View new record hints on Family Tree from 11:00-11:30am (MST) this morning
edited October 21, 2021 in Suggest an Idea
John Huff said: We will be putting new record hints on Family Tree this morning from 11:00 am to about 11:30 am as part of a test. During that time we will be focusing on identifying hints that are "Not A Match". If you can focus on marking "obviously incorrect hints" as Not A Match during this time period, it will help us improve this new set of hints. So any time you can take to look for hints that are Not A Match at 11:00 am today for about half an hour would be greatly appreciated.
Tom Huber said: I used the descendancy chart and saw a lot of hints and none were Not a Match. I was mostly working with Washington, Oregon, and Idaho records.0
Brian Jensen said: Thanks for those that helped with this test. We'll be reviewing the feedback to see if we need to make any adjustments.0
Shanna Jones said: Every hint I tried was "Not a Match" and I used McDonalds because they are so common in Ontario, Canada.0
Paul said: Believe I just found one attached to MGT5-K38 - Ohio death for individual of different surname who lived in England all her life. Marked "Not a match".0
Jeff Wiseman said: I cam across something that may have been related. Some of the sources that had been attached to a person showing their relationship to some of their children (marriage records again) disappeared from the person's source list, but did not show up as hints of them.
I knew that those sources existed somewhere, so I went to the child's PID and looked at their source list. I discovered the marriage source was actually there in their list. HOWEVER when I did a Review Attachments on that source, it was not currently attached to ANYONE including the child who's source list it was in!
So I had a source that even though it was NOT attached to the person, it was still in that person's source list!
So at this point I went ahead attached everyone in the source (using the source linker). Lo and behold, a Brand New citation for that source appeared in the source list with a new URL. It also appeared in the source lists of everyone else that it was attached to.
However, although the original citation (different URL) that I had done the review and attach on was still there, it was now showing that it was "properly" attached to all of the different PIDs it referenced. But this appears wrong as instances of that citation with that URL were NOT actually in the source lists for all the other PIDs!
If you look at the child Ada Harrison's source list, you can now see two distinct source attachments, each with a distinct URL:
If you do a Review Attachments on both of them, you can see that both of them are attached to her father as well:
But if you actually go to her father's source list, there is only ONE instance of that source attached there.
I discovered this on a couple of sources (all marriage sources). Really strange.
The reason that I think these may be related to the hinting issues are that at one time, they were in the source lists and had all of a sudden had the 410- Record Removed error so someone had gone through and detached them, but they showed up again in the hints list even though they still had the record removed errors. At that point I had Dismissed them so that they would not get lost until the removed record errors were corrected.0
Paul said: Sorry, Brian, it's not a case of WHETHER you need to make adjustments, but of WHAT adjustments. Having to unmerge IDs that have been combined on the basis of "FamilySearch" suggesting highly unlikely, if not impossible, "possible duplicates" is wearing me down and I believe I am far from alone here. Oh, and that's without my having to dismiss hundreds of suggestions as "Not a match" before some inexperienced user can do some serious damage.0
Paul said: Brian
Further to my response (above) to your, "...to see if we need to make any adjustments" comment, suggestions like the one below show there HAVE to be adjustments.
The only thing these two IDs have in common is a first name "Elizabeth" and a common christening date in England. I'm surprised there were not (at least) hundreds of "Elizabeths" christened in England on this date, so am immensely pleased the FamilySearch routine has only found this one, and I did not have to dismiss them all as "Not a match"!
A common surname and/or at least one other matching factor must surely be part of the criteria in a revised algorithm.
Must dismiss this "hint" immediately or, from past experience of many similar instances, I know some careless user will merge these two IDs.