Seen just now:
Got to the linker page from here:
The linker is confused, which is not normal. No action is needed from me other than to report that the linker is confused.
Linker confusion is likely to confuse contributors. This may be how cross-linked sources are created. I find those occasionally and when I find them I fix them. But better would be if the linker does not contribute to the creation of cross-linking.
This article will explain that the triangle is indicating that this source has been attached to another person. You will need to search that other person to determine if the two need to be merged or if the source has been attached to the wrong person. You can then detach it to attach to your orignal person.
Source Linker displays that "detach" triangle if the entry is attached to a different profile than the one it's currently lined up with. This means that it's likely all attached correctly, but for some reason, it's switching the couple around and therefore getting confused.
The triangles should go away if you change the focus person to James on the left or to Sarah on the right. I think. I have no idea why it's happening, but I hypothesize that something about the relaxing of the "man on top" rule is causing things to not quite work properly.
Not a bug. This is a feature to warn you that the source is linked to someone else in addition to the person shown here.
Going to the page, if you click on the warning triangle you will see that the source for James is attached to both James and Sarah and the source for Sarah is attached to both James and Sarah.
Detach James' source from Sara and detach Sara's source from James and everything will be fine. Don't know how the user that did these attachments back in 2017 managed to do this, but I think it is harder to do now than it was then with improvements made in the the routine for attaching sources over the past four years.
(dontiknowyou: Please just ignore this post. I know you are happy with your conclusions and do not need more discussion of this issue. But I feel an obligation to others who read this post and to FamilySearch's engineers who may read this post to give them as full an analysis of what is going on as I can.)
I've been looking at this issue again this morning and have modified my opinion a bit. I'll reiterate that the source linker is not broken and not even all that confused. It is just taking the source information on the Family Tree records and displaying it. The problem lies elsewhere in the past and is not currently something that will continue to occur.
On and off over the past few years, FamilySearch has apparently been working on ways to attach marriage record sources to marriage events in Family Tree instead of just individual bride and groom sources to the bride and groom respectively. I've seen a few varieties of this come and go. None of them have lasted very long. There must be something just inherently difficult in creating sources for groups of people in FamilySearch's sourcing routine in which a FamilySearch source is not just text, but contains a rich amount of metadata that can be used to find more information about an individual.
But back to the case at hand. If you look closely at the source on James and Sarah's couple relationship flyout, it is in a format that FamilySearch only used for a few months:
It is not an individual source, but rather a couple source. Viewing it shows this:
Note that it is attached only to the couple relationship and not to the individuals. This type of source is no longer created when attaching marriage records, at least I don't recall seeing them recently, so there must have been some problem with them and how they interact with the individual sources they were derived from which may be what is causing the source linker to display the source the way it is.
Examining this source further shows just one URL leading to the actual source which is actually not a couple source but a regular individual source:
However, this source is messed up because the title on top should be the name of the person in black lower down. so this should be the source for James not for Sarah. Clicking on Tomas Hill properly brings you to the source for him. Clicking on J W Summerfield properly brings you to the record for him. You can't click on James E Lewis. Clicking on Sarah D Lawrence, which you should not be able to do, brings you back to this same source. There is no way to come to version of this record with the title James E Lewis.
This same source with the title "Sarah D Lawrence, "Indiana Marriages, 1811-2007" does appear on Sarah's source page but there the proper links are present so you can click on James E. Lewis and jump to his version of the source:
I have no way of telling if these couple sources were uniformly created with problems or whether it was just this batch of Indiana marriages. I also have no way to tell if the source for Sarah originally had the linking problem and that is why the couple source is causing problems or if the creation of the couple source corrupted Sarah's source. I don't even think these couple sources actually exist in the databases of indexed records. There is certainly no way of searching for them when searching the databases.
Summary of current theory:
I assume that FamilySearch could come up with a routine to go through and detach all these problematic couple sources, thereby removing the display issue in the source linker. However, such wholesale editing of Family Tree records has, historically, been fraught with unforeseen complications that have caused more problems than they have solved. I'm glad FamilySearch is leaving it up to us users to remove these badly structured couple sources as we run across them.
I leave the debugging to the engineers.