I agree completely.
I write my own citations and always have. I tend towards a Chicago-style citation, but this is what Mills based her Evidence Explained style on, so it's pretty close.
I find Ancestry's citations appalling and FamilySearch's somewhere between mediocre and bad with an occasional foray into appalling.
Their citations that refer to their own database names are ridiculous. It tells you nothing about the actual record, where the original record was from, etc. Sometimes it's the equivalent of a citation that tells you the street address of the library you were in when you looked up a book in their card catalog and the book's Dewey Decimal number for that library, but nothing more. Title? Author? page #s? Publisher? Year published? Nope.
I don't really care where you were when you saw a copy of the record; I want to know what the original record is.
I feel like a broken record on this sometimes, but a good citation is going to tell someone else where to find that original record when those websites and databases no longer exist. It should do that without needing a separate extract of the entire contents.
This (for example):
"United States Census, 1910", database with images, FamilySearch (https://www.familysearch.org/ark:/61903/1:1:M2Z2-T24 : Tue Jul 18 17:25:53 UTC 2023), Entry for Walter J Montgomery and Mrs. Nellie M Montgomery, 1910
doesn't tell me where to find this record if I can't look at it on FS. But this:
1910 United States Census, Greenleaf, Washington, Kansas, 15 April, 1910, page 7601/stamped 85, SD 5, ED 151, Sheet 1A, family 1, household of Walter J. Montgomery, lines 1–3, https://www.familysearch.org/ark:/61903/3:1:33S7-9RK3-CZL, DGS 004971572, image 730, citing National Archives and Records Administration publication T624.
tells everyone exactly what I was looking at, lets them find it in another repository if they want or need—and even lets them see that maybe they're looking at a slightly different record for those cases we've all seen when someone got double counted.
Beyond the census records, I find they're often not clear about exactly what a record is in their description. They might say "Marriage record" but it's actually a bond. Or it's a transcription of the bonds. Or an index.
I don't have hope for improvements from any of the big websites and just keep doing my thing.