Impossible hint listed for three ancestors
Impossible hint listed for three ancestors
The hint is listed as a Research Help, and is the Death, in 1940, for Susie Vanderzee, born in the Netherlands in 1855, who is listed with a father, Age Westra.
This hint has been listed as a Research Help for at least six different persons whose name was Age Westra. These are all listed below.
For three of these persons, the listed Research Help is impossible. Suzie was born in 1855, and three of the persons listed below died before 1855. One of them was a baby who died at age 1, and two others died as young men, and were never married. So none of those three could possibly be the father of Suzie.
By placing impossible hints for ancestors, Family Search is making its data base vulnerable for careless people who blindly attach the hint as a source, and in doing so, use the Source Linker to add Suzie as a child to those persons, resulting in red flags, child born after father's death.
We have spent some time in cleaning up these problems.
Two of the other persons named Age Westra did not have a daughter in 1855. Only one of the six is a correct hint, and it has been attached to the correct father.
Age Klazes Westra Male 10 November 1821 - 19 April 1907 L4JQ-MQR Correct hint, now a source
Age Douwes Westra Male 19 October 1829 – Deceased KZP4-75L Incorrect hint was listed, now not listed
Age Tjallings Westra Male 25 July 1832 – Deceased K86B-WHK
Age Paulus Westra Male 5 July 1820 – 9 June 1841 KCS9-7D4 Impossible hint
Age Harmens Westra Male 17 August 1813 – 8 January 1815 GVHQ-**** Impossible hint
Age Riemers Westra Male 14 September 1824 – 2 June 1847 K8T8-SBS Impossible hint
The hinting algorithm should consider more things than just a match on the name. They should do at least a cursory sanity check to see it a hint is even possible.
When we now look at Age Klazes Westra Male 10 November 1821 - 19 April 1907 L4JQ-MQR, where the hint was correct, and is now a source, and click on his sources, and click on the source from 1940, we see the death record. Then when we look in the rightmost column under Similar Records, we can see that it lists all six of the above persons, and states that this source is attached to all of them. That is not true. The source is only attached to one of the six. Perhaps, because this hint is still listed for four of the other six, it thinks that the source is attached to all of them, but it is not.
In checking again, sometimes the number of Similar Records we can see is only five, and not six.
Hopefully this problem will be brought to the attention of the engineers who could consider cleaning up the code where it is warranted.
I realize I could just Dismiss the errant hints, but that would tend to sweep the problem under the rug where the real problem would then be hidden.
Answers
-
If I'm understanding your post correctly, then part of the problem is that you're misinterpreting what FS is telling you.
On the index detail page for the 1940 death record (https://www.familysearch.org/ark:/61903/1:1:Q2M6-F9VX), the list in the right-hand column is of similar records: indexed records that are not the one that's attached to Age Klazes, but are similar to it in some way. The list is currently showing seven different index entries, one of them not yet attached to any profile, the other six attached to six different profiles in the Family Tree.
Part of the confusion is likely to stem from the fact that the first one -- the one that's not yet attached anywhere -- appears to be another indexing of the same 1940 Chicago death registration. It's recording the same event, but it counts as a different source in FS's structures, because it's a different entry in their database. (From what I can tell, the 2018 [already-attached] index was of the certificates themselves, while the 2022 [not-yet-attached] one was an index of an index. I'm guessing based on what the catalog says about the film/image group numbers; being Chicago, the images are FHC/AL only for a public FS account like mine.)
Even when researching someone with an uncommon name, I find little use for the Similar Records list: I don't think it has ever found a record that I didn't find by other means. When researching someone with a common name, it's best to just ignore that list; it's highly unlikely to tell you anything useful.
2 -
Julia Szent-Györgyi ✭✭✭✭✭
Thank you for setting me straight on the Similar Records feature. The list of Similar Records appeared when we were looking at the record for Age Klazes Westra Male 10 November 1821 - 19 April 1907 L4JQ-MQR and looking at the source for the death in 1940, for Susie Vanderzee, born in the Netherlands in 1855, who is listed with a father, Age Westra.
The Similar Records listed six persons in Family Tree (FT), who were named Age Westra, and for each one, it stated Attached To: I assumed that it meant that this death source was attached to each one, but that was fuzzy thinking on my part. The Attached To : was stating that there was some other record attached to each Age Westra.
However, the main reason for posting this problem was to show that Impossible hints were listed for three ancestors, and the Similar Records did not substantially affect that problem, so I will rewrite my original post, emphasizing the impossible hints, and de-emphasizing the Similar Records.
Impossible hint listed for three ancestors
The hint is listed as a Research Help, and is the Death, in 1940, for Susie Vanderzee, born in the Netherlands in 1855, who is listed with a father, Age Westra.
This hint is attempting to tie a father named Age Westra to this Suzie. Suzie is already listed in FT with her correct father, but FT is apparently looking for other fathers named Age Westra. The source for the death of Suzie lists Similar Records for six persons named Age Westra. The hint is listed for three different persons whose name was Age Westra, who were in the list of 6 similar records, where the listed Research Help is impossible. Suzie was born in 1855, and those three persons died before 1855. One of them was a baby who died at age 1, and two others died as young men, and were never married. So none of those three could possibly be the father of Suzie.
By placing impossible hints for ancestors, Family Search is making its data base vulnerable for careless people who blindly attach the hint as a source, and in doing so, use the Source Linker to add Suzie as a child to those persons, resulting in red flags, child born after father's death.
We have spent some time in cleaning up these problems.
The three persons named Age Westra, who list that impossible hint are:
Age Paulus Westra Male 5 July 1820 - 9 June 1841 KCS9-7D4 Impossible hint
Age Harmens Westra Male 17 August 1813 - 8 January 1815 GVHQ-**** Impossible hint
Age Riemers Westra Male 14 September 1824 - 2 June 1847 K8T8-SBS Impossible hint
To see the impossible hint for each of the above three persons, go to the person page for each one, and look at their Research Help.
The hinting algorithm should consider more things than just a match on the name. They should do at least a cursory sanity check to see if a proposed hint is even possible. If a hint is trying to attach a father to someone, FT should determine if the person considered could even be a father.
Hopefully this problem will be brought to the attention of the engineers who could consider cleaning up the code where it is warranted.
I realize I could just Dismiss the errant hints, but that would tend to sweep the problem under the rug where the real problem would then be hidden.
0 -
The impossible hint for Age Harmens Westra Male 17 August 1813 – 8 January 1815 GVHQ-****, got messed up in the above list. This is the correct ID.
0 -
The system will not copy and past the ID for the above person correctly. I will type it in: GVHQ-****.
0 -
Again, it will not let me type that ID number. The last three letters are ****. You can see that ID correctly in the list of Similar Records.
0 -
Again it will not let me enter those three letters, F as in Frank, M as in Mary, L as in List. That is surely strange. Perhaps the subject of another post.
0 -
(I haven't a clue what awful thing this Community's nanny-bot thinks F M L means. The only thing I can think of is First, Middle, Last, as in names, and I cannot imagine any possible objection to that.)
Age Paulus and Age Riemers could theoretically have fathered children -- Age Riemers even has a "Research Suggestion" to that effect -- but Age Harmens could not, and it seems very strange to me that the hinting system is ignoring this fact. It also seems very strange that the system is not only suggesting a record that's already attached, but also doing so on three different profiles -- ones that the Possible Duplicates routine would never suggest, given the totally-mismatched names, dates, and relationships.
In a broad view, I can understand why the algorithm wouldn't discard a suggestion for a single mismatch or absence. For example, when the index includes only a father, not a mother (as here), I can see why it would ignore a man's marital status in its matches. However, there has got to be a way for the system to look at the whole picture: when a man not only doesn't have a wife, but also doesn't have a child (never mind one with the right name), and he died a decade before the supposed child was born, the algorithm should be discarding the hint without ever involving a human.
It is my impression that dismissing the hints is actually the best way to get the data to the engineers. It gives a "thank you for your feedback"-type message when you do so, anyway. While I doubt that programmers have time to read our reasons for dismissal, I'm sure there's some process in place for counting how many hints with particular types of mismatches were dismissed.
(Edited to add: yay, it worked! I made Age Harmens's link evade the nanny-bot's blind search-and-replace by using URL encoding on the last letter.)
0 -
I'm not suggesting that it's the case here, but the hints are only impossible IF the information on the profiles is accurate. We've all seen profiles with inaccurate or just plain wrong information.
I once recorded a man as deceased (by a certain date) because his wife was recorded on a census, living with a son, as a widow. When the hinting system suggested it might have found him on the same census I almost dismissed it immediately. Turns out he was living further along the same street, with another son, as a widower. I suspect a major falling out, but the impossible hint was correct.
1 -
Just a comment regarding the term widow. Colin didn't mention what census he was looking at, but a while back when I was going through newspapers from the early 1900s looking for family information, I ran across several instances where it was clear that "widowed" was being used as a synonym for "divorced."
Going to the question about impossible hints.
The Hint engine does, in fact, takes a pretty comprehensive look at all the information, family relationships, and prior conclusions on a person. This includes all the existing sources on a person.
I only looked at the first example: "Age Paulus Westra Male 5 July 1820 - 9 June 1841 KCS9-7D4 Impossible hint" and while poking around found this:
Since a user has declared, by attaching a source, that Age Klazes Westra and Age Paulus Westra are the same person, the Hints engine will include that piece of information in order to find more hints. I'll leave this for @jlwestra to test out, but I suspect that if you remove that incorrect source and refresh the page a couple of times, that the impossible hint will disappear. You should also double check that that 1820 source is also for the correct Age.
0 -
Actually I have a couple more minutes. Looking at the other two, I agree that there is no clear reason why those have the same hint on them. About 2 or 3% of the time, hints are just wrong. So that incorrect hint may not disappear after detaching the incorrect source after all.
Another reason for dismissing hints like this, is that the Hints engine also looks at dismissed hints when considering additional hints to determine who people are not and to not present a hint. This is why it is important to attach all duplicate hints and not dismiss as "not-a-match" a correct hint in the name of keeping the Source page from getting cluttered. When you dismiss a correct hint you are decreasing the chance of getting additional correct hints.
2 -
Gordon Collett ✭✭✭✭✭
Thanks for your input for "Age Paulus Westra Male 5 July 1820 - 9 June 1841 KCS9-7D4. I did delete the errant source. That didn't remove the Impossible hint. Then I saw that the errant source also put in an entry into the residence part of Other Information. I deleted that also. That still didn't remove the Impossible hint. So I finally dismissed the errant hint for the three persons named Age Westra. That did the job. I could see that the dismiss process does allow for adequate descriptions of the problem, and it triggers a Feedback.
0