It’s kind of problematic, of fetish content being marked mature, since there really aren’t clear lines. A lot of fetish content comes across as innocuous, though perhaps a bit weird, if you don’t know it’s a fetish. In some cases you could find something quite similar in a children’s cartoon. Indeed, that’s probably where many of these fetishes originated. You might have two artists produce drawings with similar content, and it will be a fetish for one but not the other.
Exactly, fetishes that have SFW variants shouldn't be marked as mature/adult. However, what FA considers to be graphic can be caught by a tag-based safety net.
Tags that the "Mature"-filter can "safety-catch" is obvious things well-agreed in spirit of the ToS like "gore", "censored", "nudity + <list of genitals>", "torture".
It shouldn't be considered a full solution but rather an extra catch to protect minors, etc. It makes it so in the edge case "people forgot to filter but tagged correctly" that it doesn't get seen by minors. No blacklist is a solution, but I can tell you for sure that personal blacklists has nothing to do with the issue of bad NSFW filtering.
I think it's pretty bad to remotely suggest that kids as young as 13 should have to remember sexual tags so personalised blacklisting is no method to resolve the issue of minors (and others who want SFW) from seeing NSFW content.
Let's consider the possible form types of blacklist:
1. A predefined mature blacklist: "gore", "censored", "nudity + <list of genitals>", "torture"
2. An predefined adult blacklist: "<list of genitals>", "<list of adult objects>", "<list of sex acts>"
3. A personal blacklist that aims to blacklist mature or adult ends up identical to the obvious above, and like the above "is a NSFW filter".
4. A personal blacklist that aims to blacklist fetishes ends up censoring fetishes regardless of the NSFWness, and unlike the above "is not a NSFW filter".
Since most arguments in this thread by the supporters of blacklists are following the notion of 3/4, when 3 and 1/2 have the same effect. Then I think it's clear to conclude there is most definitely an ulterior reasoning to have blacklists than "NSFWness", which people have said is to a) blacklist fetishes (which search already can do), and b) blacklist identities they don't like. Types 1 and 2 work well for safety nets to protect minors, 3/4 do not, as said minors/etc would have to each manually enter a blacklist full of NSFW terms which isn't minor-friendly. To resolve (b) within 4, is to prevent the blacklisting of identities, at least with SFW content.
So while 1/2 and 3 are identically effective, 3 puts the burden of responsibility of making NSFWness-guaranteeing blacklist lists on the user (e.g. minors/etc having to know sexual terms which isn't very PG) which doesn't at all protect kids (or adults who are triggered by NSFW and its terms) from interacting with NSFW at all.
While 3 + 4 or just 4, is equally effective as 1/2/3 with the addition of the caveats such as oppression blocking and apartness on everyday users, while the search system already provides adequate solution to 4. Type 4 is irrelevant to preventing NSFW leaks (unlike 1/2/3) and shouldn't use NSFW leaking into SFW as an argument as it's a false-equivalence.
It doesn't resolve NSFW content leaking into SFW mode. 9 in 10 times when someone sees a NSFW piece in SFW, it's when someone forgot to filter (or is unsure about the rules for the filter) but remembered to tag with NSFW keywords. I think this is because they use poster bots like PostyBirb which have a user interface/template-system that is confusing to some as it has been a complaint. Often it leads to a user entering the right tags but forgetting to select the right template setup. Both failing to filter and failing to tag happens, however, so the solution is not as trivial.
In the end of the day, blacklists won't at all fix the issue at all of minors seeing NSFW in the SFW sections, but least those with pet-peeves towards certain subjects will be happy. These complaints won't cease after blacklists exist.