db_error comment

+17 votes
368 views

Hi! I was looking at this page and saw the following:

  • 400 ... to find all partners with same LNAB. Generally they shouldn't be the same except in rare cases like when siblings marry.

Just want to suggest that this is not as rare as implied. In Colonial America, extended families traveled together to new locations, and you'll find marriages between folks with same LNAB fairly often - not between siblings, but maybe cousins & in a Peyton/Payton marriage I just worked on, between two different family lines.

Thanks!

asked in WikiTree Tech by Liz Shifflett G2G6 Pilot (308k points)
retagged by Ellen Smith
even now too... my daughter is dating someone with the same last name, and as far as I can tell, we're not related.

Actually if your name is Smith I guess the most common name for a marriage partner is Smith.

But I think the number of genuine cases will be small compared to the number like Mrs Morgan Howell

However, false errors do produce a high annoyance level.

 

Rj, It is not as uncommon as you would have everyone believe. I am finding that happen with names like Barringer, Blackwelder, Earnhardt, and many more.
I have one branch that hails from a small German town with limited # of unique surnames. Jauchs were regularly marrying other Jauchs, Benzings other Benzings and Schlenkers--the Smiths of the town-- were regularly marrying other Schlenkers. etc etc.

In fact, I chose not to upload my GEDCOM for this town because the matching would be a friggin' bear!
I have a Marshall/Marshall marriage I've been working on in England in the mid 1800's.  There is nothing to suggest that the two Marshall families are related in anyway.  I just recently noticed a Marshall/Marshall marriage announcement in our local newspaper too.
The occurrence of its reality needs to compared to the occurrence of data entry based errors. Basic error theory and statistics: if your guess is as big as your margin of error then you have nothing substantial.

A good guess is strictly smaller than its margin of error.

The question is how common do you think same LNAB marriages are compared to the total number of marriages?
If it was a 50% (different LNAB) vs 50% scenario (same LNAB), I would think it isn't worth flagging profiles because it is as likely to be one as it is to be the other.
If it is a 51% vs 49%, I would say it is marginal and not worth flagging for the general public though it would be worthwhile for statistical analysis or for machine correction.
If it is 67% vs 33% then it would be clearly worthwhile to flag and review the profiles. Colloquially, it might be more proper to refer to it as uncommon.

Do you know how common it actually is? And I don't mean for your particular profiles; the plural of anecdote is not data.
Would you bet on it being close to 50%?
Do you know how common same LNAB between spouses is in genealogical databases?
Do you know how commonly people erroneously assume that a person's LNAB is their marriage name?

We will have some hard numbers to use to make these estimations by adding the error to the error checking project and flagging all the profiles with what amounts to a warning.

I mean do we happen to have a study that shows us just how common this is globally and historically in terms of quantities? How does that relate to the error rate in genealogical database entries?
The only way to find out is to have a look.  It doesn't seem like an expensive check to make.  A failed experiment can be made to disappear very quickly.

To avoid over-reaction, it might be better if convenient just to produce the static lists and not include the error in the personal reports.

Obviously, given a list, the trick would be to go through it and tick off the false errors asap before they annoy people.  Once ticked off they disappear for good.

The odds will be improved a lot by only counting wives with no father attached (though you do get a few where somebody has found some Plodson parents for Susan widow of Robert Plodson).
Ian - nope, I know none of the answers about percents/statistics/etc. you ask. I just wanted folks to be aware of the possibility that same LNAB can occur more frequently & in more circumstances that what the writeup for the error said (and still says): "they shouldn't be the same except in rare cases like when siblings marry"
Then the appropriate course of action for you is to submit an edit for the description detailing known or expected exceptions. We'd all appreciate that.
I'll give it a go, although I don't know anything aobut he workings or tech for db_errors.

Please edit

    400 ... to find all partners with same LNAB. Generally they shouldn't be the same except in rare cases like when siblings marry.

to read


    400 ... to find all partners with same ^Given and^ LNAB. Generally they shouldn't be the same  [delete - except in rare cases like when siblings marry end delete] ^ and will mark as error those profiles who are styled as "Mrs. [Husband's Name]".
Finding the Mrs John Smiths would be a first step.

But I'd also want to go on to find all the Mary Unknowns who marry John Smith and arrive in the gedcom as Mrs Mary Smith.

The problem is the ambiguity.  If John Smith did marry Mary Smith, the profile may be technically correct, but you need a [sic] note on it to say so, because it looks like somebody's just filled in the married name as they so often do.

Of course we also have a lot where John Smith married Mary Brown, who was really Mary Unknown, widow of Brown, and a lot more where we don't know whether Mary Brown was a widow or not.  But they're harder to find.
Todo list on project page is just general guigeline for me to have a list of what could be done. When i will start to work on item, i will check and decide on details about implementation.

Of course this check will be performed only on wifes without parents.

I dont think prefix Mrs tells us anything, but i will check that.

Then we will estimate the false errors ratio and decide to keep an error or not
Mrs is common in Ancestral File etc, but usually dropped on importing into WikiTree.  This is why the Morgan Howells look bizarre until you realize.

Sometimes I've been proposing a merge with the husband before I realised she was supposed to be the wife.
'Mr' and 'Mrs' are actually titles for landed gentry in 16th, 17th, 18th century England.  They should not be removed. (It also helps me, among 30 John Blagdons, to see which one is the landed gentry.)

Thanks so much Aleš Trtnik - I am in awe of the database you created!

2 Answers

–5 votes
Liz, I agree. I am working on some lines from North Carolina in the 1800's and there would be a large number of profiles added to this "error" from these lines that are not errors. There are far too many errors that this project reports that are not really errors and just adding sources can prove this, but they just want to fix these false problems to fit their view as to what should be. I fear that this project will evolve into something that will make WikiTree into a very poor source for information in the near future.
answered by Dale Byers G2G Astronaut (1.2m points)
I downvoted this answer because you're lumping all db_errors into the same bucket. Some of them are obvious fixes and easy and they don't generate much if any controversy. A few of them are showing up as problematic. It's our job as a community to figure out how to make this work. We should not throw out the entire project just because we've encountered some challenges.

I also downvoted this answer because I'm tired of seeing comments along the lines of "they just want to fix these false problems to fit their view as to what should be". It comes off as conspiracy-based, like someone is out there trying to get us, and they spend all their time thinking of ways to piss off the entire wikitree community. Such wording reflects thinking the worst of "they" (whoever the heck that is) vs. assuming best intentions.

I choose to believe that the db_errors effort was created by very well-intentioned people to address the amazing amount of incomplete or bad data that-- if we were to do all of this manually-- would never get found much less dealt with. And that the people who are involved in checking the errors are also well-intentioned. Is the process buggy? Obviously. Is it resolvable? Absolutely.
Jilliane, First I do think that some of the errors reported are valid and should be fixed but I have seen far more that are not errors on even sourced profiles and because the error report can not see the sources it still thinks it is an error. Now the project has on it's to do list adding something that can increase the number of false errors reported. The real problem is not the report but rather some on the project think that by making fast fixes without looking for sources or even thinking about how the changes will affect others they are doing something good. Just today the majority of the "errors" that I dealt with on the list based on my watchlist were actually False Errors based on names. how is this helping  WIkiTree by making active managers check and mark all of these false errors? I have no problem fixing actual mistakes but this project is making more work that does not need to be done.

I agree that this particular error would produce a lot of false errors but it will only be changed if a human makes the change. I have some faith that people won't make these changes willy nilly.

For that reason,  I don't agree that it will make wiki tree a poor source of information. The project has the potential to improve so many of the profiles that are at the moment a poor source of information .

 Some of these changes are simply common sense and don't take that  much time, others require quite a bit of research and I'm certain that I'm not alone in taking time to do it,

eg http://www.wikitree.com/wiki/Burt-1144 There are lots of date errors here (birth before and after fathers death)  You can see by my notes that I've had to find baptisms and find and read the father's will.  I'm not there yet as I can't simply change the dates to the correct one because the  George linked to Thomas here has another set of children and lives in the next county as well as the previous century. He  must fit into a family somewhere .

Dale,  I wouldn't down vote you, but I do think you're way off base.  I think, whether you're aware of it or not, is that you're being affected by the word "error".  The negative vibes of Error make us think we're being accused of something horrible.  Perhaps we could rename it the "please check and see if this is correct and fix it if it isn't" project, but I don't think it matters.  An automated system is almost by definition prone to false positives and this system has a simple way to weed them out by interested humans doing the checking.  I've clicked several unique names as false errors without even having to go look as the're names I'm sure of.  If I was working on someone elses profile, it's only simple politeness to either source any change and/or notify the PM before or after the change.  I don't think there's going to be any big problem from false errors, especially after the numbr of errors are wittled down a bit.
The problem Helen is I have seen just today someone make an unneeded change that affected others in a way they did not understand, and it was a date error. The dates worked but because the error report software did not understand that 1941 would fall into the after 1909 range it was reported and some human then changed one of the dates . I will not go into the problems that will cause but instead of just marking it a false error someone felt they had to make a bigger change. Like I said it is causing more work for me than I care to do so I am reducing my watchlist size and concentrating my sourcing on other sites.
Dave, see my reply to Helen, it is happening and I witnessed one bad change that will be difficult to undo just today. I an done with this thread as well as any other about this project.
The only error, that has many false errors is 511, Up to now there were 5000 false error reports, and 4087 are for error 511. That is a lot, and I also reduced number of errors reported from initial 470K to 240K by changing algorithm. From reports of users, around 50% are actually misspelled, for some it cannot be determined and others are false errors. Anyone is welcomed, to count actual errors 511 and false one's and report back to me. I might remove it, but for now I think it corrects enough spelling errors. And don't make up the numbers, I will recheck them.

Also nobody is correcting the list systematically, since it is too big, and you can not know what is correct. So this error should be corrected only for one's close family, where you have personal knowledge of how name should be spelled.
I have an easy approach to the db_errors project: DON'T TOUCH.  If it is a small and simple thing, like gender can be, then yes, change it, and always always always leave a note in the box to say what you did.  But if it is anything else, such as dates, locations etc, then don't fix it yourself.  Read all over the profile, in case it is already being worked on or there is a previous message.  Drop a note in the comments box (such as 'Fred is unlikely to be the son of Joe, as it seems Joe died twenty years earlier.  What do you think?'), hide the 'error' temporarily for 30 days, and Step Away From The Keyboard. LOL

It's been working for me so far, and all I have received are thanks from profile managers.
upvote for Ros

Well I didn't down vote, because I think it isn't really clear to all members that from all reported ''errors'' many in fact are ''possible errors'', it are not all clear and certain errors that need to or can be ''fixed'' immediately, but many are possible errors that maybe are but maybe are no errors at all, that's why all (possible) errors need to be checked and looked at by real people. 

Also the project has many project and error pages where all is explained, so I suggest maybe people should read those pages as well, before they start to work with the error report 'tool' And there are different options for different situations : 

1. If the error is no error at all , so a false error you can report  it as false error and it will not be reported as error again (after the weekly update) 

2. Temporarely hide the error If you encounter an error, that you cannot fix and you posted a message to profile manager or you proposed a merge, you can click a link on the right to tell the system to ignore this error for a month. If profile manager will correct the error, it will no longer exist, otherwise error will reappear after 31 days so other actions can be taken. Error will be hidden at latest on next recalculation (on monday).

And if the error report is making people add sources to make clear if something is or isn't an error, than of course it only will help making WikiTree a better source for even more accurate information in the future ;) 

My favorites in this thread:

Jillaine:

It comes off as conspiracy-based, like someone is out there trying to get us, and they spend all their time thinking of ways to piss off the entire wikitree community. Such wording reflects thinking the worst of "they" (whoever the heck that is) vs. assuming best intentions. 

Aleš:

The only error, that has many false errors is 511

Dale:

I an done with this thread as well as any other about this project.

Well, I am done too with this thread and all others accusing people of wrecking WT, using only examples in their own small world where things went wrong, suggesting without quantiftying that happens on a massive scale, suggesting "they" are in a hurry to clean up, etcetera, etcetera...

Over a month now I am working down the list of errors in the Dutch province Noord-Holland. Starting with over 700 I am now down to a little under 500. So the context of all concerned profiles is 100% clear to me, I know official records of all people in the Netherlands are available online, I add these sources, make changes based on that sources only.

That's another story than the accusers like to present.

Okay, let's all take a step back and a deep breath and not post here til we can do so respectfully.
Where is the lack of respect in my comment?
That was intended in general, not necessarily you specifically.
0 votes
George Taylor, Signer of Declaration of Independence married Anne Taylor no known connection. This seems common to me  In fact in my own family there are 2 different Miller families that have 3 marriages between them. Henry Morgan the privateer Married his cousin also a Morgan. It is pretty easy to find these.

Art Morgan
answered by A M G2G3 (3.6k points)

Related questions

+4 votes
1 answer
69 views asked Aug 2, 2017 in Policy and Style by C. Mackinnon G2G6 Pilot (123k points)
+24 votes
1 answer
97 views asked May 17, 2016 in The Tree House by Anonymous Vickery G2G6 Pilot (236k points)
+5 votes
2 answers
+2 votes
1 answer
132 views asked Jan 21, 2018 in Genealogy Help by Juha Soini G2G6 Mach 6 (67.3k points)
+22 votes
9 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright

...