Direct link to source linker
On many pages on the FamilySearch site, you can directly access the source linker without viewing the person page (on the home page, Global hinting initiative see https://www.familysearch.org/en/campaign/hints-global?location=1927071). The problem with that is you take three things for granted when you just use the source linker and nothing else:
- The record is correctly indexed
- The person page on the tree is correct
- The record itself is correct
The explanation on the page of the global record hints initiative is way too simple.
I came across a death record for my ancestor where the informant was his son but he mentioned his uncle and aunt as the parents of his father. I needed to research multiple sources to come to a conclusion it was incorrect, only the date of death was correct.
When reviewing hints from my home page, I always go first to the person page. The linking of the record via the source linker is really the last step in the research process, I honestly don't review the data in the source linker as I reached by that point already my conclusion . See also my reply in the following discussion where the BYU Record Linking Lab is also mentioned (they only work with the source linker with all the consequences):
So I would strongly advice that FS forces the user to go to the person page first before being able to attach a hint and to remove the possibility to go directly to the source linker. If we all care about the quality of the tree, we need to be cautious when attaching hints. Do not understand me wrong: I love hints to be able to grow the tree but you need to do research before attaching it.
Answers
-
@Loïc Baert thank you for raising this.
We raised a similar point in 2024 in relation to some of BYU RLL's (non-automation-related) activities.
'Those using the [BYU RLL] Power Linker, Source Linker via BYU RLL hint emails, etc., don’t necessarily understand the impact a resulting change would have on the FT before they make or approve it ... Starting from a BYU RLL-chosen candidate match will always mean that the user has had no chance to judge whether this is the
best match availableto existing FT data.'It is clear that we are in very much the same position here with FS-generated Record Hints shown outside the specific profile.
Unfortunately we have made no more progress with BYU RLL on this matter than on their automation projects. Again quality seems to take a back seat to quantity.
Reference:
[Ignore the bit about duplicate algorithms at the end, I had failed to notice one of the available APIs]
1 -
I rarely use hints from my home page. In fact, I rarely visit my FS homepage. I usually start either from a person profile or from a record set. Context is essential, and hints from the homepage don't have that context.
4 -
I wholeheartedly agree, Loïc. There isn't enough context on the source linker (and especially the stripped down RLL "power linker," a definite misnomer), to make good decisions about people or data. A situation I've run into from time to time is blended families—i.e., a second marriage for one spouse or the other, or both, and they've brought children from previous marriages. But this often is not obvious in census records or other records. To confuse things even more, even if the children's surname is different, they're often listed by the step-parent's surname in just that one record.
As a result, the children get attached to the wrong couples. They've often duplicates but they aren't found because of incorrect relationships and even names.
I also see a lot of parents added via AI hints—parents with no vital information at all, nothing but a relationship to a child. In my experience, these turn out to be duplicates nearly 100% of the time.
It doesn't help to take shortcuts to get names quickly into the tree if the time saved is more than offset by having to do major cleanup afterward. As Albert Einstein is purported to have said, "Make everything as simple as possible, but not simpler." Building the tree isn't a race; it needs to be done with care and attention in order to create a tree that is at least reasonably accurate.
Kathryn
2
