Quick forum link: forum #193528 - Unofficial PSA: invalid_tag vs. *_(disambiguation) (May. 2016)

Ok, this is getting ridiculous. Let's address it before things get out of hand.

As of this post, there are:

~300 400 approved invalidated tags

~40 25 pending invalidations

• The more invalidated tags that exist, the less valid ones people can use to tag posts with
• The less valid tags people can use, the lower the possibility of a post being tagged with the more basic tags (see list & link above)
• The more posts without basic tags, the harder it is to find the post

Instead of invalidating them, these tags should be disambiguated (have the *_(disambiguation) qualifier added to the end of the tag) instead

i.e

• arm -> arm_(disambiguation)
• arms -> arms_(disambiguation)
• tail -> tail_(disambiguation)
• soft -> soft_(disambiguation)

etc.

So people can still use the tags to search/find content relevant to what's being tagged, while still being able to have less ambiguous tags for the more specific situations

-

There is an entire macroforum about this topic with links to example cases if you're still not sure why this could be an issue

Here is an actual example of what's being proposed in this forum topic:

forum #183516 - Tag Alias: shot -> invalid_tag (Feb. 2016)

Nah.
Hardly anyone bothers to actually clean those: https://e621.net/tag?name=*_%28disambiguation%29&type=&order=count.

So overall, the *_(disambiguation) experiment has been a failure.

Genjar said:
Nah.
Hardly anyone bothers to actually clean those: https://e621.net/tag?name=*_%28disambiguation%29&type=&order=count.

So overall, the *_(disambiguation) experiment has been a failure.

..Why is that considered a failure?

Was it really necessary to spam that message in 28 threads?

Genjar said:
Nah.
Hardly anyone bothers to actually clean those: https://e621.net/tag?name=*_%28disambiguation%29&type=&order=count.

So overall, the *_(disambiguation) experiment has been a failure.

I don't think it's entirely a failure. The biggest problem is the tags that have been disambiguated are ones that most people aren't going to care one way or the other about (precious stones, for instance). I think being more selective and using it as a tool rather than another way to invalidate tags would make cleanup a lot less daunting.

parasprite said:
I don't think it's entirely a failure.

I was mostly referring to the suggestion of changing all invalidated tags into disambiguated ones. Including ones such as the arm and mouth. Those would just need be manually removed, except for rare instances caused by forgotten underscores.

Disambiguation has occasional uses, but the amount of work it generates shows that it should not be used as the default.

I certainly don't see how being more selective about what to disambiguate is somehow 'ridiculous' or 'getting out of hand', like the op claimed. Disambiguation takes more work than simply invaliding the tags, and therefore we should consider carefully which ones are actually worth it.

or alternatively instead of using the clunky *_(disambiguation) suffix, "ambiguous" tags can simply be used as umbrellas

intersex is precedence for this

null0010 said:
or alternatively instead of using the clunky *_(disambiguation) suffix, "ambiguous" tags can simply be used as umbrellas

intersex is precedence for this

The suffix is to make them easier to find and clean up, but I do prefer umbrella tags over disambiguating.

Stuff that has too many mixed meanings doesn't work well for that and benefit from being an actual disambiguation tag. Bad example, but think if we had husky_(animal), husky_(artist), and husky_(body_type) to deal with. husky being an umbrella tag for that wouldn't really help much here.

Edit: Ew that dtext parsing

[I'll reply to everything in here soonish]
-
Here's another possible issue of the invalidation/disambiguation method:

TL;DR - manually move tags -> *_(disambiguation) instead of via alias, and leave the original tag as a catch-all for every ambiguous sub-category so people can use it without having to hunt for the disambiguated tag

I'm not addressing tag invalidation in the above because that was already done in the OP, and this post is already too long

NotMeNotYou said:
Was it really necessary to spam that message in 28 threads?

Sorry :c

But yes, I think it was/is;

because if we continue to invalidate the most basic, common tags on the site,
then people will have to suddenly become intimately familiar with a wiki and tag system that's already arcane to most

-
Nobody likes typing out *_(*) any more than necessary- even before disambiguation tags,

and if this keeps up, that's going to become the standard

Search suggestions, wiki guidance, *_(invalid) and other comments:

Personally I think the only failure with the disambiguation system is that it makes it harder to search for things easily without initiation. However, if the tag suggestions while searching worked better it could alleviate some if not most of the problems with finding stuff.

I.e. multi tag search query suggestions and click to update search, e.g. searching for diamond ring would suggest diamond_(gem) and clicking that would search for diamond_(gem) ring and not just diamond_(gem). Additionally adding wiki links ('?') to suggested tags might be interesting as well.

And another thing; regarding an alternative umbrella term, tagging diamond for both diamond_(gem) and the multitude of characters seems a bit weird to me personally, but I guess it could help people finding everything remotely associated with diamond easily, and then having to be a bit creative and figure out what the term they really wanted is called (unless they are smart and consult the wiki).

In my opinion a better search suggestion system would be more efficient and people would not have to hunt for the tag(s) they want since they would be presented with them anyway. One could additionally even put some extra marking/coloring denoting that one has searched for an invalid tag or a disambiguated tag.

Also why not instead of aliasing terms to invalid_tag or *_(disambiguation) instead alias them to *_(invalid)/*_(invalid_tag), that way ambiguous tags are kept separate from purely invalid tags, but the invalid tags are still there and visible but clearly marked so that they can be exchanged for a better tag instead of just being "blurred" into invalid_tag.

invalid_tag is pretty useless as it is now, it is often no more useful than aliasing something to nothing (if that was a thing). I'm not sure if this should be done for every invalidated tag, but different tags have different "themes" and grouping them all together seems a bit excessive.

That the disambiguated tags hold "a lot" of posts is of less concern to me. Also one way to limit the number of posts in disambiguated tags could be to clean them out before they are disambiguated, often the high number of posts in these tags is due to the following event chain:

• Someone announcing "hey, this is used for lots of different things".
• People go "yeah, sure".
• Admin agrees with "approved" and an alias is created.
• A wiki is hopefully created to list alternative tags.
• Everyone forgets about it, because the problem is "solved".

As a note: A high post count can also discourage people to change things, while if you add a tag and notice that it has only 1 post people will be more inclined to replace that with something else.

titanmelon said:

TL;DR - manually move tags -> *_(disambiguation) instead of via alias, and leave the original tag as a catch-all for every ambiguous sub-category so people can use it without having to hunt for the disambiguated tag

Just to make sure I don't misunderstand something, e.g. are you suggesting something like (counts are random):

Instead of:

• diamond_(disambiguation) (300) <- diamond
• diamond_(gem) (200)
• diamond_(...) (100)
• ...

You want:

• diamond (800)
• diamond_(disambiguation) (300)
• diamond_(gem) (200)
• diamond_(...) (100)
• ...

I suspect that would work worse when it comes to keeping things organized (and disambiguated) in the long run. "Cleaning it out" requires searching for the tag itself and negate all disambiguated tags, something which is quite cumbersome (note that negative wildcards don't work), especially if there are many of them.

The only positive thing would be easy, but not so useful searching of the umbrella term, but I'm not sure it outweighs the negative. But I get the feeling this is just my personal opinions talking.

titanmelon said:
[...] because if we continue to invalidate the most basic, common tags on the site, then people will have to suddenly become intimately familiar with a wiki and tag system that's already arcane to most

I'm not sure if that's what we're doing, we're invalidating generic tags, tags that usually have very little search value while sometimes spanning a ridiculous amount of posts. E.g. image and leg. But I see see your point, it can get excessive and maybe is already inching that way.

titanmelon said:
Nobody likes typing out *_(*) any more than necessary- even before disambiguation tags,

I'm just curious, why would you want to search for that unless you're dealing with something directly related to tagging? Sure you'd find all kinds of various posts related to what you want but not what you really want. As a note you can usually also get away with xxx_*.

However this problem of finding something specifically exists no matter what you do:

• If using an umbrella term/doing nothing:
• diamond wouldn't help you find diamond_(gem)
• If using disambiguation:
• diamond wouldn't help you find diamond_(gem)
• diamond* wouldn't help you find diamond_(gem)
• diamond_(*) wouldn't help you find diamond_(gem)

The only thing that would help would be:

• Better search suggestions (or live search suggestions)
• More efficiently lead users to the wiki (if that's the issue), especially if the system detects a user that is having problems finding something.

titanmelon said:
I'm not addressing tag invalidation in the above because that was already done in the OP, and this post is already too long

Hah, your structured and summarized content is nothing compared to my long chatty monologues, I think people sometimes get tired of hearing me talk :P

Speaking of which, a completely off topic question: I sometimes try to "hide" my content in closed sections, do you guys think this is a bad or good idea? I.e. easier to navigate page vs unable to search and have to open them.

Chessax said:
Speaking of which, a completely off topic question: I sometimes try to "hide" my content in closed sections, do you guys think this is a bad or good idea? I.e. easier to navigate page vs unable to search and have to open them.

I'd say that it's slightly on the 'bad idea'-side, simply because sections make it harder to reply to parts of posts. Partial formatting usually needs to be removed from the reply, or else it ends up messed up.

@Chessax:
Those are some really good points, I'll try to finish writing a full reply soonish

One major thing you mentioned somewhat indirectly was the use of wildcards to find all the disambiguated posts (diamond*)
Which could be a potential substitute for keeping diamond searchable normally

The problem there is whether on people would intuitively do a search for diamond*, rather than just diamond. And knowing when there's a need to use wildcards or not

Especially since there's not current standardized form for disambiguated tags, aside from an underscore (*_*)

Here's an example of an invalidated tag I think is worth bringing up again:
fluids

What are people using to identify generic fluids?
ambiguous_fluids?
There's still the issue raised in the above posts + OP;

If we disable the most intuitive tag options,
How do we know if people are aware of the alternatives?

Keeping in mind that anything aliased to invalid_tag is automatically made useless for regular searching/tagging purposes

Invalidated humanoid appendages

Another example, regarding invalidated (humanoid) appendages:

post #942318

On character sheets, appendages are a prominent area of focus, but you won't be able to search for character sheets based on what appendages they contain (arms, hands, legs, tail etc)

Instead, you have somehow know in advance and search *_arms, *_hands, *_legs etc.

-
Regarding the counterpoint, 'you can use alternative tags to find the invalidated one'
has been addressed here: https://e621.net/forum/show/200358

tl;dr - The problem still exists because it assumes the person searching for something already knows what tags are available in advance

Not mention how roundabout and time-consuming it is in comparison to just...search for the tag you want to find

Speaking of tail, there are >50 listed tail-related tags

But want to search for just posts with tails in them?
Want to blacklist posts with tails?

Too bad. You'll probably have to figure out all the possible wildcard combinations and hope the specific ones get tagged in

titanmelon said:
Speaking of tail, there are >50 listed tail-related tags

But want to search for just posts with tails in them?
Want to blacklist posts with tails?

Too bad. You'll probably have to figure out all the possible wildcard combinations and hope the specific ones get tagged in

Funny thing is, that was discussed yesterday when I tried to use "tail" on my post and got redirected as an invalid tag. However, I don't know if we should use just "tail" as an umbrella tag, because it can also be used for certain hairstyles.

titanmelon said:
Speaking of tail, there are >50 listed tail-related tags

But want to search for just posts with tails in them?
Want to blacklist posts with tails?

Too bad. You'll probably have to figure out all the possible wildcard combinations and hope the specific ones get tagged in

I imagine this to be a small problem, it would be easier to use a dedicated tag due to almost all standard furries having tails, e.g.tailless/no_tail, i.e. as tailless/-tailless, however that tag is ridiculously underused (e.g. see just worgen, which dwarfs (hehe dwarf) the tailless tag).

Slightly related I'm wondering if the blacklist could somehow be improved a bit to allow semi-wildcards, i.e. you can add *_tail or tail_* which will be internally saved as the entire list of expanded tags, but presented simply as *_tail or tail_* (for editing purposes), the problem with this is that the available space in a cookie is very limited so it would more or less require it be converted to local storage instead.

Another issue would be when to update the blacklist to reflect any new changes to tags (or how to note the user about that it might be needed), since the blacklist is static and does not respect changes to aliases etc., IIRC.

I dont understand the concept of "invalid_tags" its tag what you see right?

Ok sure some tags are arbitrary. But you dont always see a tail. Does that mean I should tag no_tail when I dont see it?

No. It doesnt.

titanmelon said:
One major thing you mentioned somewhat indirectly was the use of wildcards to find all the disambiguated posts (diamond*)
Which could be a potential substitute for keeping diamond searchable normally

The problem there is whether on people would intuitively do a search for diamond*, rather than just diamond. And knowing when there's a need to use wildcards or not

Especially since there's not current standardized form for disambiguated tags, aside from an underscore (*_*)

My hope is that forum #201842 will fix many of these issues if implemented.

titanmelon said:
On character sheets, appendages are a prominent area of focus, but you won't be able to search for character sheets based on what appendages they contain (arms, hands, legs, tail etc)

Instead, you have somehow know in advance and search *_arms, *_hands, *_legs etc.

Going on slight rant here, I couldn't bother condensing it, so...

While I won't argue about it being easier to find something if it has a dedicated tag, but there are a few issues:

For it to be effective, considering these tags spans so many posts, they have to be very well tagged. Because using the tag alone is pretty much pointless which makes it much more interesting to use in combination with other tags, i.e. model_sheet however if 95% of all posts are tagged correctly that still means every single model_sheet post could be missing the tag. Of course you could set up a project to add these to all model_sheet posts but then you will have other places lacking. It's just a massive undertaking to get this working effectively. Not saying you can't only that it's a lot of work for what might be very little payoff.

Continuing along these lines, it's also the question if people would use these tags when a tag is as general as these would be. I don't know but I feel like searching for arm, knee, leg, head, body and finger etc. are not a priority for most people. I could imagine artists or character owners doing this for references, but then it would be easier to simply look through model_sheet, or using a more specific tag like foot_focus, headshot_portrait, or something else in combination with close-up. However even in a model_sheet this seems superfluous since people pretty much expect to find these things when they search for model_sheet. Ear is the only one I could see actively used due to scalies/marine/etc.

GDelscribe said:
I dont understand the concept of "invalid_tags" its tag what you see right?

Ok sure some tags are arbitrary. But you dont always see a tail. Does that mean I should tag no_tail when I dont see it?

No. It doesnt.

Except not all tags describe visible things, here are some obvious examples of invalid_tag:

• image (useless because of metatags, type:png, type:gif, type:jpg, also ambiguous (e.g. drawing, painting etc)
• 1:1 (ditto, ratio:1)
• stuff (what is stuff?)
• large/big (large/big what?)
• hot (completely subjective (as well as ambiguous))
• not_gay (yeah, let's tag everything with not_flying_spaghetti_monster, but of course not on the 7 flying_spaghetti_monster posts)
• cool (everything is cool, that's why cool is cool (also ambiguous))
• (artwork) (mistake because someone used instead of _)
• etc. etc. etc.

A few might be better disambiguated or even kept (as discussed earlier), but many are completely useless.

And I don't think anyone ever suggested using "tag what you don't see" instead of twys. Also it's not literally "tag what you see" it's "tag what the general viewer sees", i.e. not subjective but objective tagging. You would tag tailless on a character whose tail would be visible if it had actually had one. Especially if on a character where it would be assumed that they have a tail, e.g. on a species such as worgen which implies the more general werewolf.

GDelscribe said:
I dont understand the concept of "invalid_tags" its tag what you see right?

Ok sure some tags are arbitrary. But you dont always see a tail. Does that mean I should tag no_tail when I dont see it?

No. It doesnt.

Chessax said:
And I don't think anyone ever suggested using "tag what you don't see" instead of twys. Also it's not literally "tag what you see" it's "tag what the general viewer sees", i.e. not subjective but objective tagging. You would tag tailless on a character whose tail would be visible if it had actually had one. Especially if on a character where it would be assumed that they have a tail, e.g. on a species such as worgen which implies the more general werewolf.

We have a few "inverse" tags that are used when something doesn't exist rather than when it does. Examples are no_sound and no_sclera (and maybe featureless_crotch?).

TonyLemur said:
We have a few "inverse" tags that are used when something doesn't exist rather than when it does. Examples are no_sound and no_sclera (and maybe featureless_crotch?).

Also unseen_character. I think these four tags are a good example of when to tag things which are not there: because these things are expected to be there given the content of the post, but for whatever reason, they're not.

TonyLemur said:
We have a few "inverse" tags that are used when something doesn't exist rather than when it does. Examples are no_sound and no_sclera (and maybe featureless_crotch?).

Yeah, maybe my thoughts didn't translate that well, I meant what @DragonFox69 said. I.e. tagging things like not_black_hair, not_female, no_coffee, etc. is not only pointless, but harmful, while other no_* tags are fine for reasons explained.

Chessax said:
I bet you could get some humor post which accurately depicts no_coffee.

It ain't under humor, but post #911265 could fit "no coffee," though that's more of an assumption than an accuracy.