NSFW Levels: A flexible and practical proposal to label NSFW content with more granularity

Note: this post in marked as NSFW because it contains descriptions of sexual acts and non-specific descriptions of NSFL content for the purposes of classification.

As an admin of a furry instance that also allows nsfw furry artwork, I am always wary about the impact our content has on remote lemmy instances.

There is a delicate balance between allowing our users to post the content they enjoy and making sure that we won't be defederated from remote instances whose users might not appreciate our content.

I believe that users should not expect the ALL feed to curated to their interests, but even then I can understand that some users and instance admins might want to find a way to avoid certain type of content even on the ALL Feed.

In order to solve this problem I have tried to come up with a flexible and practical way to label NSFW content with more granularity.

The goal of this system is to more effectively classify NSFW communities in a way that is more practical than simply having a large one-size-fits-all label, while at the same time not increasing the overhead for instance admins to moderate this content

Advantages:

Can be implemented in the current version of lemmy
The burden of classification falls on the creator of the community or the admin of the instance that hosts it
Absolutely no impact on performance
It is granular, giving remote admins choice on which nsfw level they want to allow, hide or block
Once set up it would work for existing and future communities without having to do any extra work

Disadvantages:

It's a new spec that would have to be communicated
It requires some sort of database trigger or other simple tooling to implement any sort of filtering functionality.

How it works:

It works by adding a specific 'tag' describing the level of NSFW content in the description of a community. Instance admins can create a database trigger to automatically hide or remove communities with a NSFW level above their preferred threshold

List of NSFW Levels:

NSFW-level-0: alcohol, drugs, gambling and other non-sexual content that you'd generally want to keep children away from.

NSFW-level-1: Artistic nudity, general nudity, adult humor, Mature artwork or images

NSFW-level-2: Vanilla sexual explicit content

NSFW-level-3: Common sexual interests such as foot fetish, food play, being tied to a bedpost, etc...

NSFW-level-4: Niche sexual interests such as bdsm, tentacles, inflation, etc...

NSFW-level-5: Heavy version and much more niche sexual interests such as heavy bdsm, breath play, rubber encasement, etc... This category might also contain contact with bodily fluids that some people might consider triggering

NSFW-level-6: Extreme content potentially involving fictional non-gorey mutilation, torture, non-con, as well as consumption of bodily fluids that some people might consider triggering

NSFW-level-7: Equivalent to NSFL. Might contain heavy or extreme versions of fictional mutilation, torture, non-con, etc... as well as content such as videos of accidents resulting in death, amputation, visible gore, etc.

NSFW-level-8: Extreme NSFL, usually irl content. I don't want you to imagine it, just know that it exists and as such needs to be added to the spec so that it can be automatically be blocked.

The description of these levels has been left intentionally vague so that it can grow organically.

The idea is thus that each NSFW community adds a classification to their description. This could even be done in from of an invisible markdown link such as this:

[](cw:nsfw-level-3:foot fetish, light retrains)

The text description is optional but could help give admins necessary clues about the content.

With communities tagged, instance admins can easily and automatically hide or remove communities above the threshold they decide.

Hiding a community is achieved by setting the "hidden" property in the database to 'true' for that community. Only people who have explicitly accessed the community via url and then subscribed to it will be able to see any of its content. It will not be visible in the All Timeline, Local Timeline or even search results.

There might be a situation in which the admins or moderators of nsfw community are so used to certain content that they classify it with a lower score. If that happens it's just a matter of lowering the tolerance for that instance and instead of filtering out at level 6 and upwards, you filter out starting at level 5 (for example).

Anyways, the core idea is that by classifying NSFW content with a numerical scale, instance admins can automatically filter out content above a certain threshold.

2 comments

It seems suboptimal to put sex and gore on the same axis. The different levels also seem fairly arbitrary, e.g. you put foot fetish at a lower level than BDSM (a catchall descriptor covering tons of different acts including bondage, which you have an example of in a lower level). It's also very, very difficult to quantify normal/niche/extreme. Remember when England tried to classify face sitting as extreme back in 2014? https://www.cnn.com/2014/12/12/world/europe/uk-porn-protest/index.html Some common fantasies could also be considered extreme ones, like this one small study shows https://pubmed.ncbi.nlm.nih.gov/19085605/ But basically the vagueness foists the decision of what fits into these categories onto mods and I think that's a bad idea. Tags might be an ok way to filter things out but by the time you know a tag you don't want to see, you've seen it.
Stumbled upon this and have some thoughts on it; this is something I've been thinking about myself for a while.

I agree with Xero's concerns that the levels seem a bit arbitrary, and I've got a "counter proposal" for how a tag based system could work:

The local instance's community description contains a number of entries like the following:

[](cw:defaultallow:yiff)

[](cw:defaultblock:feral)

[](alwaysblock)

The instance itself will have a list of these tags and what each means available, both for community moderators and remote instance admins.

The remote instance, in its settings, has two lists of tags, a whitelist and a blacklist.

A community is automatically marked as hidden when any of the following are true:

A defaultallow or defaultblock tag apperas in its blacklist.

A defaultblock tag does not appear in its whitelist.

The alwaysblock entry is present.

This would allow remote instance admins to freely customize what they allow and disallow, while giving the local instance the ability to set sensible defaults in case something is missing from the remote's blacklist. This does put a bit of pressure on the local instance admin to specify what it wants this "sensible default" to be, but that's something that they already need to do when specifying where they draw the line between what's NSFW and not NSFW anyway.

Some advantages for this is that it could be expanded to handle content warnings in post titles as well, and allow instances to filter things that wouldn't traditionally be seen as nsfw (such as politics, bot posts, AI art and so on).

Just some thoughts I had, hopefully it all makes sense.