Correcting indexing errors, and making it easier to link records
I've found some weird indexing errors, especially when I'm working with German handwritten records. Places where the records change year, but the indexer did not. Indexers who confuse "Beerdigung" with baptism. Often I can find the record image on Ancestry, when it isn't available via Family Search--if I can, I'll make the correction, but sometimes that's impossible.
Also, I get really frustrated with name variants. I would really like to be able to ignore the first name and use only the middle names for comparisons. Possibly some kind of switch? Consider: Hans Jerg and Johann Georg are actually the same name. Either would show up in some records as Jerg or Georg. But they would NEVER be the same as Johannes. Nor should they be confused with Johann Conrad, Johann Martin, Johann Michael, or Johann Xavier!
One thing I'd REALLY like would be a checkbox option when I choose Show All under Research Help. I'd like to be able to select multiple record "matches"--and delete them all at once. That would be especially nice when my subject is one of the Johann "somethings". Please, please, on this one?
Well, when you dismiss multiple hints in a row, it copies your reason from the first one into the box for the subsequent ones, so at least you don't need to retype your explanation each time. It's not quite the same thing as dismissing them all at once, but it does speed things up considerably.
FamilySearch's matching algorithm is weird about which name equivalents it "knows" and which ones it doesn't. I searched for Farkas and got mostly Wolf, but if I search for Imre, it doesn't find Emeric, and if I search for Ferenc, it doesn't find Ferencz. There are also some names that it equates incorrectly, at least for Hungarian. (For example, it files Örse with Ursula, when in fact it's a form of Elizabeth.) I don't know of a way to "teach" the algorithm that Jerg and Georg are the same thing. (I suspect that it does know about Hans and Johann, but I haven't tested it.)
I think the main reason it doesn't handle multiple names correctly for your purposes is that there are too many variations in how languages or cultures handle them. It seems to me that the algorithm is weighted mostly toward American usage, which actually makes some sense, since that's what the programmers know and what most of their users need most of the time. When searching American records for a John George, you need the algorithm to cough up all the plain Johns and the John Gs, because that's how he'd have been recorded most of the time. You also generally want to at least look at all the John SomethingElses, too, because since middle names are not actually official, they're subject to a lot of variability. Hence giving you Johann Martin when you searched for Johann Georg is usually a feature, not a flaw.
The way around the flaws of the search algorithm basically boils down to multiple searches. If you know a name equivalent, but the algorithm doesn't, you'll just have to search both ways, and if your actual matches are getting lost amidst the Johann SomethingElses, you can try a bunch of searches with the "exact" box checked. Or, hmm: what happens if you search for Jo* Georg*? That combined with the exact box may be a way to remove the Johann Martins from your results. (You'll still get the Josephs, though.)0