Introducing the Geni Consistency Checker

Hello, Geni,
Am I going to get any feedback to my comment on 22nd November? The number of false inconsistencies of the type mentioned has now moved from 19 to 59. I expect more to follow unless this issue is rectified.

Hi Gervais de Matas, i have the same thing happening but if you hit ignore and choose mark as correct for all users, it turns green and then you know you do not have to deal with it. It no longer shows up on the profile when you do that. Same for other issues where there is no way for a user to change anything to make it go away, like parent too old, lots of 60ish guys have had children at that age. But you are right it could be dealt with in the coding, ie disappear completely if marked correct, since the box it is in will still remain, but you will see the green when you open it. Overall i have found the consistency checks helpful for making corrections.

Hello, Linda Snider. I agree with you that the checker has been very helpful in drawing attention to situations of obvious profile errors. My issue is that it has also created a "monster" by pointing out errors where there are none. Telling me that the "de" in my surname is a possible suffix in the wrong place, when it isn't, is giving me unnecessary work to do to officially ignore them, but to still have them accumulate, hidden away in the box, is not my idea of rectifying the issue. There are hundreds of people with my surname in my tree. I've ignored 59 of them so far. Am I to continue ignoring them as they turn up until they are all in the box? Surely, having taken the time to create this checker, they can take a little more time to fix issues like my own.

Gervais de Matas,

The checks that are depended on language still need work, there is the problem most users use the default tab for names in other languages than English (US).
This is causing problems with those checks because the checker assumes those names are in English.

Instead of ignoring profiles for those checks you can at the moment better try to handle those checks that are not language dependent first and use the arrow on the right of the title for these language depended inconsistencies to hide them for now.

Please also note that the default group for inconsistencies is relatives, but that there is also a group for the profiles you manage. This group can show inconsistencies for other profiles than that of the relatives group.

Home of the World Family Tree

Daily Digest for December 8

View all the discussions you follow
View recent public discussions

Maria Teresa de Marco (Cervi) - Please check the following, English (default): de

You replied to this discussion

Luigi Gino Marchelletta

The new software is creating so much more work and what's more we have to go back to correct these so called suffix's such as de , di, which precede the surname of the person. In Italy these are actually on the birth certificates and they are actually part of the individual's names. What Geni would like the name to be is; eg,, the surname de Rubeis, (DeRubeis) the surname di Bona (DiBona) etc,,, these Americanized surnames for people born in Italy are not correct...

Another country heard from! The tunnel-vision Americanism of the Geni Consistency Checker is biting *everybody* else in the butt.

The results of this checker are disastrous to the big family tree. Profiles have stayed untouched for 5 to 10 years and now every new person joining geni.com has this urge to change them, as system says there's something wrong (although everything is OK).

My suggestion is that let the checkers regarding age and dates stay, as they might draw your attention to some possible errors, but remove all those which have anything to do with names. The only impact of those name checkers is to destroy the family tree and ruin the 10 year work of people who have contributed to the big family tree.
Personally, I can’t understand what is the point of those name checkers. Has anyone found any help from them?

* 'birth surname looks like married name' – So, it was quite common you married someone with the similar last name. Generally you married with someone from your own village, so you did not have that much families with different family names in the area.

* 'children with different last names' – was pretty common in the past, that children took different family names. Not to the mention that when you moved to another village, you might still have the same name, but it was written differently (which for name checker would be an error). Say my own name Vuks: in different places it could have been Vuks, Wuks, Wuchs, Fuks, Fuchs.
Or say Livamägi, Liivamägi, Livamäggi, Liivamäggi, Livvamägi, Livvamäggi, Livvamägi, Livvamäggi, Liwamägi, Liwamäggi, Liiwamägi, Liiwamäggi, Liwwamägi, Liwwamäggi (or that same name in this particular family was in some records written German sounding Sandberg, direct translation, what sometimes happened). Still the same name all the time, but error for name checker.

And names didn’t change just when moving, but also through time, when you were staying in the same place. When you were born, your name was written in one way, 20 years later, when you married, it was written differently. Again error for name checker.

* 'missing birth surname' – I would guess, that in lot of countries around half of the tree falls into time, when the family names just did not exist.

* 'siblings with the same first name' – I would say that if you go back hundred something years and you find a family where there were no siblings with the same first name, then quite likely some children are missing from that family. With mortality rates high in the past, when child died, the next child born into that family usually got the same name. It was typical families had 2 sometimes 3 children with the same name. And often the siblings with the same name were all alive at the same time.

There have been sayings in this discussion earlier, that you can ignore this. Come on, think of the number of participants in this discussion versus the total number of Geni users (who are not going to ignore that checker).

Randel Vuks I agree, too many not errors notified as possible error.

A few days ago a new star started to set the husbands name on a lot of women pre 1700c, likely cause, this tool checking, please, rebuild the tool, it would be enough that it checks for date errors such as children born before parents or too young parents etc.

Randel Vuks brings up some important things we have said but with a new caveat: All geni users (I would venture to say most) are not part of the discussion boards and are not getting the message maybe to "ignore" things. Frankly, the inconsistency marker is annoying when you are looking through your tree going back in time particularly. The probability of tree tampering is greater to make something fit that just isn't right.

It has been helpful in rectifying a few things for me, but really, I have some fears of an "unknowing" user coming through whole lines and undoing some important work. Perhaps dropping the standard naming conventions that don't make sense for many countries would be the simplest fix for some of the potential problems. I like the fact that my family names have changed over time as it is part of my family's story and really hate to see it possibly wiped out. With each change came a new era, place, and/or generation of individuals.

I noticed that new feature too, Private User. That definitely needs to go back to what it was before where we manually entered names. I believe that women in my Finnish side did not take their husband's names formally until recent times. I believe Private User made a great post about this some time back.

Yep, as many other countries, women kept their fathers name. They did not take husbands name until 1900's. also in Sweden as in Finland nobles always kept their own name, also Scotland, Wales, Ireland, most of medieval Europe but not Britain. I always check the records and take the wrong name out when see one. It is important that records match the right person. also people outside of Nordic countries sometimes use patronyme of the man to the woman also. Everybody has their own patronyme. If the wife name is unknown, it is to be marked as N.N - not "mrs Nilsson" .

I tend to use "Wife of (full name of husband)", all on the First Name line, because if you pile up too many N.N.s (or NNs or Nns or other "who dat?" significators, you start getting bad matches.

Re: I tend to use "Wife of (full name of husband)"

I do too in congested areas in Colonial America. Last name was her last husband’s last known name.

I have done similar (eg, "Mother of (first name)" with mutual last name in Last Name field) -and the Consistency Checker includes under "Title in name" - should they just be marked as "Ignore" and "Mark this correct for all users" - or do you just ignore the consistency checker there - or ??

I lock them down and mark them ignore for all.

Mike Stangel,

Could you make the language depended checks (like title in name) for the default tab optional for single language profiles on a user basis?
That would help those users who use the default tab for other languages than English on single language profiles.
A bit more complicated option would be to ask the user for the language to use for those checks on the default tab for single language profiles.

Could you exclude the alternatives of N.N. (see messages above) from the title in name check?

Took another, more closer look at this checker. Discovered something really bad. I thought the checker's idea was to draw your attention to some inconsistencies, so you can take another look at the old records and re-check. But appears, it has this button 'one click fix', which allows users to change (read: ruin) the profiles in one click.
It is not normal to change any profile without doing any research and studying the old church records first. Quite often you spend hours and hours to find out information about someone. And now all your work is destroyed in one click.
I have informed people all the time, that in case, if they change something, they should put previous names on the nickname field (so other people and/or original managers of that profile will still be able to find that profile). With this 'one click fix' the original (correct) name will be lost and you will even not be able to find that profile, as it has some different name and you don't find it with search engine with previous name. Terrible situation.

It's completly useless as long as Geni still allows people to upload hundreds of profiles with a single click, yesterday, one new user uploaded 250 profiles in a row back to 1500c, they had to be manually merged, all the resolved fields had new inconsistencies in a blink of en eye!

Thank you Geni for allowing this to happen, of course, the new users have almost never engage themselves to resolve any problems in the wake of their own actions, just leaving the mess to be solved by others thus creating some kind of eternity work.

We will now never be completly done with any line, thank you again!

The problem Private User is pointing out has nothing to with the consistency checker. It even help to track the problems and solve them.

I like to point out that Ulf thanks Geni staff and I think that it is very kind of him. And very positive of him.

I share and again with Private User the work on Geni is never done. And that is also a positive point of Geni.

Private User,

A GEDCOM import will import 5 generations max on first import and it will halt the import on any branch where duplicates are found and offer the user to abort the import on that branch.
BUT if someone does a merge the import will continue.

So it best to give the user some time to decide before doing any merges.
For a new user solving conflicts is not easy so they may need weeks or months to do it.

So it may be wise to contact the user and ask if duplicates should be merged, before doing any. That way a user can look at that branch and decide, before others do the merge that may not be wanted.

Problem is not the existence of similar profiles (due to Gedcom import or otherwise) or the existence of inconsistencies.
Personally I find the existence of inconsistencies very useful in research. Say for instance person's birth year. Quite often, it could happen, that every record for some person gives different birth year (as quite often they were approximate). So, having multiple birth years in his/her profile, enables to check those specific years in the search for birth record. If the multiple birth years are removed from that profile, it would be much more difficult to search for the correct year.
So, I repeat what I said earlier, biggest problem now is that 'one click fix', which not only enables, but even urges to remove alternative information/data about that person and leave there just one version of information, which might be completely inaccurate, as this is not based on any old documentation.
According to my own experience, most of the suggestions, which I have seen, are completely inaccurate and have absolutely nothing to do with the person whose profile they are meant to change.

Have you considered creating an interface so that the user can select which of your consistency checks he/she wants to see?

Another idea could be to add an API to allow external partners to register alternative consistency checks that a user can select to run on his/her own part of the tree without running the whole tree analysis on the external site?
Such an API might consist of a callback delivering profile / family data, and allow the external tool to return a list of possible inconsistencies, that could be shown on the geni inconsistency page. (I know the current Geni.com API can be used in that way and use it for that already, but that puts the result data on my site.)

Previous system, when someone wanting to change some of your profiles, just sent you their suggestions worked much better.
The main problem with current 'consistency checker' is that I would estimate 95+% of geni users won't check the historical documents and rely on geni suggestions only, which regarding the names, is a complete disaster.
Historically people had different last names, inside one family (that occured for multiple different reasons). Names changed in time and when moving. The current system suggests names, what it founds among surrounding profiles. And most of the times, these suggestions have actually nothing to do with that particular person.
Earlier, if you received such false suggestions, you could just reject or ignore them. Current system enables everyone to change the profiles with 'just one click'. You even don't have to type that new name anymore. And previous name(s) of that profile (usually the correct ones) get lost.

consistency checker' - Shouldn't be used to change surnames - Back in History, a few families at times in the same generation used name variations- many kept the variation, a few went back to the original spelling, but we can't just automatically change all names and I don't change them- would have to personally delve into detailed lineage before to justify the change of info. I override the system to ignore the surname - and then refresh - then all is ok --- Do you think that many people pay attention to the Checker and are using it? , most people never read the profiles before making mistakes merging.

But it only offers the one click thing if there is nothing already in the birth surname field, ie it adds information, doesn't change it, afaik? To me this is a help since it then brings these people into the search for that name and possibly allows finding duplicates or parents one might not otherwise have been able to search for, even if they didn't actually use the surname

I have mostly seen suggestions, which have nothing to do with the particular family, e.g. the surnames of surrounding female profiles, who were in-laws, coming to that family by marriage, not directly by blood line.
It is different in countries, here in Estonia, the family names were first introduced in 1820s, in some places later. Before that, people just did not have any family names. If children were living in separate households, they might have taken different family names.

The main problem with that 'one click' thing, is that enables (not to say encourages) to change profiles without doing any research in old church books.
Personally I think, that changing profiles and/or adding any information without research, but just by feeling, that it might be suitable, is most irresponsible thing to do and should be avoided (at any cost).

Another problem is that when different branches of some family had different names (even if it is just one letter difference), it is shown as inconsistency, meaning there are plenty of people, who try to fix it, so they change the name, system says: it is OK now, no mistakes, but the name is wrong and you (and all the other people who might want to research that particular name) will not be able to find it anymore as you/they search for a different version of that name, which was used in that particular branch.
Normally, when someones has to change a name of some profile, it is most recommendable, that you add the previous version(s) of that name into the 'nicknames' field. With current 'one click fix', no-one bothers to do that.

I will take a random example of this 'consistency checker':
https://www.geni.com/inconsistencies/for_profile/6000000044729463449
The person's name was Ilmar Pavel. And it's all correct. However, the 'consistency checker' is not happy with that, because it's different from his parents' last name and he is missing birth name (maiden name for a boy??).
System offers 3 possible solutions, which should fix this "mistake":
a) Paawel - father's original family name. But wrong, as somewhere in the beginning of XX century (around 1920 or so) the letter 'W' was replaced with 'V';
b) Ojaste - father's new family name. Wrong, because the name change from Paawel to Ojaste, happened only in the end of 1930s, when Ilmar Pavel was already dead; and
c) Soe - mother's maiden name. Wrong, as with marriage mother took her husband's family name.
So, the checker considers correct name to be wrong and offers 3 different versions to fix the mistake and all 3 are incorrect.
Which generally mean, that people who want to get rid of the "mistakes", might change that name according to the suggestions of the 'checker'. And with just 'one click fix' the correct name will be replaced with wrong one (because system suggested that!)
For father, actually all the versions of the the original family name Pawel, Paawel, Pavel, Paavel are correct, for son, only the ones with average 'V' (and not with 'W' anymore.

But if you hit ignore and mark correct for all users then it avoids the issue, people will not get the one click button. So it is all a matter of dealing with the profiles if you are worried about people adding wrong info. I myself only use the one click when the suggested name matches the one they have already, otherwise i look into it further or hit ignore. I rather liked it when Geni used to automatically fill in the birth name field for all genders based on the parent name which automatically came up, now it just moves female last names entered to that spot and opens up the last name field as blank, as though she doesn't have one until she is married...i would like to see it go back to the gender neutral way it was before, it worked fine, i thought.

Found another issue regarding this checker. It was among my profiles, but it's more general. We have a lot of nations (or languages), where family names are gender based. Russian for example, but of course a lot of other nations/languages as well. So, all the women in the family have different version of family name, compared to men. Russian example would be say: men are "Ivanov", whereas women are "Ivanova". And currently it is a mistake for the checker.
This consistency checker suggests that I should use men's version of family name for women, which is completely incorrect, but according to 'checker' it would fix the problem! What???
(not to mention, that this girl has currently lady's version of that family name, but this correct name is considered to be a mistake by geni, because it's different from his father's family name.)

It just needs to be coded correctly but my guess is it is backburner tasks so may not get done for awhile. Hit ignore, mark as correct for all users and that should keep too many changes from being applied willy nilly. But again it can give ideas about adding different spellings to the alternate names field so that the males and females both show up in searches, for instance.

I have around 20 000 profiles with ca 1500 inconsistencies, so I think I am physically not capable of marking them correct for all users.

And this is not the point.
Big family tree is a result of the contribution of the huge group of people, meaning a lot of profiles are added by someone else. Now it has been done extremely easy for everyone to change these profiles.
The idea of the 'checker' is to enable changing profiles without doing any research, just according to the suggestions of the system. The big problem is that the system takes the profile, where everything is correct, marks it to have errors and then suggests it's solutions, which according to my own experience are nearly always incorrect.