LegacyUser

Paul said: Though I'd leave the post here, but have just found I can dismiss the suggestion from the Landscape view. Presumably it is just a warning something MIGHT be amiss.

LegacyUser

Juli said: There was another thread about this on GetSat recently, but of course I can't find it.

I don't use the views where this warning might appear, but it sounds mighty annoying: a data problem that doesn't show on the profile anywhere, and cannot be properly dealt with when it does show. I think that either the "feature" hasn't been fully implemented yet, or someone didn't think things through.

LegacyUser

Paul said: Yes, even if I look at the branch using Isaac as the primary person no data warning appears. (see https://www.familysearch.org/tree/ped...) I guess it's just related to the apparent duplication of individuals (Isaac & Jane) in the other page view.

LegacyUser

Adrian Bruce said: Yes, there was definitely another thread on this topic recently! (grin)

I confess I'm kinda puzzled why someone has programmed this. After all, there must be lots of cousin marriages - naturally, when I looked at the Cooper branch of my family where this occurs, I haven't yet got back to putting in the common ancestor so I can't see what it looks like.

My idea of "Recurring" ("Reoccurring"?) is when there's a loop and a grandparent and grandchild (say) have the same PID - that was my immediate thought of what "Recurring" would be. I've no idea whether the programmed check will pick that up - I've not got such a scenario to test and I'm certainly not corrupting the data to test it.

LegacyUser

Juli said: Not the thread I'm thinking of, but same topic: https://getsatisfaction.com/familysea...

The general guess is that it's meant to point out "looping" pedigrees ("I'm my own grandpa"), where a relationship got entered in the wrong direction or some such, but mostly it just shows an error where there isn't one.

LegacyUser

JT said: Apparently you'll get this warning anytime a husband and spouse are first cousins.

Notice the marriage on the far left (root): Stephen Wrightson's mother Mary Ann Sedgewick has the same parents as Stephen's wife Mary Cayley's mother Jane. Jane Sedgewick and Mary Ann Sedgewick are sisters!

Not likely at all, so good catch FamilySearch Family Tree programmers!

LegacyUser

JT said: And the reason this only shows up on the pedigree, is because that is the only place such an unlikelihood would be seen.

LegacyUser

Adrian Bruce said: Sorry - but why is this unlikely? I have done work (elsewhere) with members of the Church of Ireland (aka the Protestant Ascendancy) - having a small number of people in one area, cousin or second cousin marriages were par for the course.

LegacyUser

David Newton said: If that's what triggers it then the "warning" is worthless. First cousins marrying each other is perfectly legal, albeit unwise. I have an example in my tree in my great grandfather's generation. It's a valid family tree structure and so shouldn't trigger any "warnings".

LegacyUser

JT said: Only likely in a few areas/times. Warnings are just something to be aware of. If OK, ignore it.

LegacyUser

Adrian Bruce said: Except that the inclination of many, I suspect, is that "The Computer Must be Right - I need to eliminate the flag (not ignore it) - but how????"

And when you go to Isaac's profile, there's no sign of the warning. That's really frustrating and puzzling for anyone. At the very least, the "Reoccurring" flag should appear on the profile with an explanation (as per other suggestions) that this is perfectly fine unless there is a fetish against first cousin marriage in that culture. Both those aspects (appearance on the profile and explanation that it might be fine) are needed.

LegacyUser

David Newton said: "Warnings are just something to be aware of. If OK, ignore it."

Erm no. The terminology is wrong anyway. This is not actually a "warning", it is a "data problem". A data problem is not an advisory, it is an indication of an actual problem that needs sorting out. First cousins marrying each other is NOT, I say again NOT a data problem. It is a perfectly legal, albeit unwise from an inbreeding point of view relationship.

This data problem flag is not only worthless, but actually counterproductive. It could prompt people to remove valid relationships from FSFT. It's even worse than the useless possible missing chiild suggestions as those are only worthless suggestions. This is a worthless and false flag of a fictitious data problem in the system.

This is not a looping pedigree. It is a valid familial relationship.

LegacyUser

Juli said: This flag needs to be rethought.

First of all, a flag that appears in some views but not others is a Bad Idea, no matter what: either there is a problem, or there isn't, it can't be dependent on the view! Ditto for the ability to deal with the flag: there must be a (clearly-marked, intuitively accessible) "dismiss" option on every single instance.

Second, people can recur in a pedigree for all sorts of perfectly valid reasons. Cousin marriages and double cousins come immediately to mind. If there is to be a check for "looping" pedigrees, it needs to check strictly within a direct line, i.e. for cases where a person is entered as his/her own ancestor or descendant. But I suspect that such a flag would be less than helpful: sorting out multiple generations with identical names is hard enough without red exclamation marks bugging at you while you're trying to get the correct wives attached to the correct husbands (or more to the point, detached from the incorrect ones).

LegacyUser

JT said: Thanks everyone for countering my pitch. I think I was just pitching what FamilySearch PM's would've expected to be seen pitched. Now they should have good reasons to re-think their angle.

LegacyUser

joe martel said: In the past cyclic pedigrees were a common mistake created by users, usually due to different person with the same name. It actually causes bugs in software that isn't anticipating the cycle (recurring person). So this was programmed in to let the user know of the potential problem.

It should be a warning that can be dismissed, because there are valid reasons for cycles.  If you can't dismiss please post the PID and the descendancy view that it occurs in.

Regarding the UI, that is a bit tricky. To say something is wrong without the user seeing it is a problem in itself. So if the user can't see the cycle then they can't really assess what the supposed error is. Also, the error may only show up because the code that is rendering the view sees that repeating PID and then tells the user what it sees. You can imagine with 1.2 billion person and probably 3+ billion relationships it may be difficult to render algorithmically every possible pedigree. And if it could it would have to show the exact view that the cycle shows up in. So that's why a specific view is the best way to communicate the warning to the user.

LegacyUser

Adrian Bruce said: Thanks Joe. If the warning is generated by the software constructing that view of that part of the tree, rather than a free-standing algorithm, then I can begin to understand why it only appears in certain views, thanks.

But what I still don't get is what the check is looking for and why. You talk about cyclic pedigrees - sounds a good name and I assume that what you mean by that is someone who is their own grandparent or similar - what I've referred to as a loop. But what we're seeing here is not detection of cyclic pedigrees but detection of the perfectly acceptable (in many parts of the world) practice of marriage between 2 cousins. I can see that it catches marriage of first cousins, so I imagine that it will also catch marriage of 2nd cousins? And I'm not sure how big the pedigree view - and the corresponding check - can get.

Checks for cyclic / looping pedigrees - really good idea. (That was how I first got into updating FSFT - fixing a loop in my Bate line.)

Checks for cousin marriages - not a good idea for many of us, especially if it catches 2nd & 3rd cousin marriages, etc. And I'll bet we'll see way more false positives from cousin marriages than genuine problems from cyclic / looping pedigrees. Multiple false positives are, I suggest problematic because the system's crying wolf all the time.

So can you work out Joe, just what this check is doing?

Is it simply looking for someone occurring twice in a view? That would indeed catch cyclic / looping pedigrees - but also cousin marriages. But, of course, what I don't have any appreciation of is how easy it would be to just look for cyclic / looping pedigrees while ignoring cousin marriages.

LegacyUser

JT said: You're right Joe - these can be dismissed at https://www.familysearch.org/tree/ped... (familysearch.org/tree/pedigree/landscape/GQ98-H4F) but I won't do it for the sake of demonstration. The reason for dismissal could be: "The marriage of Stephen Wrightson (1886-1948) & Mary Cayley (1880-deceased) in 1911 as being 1st cousins to each other, is valid. This is not the beginning of a loop."