Wikidata:Edit groups/QSv2/75178

Edit group QSv2/75178

Summary	{{{summary}}}	Author	Quesotiotyo
Number of edits	639 (more statistics)	Example edit	Q104535596

Discussion

@Quesotiotyo: This is not an acceptable batch edit. While one can appreciate the intent. Removing family names in palce of givne names does a fair amount of harm.--Labattblueboy (talk) 22:16, 31 January 2022 (UTC)[reply]

@Labattblueboy: Adding these values was the problem, not removing them. No harm has been done here. --Quesotiotyo (talk) 00:02, 1 February 2022 (UTC)[reply]

@Quesotiotyo: Right now the consensus donesn't appear to favour this view however I think it best to see what other comments arise before any further action.--Labattblueboy (talk) 00:05, 1 February 2022 (UTC)[reply]

(edit conflict) @Quesotiotyo: As best I can tell, it appears that you've removed all instances of family names entered as given names. Please revert this unless you plan to immediately follow it up with replacements. There are many items for people that have middle names that have items only as a family name and are entered as such; see e.g. Frank Brackett (Q99522757). It's not ideal, but by removing them with no replacement, you are destroying useful data. This topic has been raised before, e.g. at Wikidata:Project_chat/Archive/2021/05#Family_names_used_as_middle_names, and it is not at all clear that there was consensus for your bulk edit. {{u|Sdkb}} ^talk 22:16, 31 January 2022 (UTC)[reply]

That discussion is definitely not a consensus for such a bulk edit. I won't address this discussion further and will leave any further comments for the admin noticeboard.--Labattblueboy (talk) 22:34, 31 January 2022 (UTC)[reply]

@Sdkb: I have not removed or destroyed any useful data. Family name (Q101352) items should not be used with the given name (P735) property. Over the last two days I have corrected by hand nearly 500 statements where the equivalent given name item already existed. The statements that I removed in this batch did not have such equivalent values. In either case, having an incorrect value is not useful. Please remember that the values for statements of Wikibase item properties are QIDs and not strings. Just because a given name item and a family name item have equivalent labels does not make them interchangeable.

--Quesotiotyo (talk) 23:52, 31 January 2022 (UTC)[reply]

I would have no issue with this edit if there was a plan in place to replace/create replacement values. I wholehearted disagree that absent values are more favourable than "incorrect". In fact, for sister programs like the commons this edit is quite harmful. I continue to argue it should be reverted.--Labattblueboy (talk) 00:00, 1 February 2022 (UTC)[reply]

@Labattblueboy: Can you please elaborate on how an incorrect statement is valuable at Commons or anywhere else? I am genuinely curious. (It will save me many hours of my time spent fixing mistakes if this turns out to be true! :))

--Quesotiotyo (talk) 00:24, 1 February 2022 (UTC)[reply]

Middle names is one group that comes to mind as a great many times it's a family that becomes a "given name"; Generally as a an second or third ordial. Worst case are those that are commonly known by their middle name and such an edit deletes the given name entirely.--Labattblueboy (talk) 13:36, 1 February 2022 (UTC)[reply]

@Labattblueboy: Yes, but middle names are given names and require the use of a given name (Q202444) item (or a subclass of it). Using the string-equivalent family name item only appears to be correct because the Wikidata UI displays labels in place of QID numbers and the labels are identical; the statement is still fundamentally wrong.

--Quesotiotyo (talk) 16:15, 1 February 2022 (UTC)[reply]

I'm kinda confused why this edit batch is controversial. The name statement removed were wrong right? I agree it would be better to replace them with the proper statement but this is at least a step in the right direction. I wouldn't let the perfect be the enemy of the good. Then again I am not sure there's a ton of value of structuring middle names like this anyways (and have avoided getting involved in that kind of data so maybe I'm out of the loop). BrokenSegue (talk) 01:29, 1 February 2022 (UTC)[reply]

@BrokenSegue, putting aside the behavioral issues, which are being handled at the AN thread, I'll go into a little more detail on the example and my thoughts overall.

Frank Brackett (Q99522757) is Frank Parkhurst Brackett, and had the middle name expressed like this, using Parkhurst (Q21504860), the item for "Parkhurst" as a family/last name, because there is no item for "Parkhurst" as a given/first name. In that situation, it was shoehorning a family name into given name (P735), yes, but it at least gave his full name, and I'm guessing "Parkhurst" was the last name of one of his ancestors anyways, so it's not really a severe misuse. In the batch/its reinstatement, all mention of Parkhurst was removed from the item's statements, so that information is destroyed in a way that'll make it difficult to recover if the batch is not reverted.

Now, there are a few ways we could resolve the shoehorning issue, which is the topic I brought up several months ago in the project chat thread. One solution proposed there was to just create new items for any family name ever used as a given name. I don't think that's really optimal, though, since Parkhurst fundamentally isn't a given name—it's a family name that can be used as a middle name, as it is here. An item for Parkhurst as a given name would probably only ever be used correctly in this one item, but it'd very likely confuse someone at some point trying to use it as a last name. There are several other possible resolutions, all of which have tradeoffs, but fundamentally, this is something the community needs to discuss further to decide what we want. And until that happens, we shouldn't be destroying information—it'll be a lot simpler to do some sort of conversion at Frank Brackett (Q99522757) from its status quo ante to whatever system we ultimately use than it will be to recover the middle name if it's allowed to remain removed. Best, {{u|Sdkb}} ^talk 03:04, 1 February 2022 (UTC)[reply]

@Sdkb: A given name item for "Parkhurst" was created (Parkhurst (Q110772478)) and has been added to Frank Brackett (Q99522757) and eleven other individuals.

"In that situation, it was shoehorning a family name into given name (P735), yes, but it at least gave his full name, and I'm guessing "Parkhurst" was the last name of one of his ancestors anyways, so it's not really a severe misuse."

Brackett's full name was in fact not previously given, at least in the sense of structured data. The item for his middle name only appeared to be correct due to the fact that Parkhurst (Q21504860): family name is composed of the same string as the given name "Parkhurst" and therefore would have the same label. The two items are not the same, which is why they must be represented by separate items to be used with the correct property.

"In the batch/its reinstatement, all mention of Parkhurst was removed from the item's statements, so that information is destroyed in a way that'll make it difficult to recover if the batch is not reverted."

"And until that happens, we shouldn't be destroying information—[...]"

Given name and family name statements are derived from labels and aliases, not the other way around. No information has been destroyed.

--Quesotiotyo (talk) 16:33, 1 February 2022 (UTC)[reply]