![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
luzulaI was going through old folders and found this file from 2010, when I was a tag wrangler. I'd almost forgotten this, but I was actually the one to suggest that we use meta-tags for characters and pairings (as opposed to just for freeform tags and (I think?) fandoms). The plan before that was to use some sort of disambiguation pages for ambiguous character and pairing tags, sort of like Wikipedia.
Anyway, I thought it was an interesting snapshot of AO3 and tag wrangling, so I'm posting it. I'd forgotten that bit about how much work it used to be to re-name a canonical tag which had lots of other tags attached to it! You had to manually unhook all of them and then re-hook them afterward.
Suggestion for using meta-tagging to handle ambiguity in character and pairing tags:
As I understand the current situation, we are disambiguating by fandom for characters (and pairings) that share the same full name. Later, we are planning to have ambiguity tags which lead to a disambiguation page where one can choose between the various disambiguated tags (kind of like the disambiguation page on Wikipedia).
I have a suggestion for instead using metatagging for the problem of ambiguous character names. I'll explain what I mean first, and then give arguments for it.
I think we are getting hung up on having to create a unique tag for each character. But that is not actually the goal here. The goal is to organize and connect the tags on the users' works in a way that makes it easy for other users to find the works that they want.
If "John Smith" is a character name for several different characters, I suggest we simply canonicalize it and attach it to all the fandoms where John Smith belongs. If a user wants to find a particular John Smith, they click on "John Smith" and then filter by fandom.
If a user tags their fic with just "John", then that tag will be an upper-level metatag connected to all the different Johns one level below: John Smith, John Sheppard, etc... Again, a user could click on "John", get all fics tagged by any of the various Johns, and then filter by fandom to find a particular John.
There's a place for disambiguated tags here, too, because they are sometimes useful for users. The disambiguated tag "John Smith (Fandom X)" would be a canonical that's connected to the (also canonical) metatag "John Smith" one level up. Note that this would require several levels of metatags, because "John Smith" is connected to "John" one more level up.
So how would the wrangler side look? Well, all pairing tags containing John Smith would be attached to "John Smith", no matter what fandom the other characters are in. Likewise, all pairing tags containing John would be attached to "John". Note that this would make it possible to collapse the cast field "John/Rodney, John, Rodney" into "John/Rodney", which is not possible today.
The tag "John/Rodney" would be synned to "Rodney McKay/John Sheppard", as today, as long as there were no other John/Rodney pairings. When someone wrote a fic tagged Rodney McKay/John Smith, then the tag "John/Rodney" would be made into a pairing metatag connected to both the tags "Rodney McKay/John Sheppard" and "Rodney McKay/John Smith" one level below.
What about synonyms? Well, they would be synned at the appropriate level. That is, the misspelling "John Smithh" would be synned to "John Smith", while "Captain John Smith" would be synned to "John Smith (Fandom X)" if such a tag exists and if Fandom X was the only fandom where John Smith was a captain. If another Captain John Smith appeared in another fandom, or if there is no disambiguated tag, then the tag "Captain John Smith" would be moved up one level and synned to "John Smith".
Okay, I think that's it.
Arguments for using meta-tagging instead of ambiguity tags:
1) It is more in line with how users tag their works. I don't have statistics on this, but I would guess that most users don't intuitively tag their work with "John Smith (Fandom X)" when they have already tagged for fandom X a few lines above. They think that it's enough to have "John Smith" or "John" when you already know what fandom the work is in.
2) It spares work for the wranglers. They wouldn't have to research whether or not a character name is unique or not, and then disambiguate it if it's not. Besides, it's a subjective decision whether a name is unique enough, and people make that decision quite differently. As the archive grows, every character name could potentially need to be disambiguated. That's a lot of work, and not actually necessary.
3) It makes for shorter canonical tags. (Shorter tags might also be good for performance, is that right?)
4) It saves work for the coders—they wouldn't need to write code for ambiguity tags and disambiguation pages, etc. Meta-tagging is something we want anyway.
5) It makes canonical tags more stable. For example, if I were to write RPF for John Sheppard the 16th century composer, then the tag "John Sheppard" would suddenly need to be an ambiguity tag, and the 1292 (and counting) fics with that tag would suddenly be stuck with a tag that's not canonical any more. "John Sheppard (SGA)" would have to be created, but it wouldn't solve that problem. If "John Sheppard" is a shared canonical between all fandoms from the start, there is no problem.
6) It eliminates the gray zone between metatags and ambiguity tags. In the wiki, "The Doctor" is used as an example of a tag which would be a metatag, linked to all the different characters in Doctor Who (and potentially other doctors, as well). But "The Doctor" could just as well be an ambiguity tag, which led to a list of the different doctors. How would you decide if it was to be a metatag or an ambiguity?
7) It makes sure that character names are not implicitly assumed to belong to a fandom just because there's no disambiguation. By contrast, in the current situation John Sheppard is automatically assumed by most people to be the SGA character, just because he's the most popular character with that name and there's no disambiguation.
8) It makes for better filtering. I don't think there's actually a way to click on a tag and bring up all fics tagged with the character you want in one step, either with disambiguation or with metatags. But metatags definitely do the job better.
So, say you're looking for John Smith in fandom X. With metatags, you click on "John". Then you get everything tagged "John", "John Smith", and "John Smith (Fandom X)". A drawback is that you also get fic tagged with "John Sheppard", but you can always filter for Fandom X afterward. Also, you don't get the fic that's tagged with just "Smith". So you also bring up the tag "Smith", then filter for fandom, and you're done. That's two different tabs.
In the scenario with ambiguity tags, you can start at the tag "John", and go to the disambiguated tag "John Smith (Fandom X)". But then you only get the fics tagged with that specific tag, so you also need to bring up all fics tagged with just "John" (which I understand would be an option on the disambiguation page), and filter for fandom. But you're not done, because there's also the tag "John Smith", which I can think of three different options for here. It's either
- unwrangled, which would be unfortunate since it's probably the most popular tag for the character (it can't be canonical or synned to a canonical, since it's ambiguous).
- synned to either of the ambiguity tags "John" or "Smith" (how would you choose which one, though?)
- an ambiguity tag at an intermediate level, so that from "John" you could disambiguate to "John Smith" and from there to "John Smith (Fandom X)".
In any case, you would need to bring up all the fics tagged with "John Smith" and "Smith" and then filter by fandom. In total, that's four different tabs, as compared to two using metatags (okay, it would be three tabs if "John Smith" were synned to either "John" or "Smith"). But I think another serious issue here is that by offering a disambiguation option, we are making the user think that they'll get all fics with John Smith in Fandom X if they click on that tag. That is very far from being the case. (In contrast, if you click a link on a disambiguation page in Wikipedia, you are reaching all the information about the disambiguated subject, because there is only one central page for it.)
I assume advanced search will eventually let you search for (("John" OR "Smith") AND "Fandom X"), so that might solve the searching issue, but people are still going to click on tags.
9) It would make tags like "John" and "John Smith" immediately useful and filterable. Even if we don't have metatags yet, we can still make "John" a canonical tag, which would let users filter on it, and then connect it up with the various Johns when we get metatags.
Potential objections to using meta-tagging instead of ambiguity tags:
1) What would we do with the disambiguated tags like "John Smith (Fandom X)"?
We would connect them to the canonical meta-tag "John Smith" (or let the rake kill them if they're unused). There are reasons why a user might want to use a disambiguated tag. Also, we would transfer pairing associations and synonyms to "John Smith" if they're non-fandomspecific (like "John Smith/Jane Doe") or let them stay with the disambiguated tag if they're also disambiguated.
2) Won't that be a lot of work?
Maybe? But I think it would be more work in the long run to keep disambiguating. Also, I think we're hoping it'll get easier to rename canonicals soon, so that we won't have to unhook and rehook all the associated tags. Besides, if we got ambiguity tags instead, it would also be a lot of to hook them up.
3) What if there's an ambiguous character within a fandom? Then we wouldn't be able to filter by fandom to differentiate them.
The users in that fandom probably already have an established way to differentiate between them. Say, "John Smith Jr." and "John Smith Sr.". These could be made into canonicals and connected to the meta-tag "John Smith". Or they might want to use "John Smith" to refer to their main character, and "John Smith (18th episode)" to refer to the John Smith who only appears briefly in the 18th episode.
4) What if someone wrote a crossover featuring two different John Smiths from two different fandoms?
Then the user would instinctively disambiguate between them by means of fandom, and the disambiguated forms would become canonicals linked to the metatag "John Smith".
5) Wouldn't it be confusing for wranglers to share tags over fandoms?
Yes, maybe? OTOH, ambiguity tags would also be shared between fandoms.
6) I don't like having characters in different fandoms share the same tag.
We already do this for the book- and movieverse versions of the same character, if they're considered similar enough (and the same for other media), so it's not really something new. We also do this for historical and mythical characters who are shared between many fandoms. Having all characters with the same name share a tag would eliminate the judgement call as to which characters are similar enough to not need disambiguating.
Okay, that's it! Please point out the issues I didn't think about. : )
Anyway, I thought it was an interesting snapshot of AO3 and tag wrangling, so I'm posting it. I'd forgotten that bit about how much work it used to be to re-name a canonical tag which had lots of other tags attached to it! You had to manually unhook all of them and then re-hook them afterward.
Suggestion for using meta-tagging to handle ambiguity in character and pairing tags:
As I understand the current situation, we are disambiguating by fandom for characters (and pairings) that share the same full name. Later, we are planning to have ambiguity tags which lead to a disambiguation page where one can choose between the various disambiguated tags (kind of like the disambiguation page on Wikipedia).
I have a suggestion for instead using metatagging for the problem of ambiguous character names. I'll explain what I mean first, and then give arguments for it.
I think we are getting hung up on having to create a unique tag for each character. But that is not actually the goal here. The goal is to organize and connect the tags on the users' works in a way that makes it easy for other users to find the works that they want.
If "John Smith" is a character name for several different characters, I suggest we simply canonicalize it and attach it to all the fandoms where John Smith belongs. If a user wants to find a particular John Smith, they click on "John Smith" and then filter by fandom.
If a user tags their fic with just "John", then that tag will be an upper-level metatag connected to all the different Johns one level below: John Smith, John Sheppard, etc... Again, a user could click on "John", get all fics tagged by any of the various Johns, and then filter by fandom to find a particular John.
There's a place for disambiguated tags here, too, because they are sometimes useful for users. The disambiguated tag "John Smith (Fandom X)" would be a canonical that's connected to the (also canonical) metatag "John Smith" one level up. Note that this would require several levels of metatags, because "John Smith" is connected to "John" one more level up.
So how would the wrangler side look? Well, all pairing tags containing John Smith would be attached to "John Smith", no matter what fandom the other characters are in. Likewise, all pairing tags containing John would be attached to "John". Note that this would make it possible to collapse the cast field "John/Rodney, John, Rodney" into "John/Rodney", which is not possible today.
The tag "John/Rodney" would be synned to "Rodney McKay/John Sheppard", as today, as long as there were no other John/Rodney pairings. When someone wrote a fic tagged Rodney McKay/John Smith, then the tag "John/Rodney" would be made into a pairing metatag connected to both the tags "Rodney McKay/John Sheppard" and "Rodney McKay/John Smith" one level below.
What about synonyms? Well, they would be synned at the appropriate level. That is, the misspelling "John Smithh" would be synned to "John Smith", while "Captain John Smith" would be synned to "John Smith (Fandom X)" if such a tag exists and if Fandom X was the only fandom where John Smith was a captain. If another Captain John Smith appeared in another fandom, or if there is no disambiguated tag, then the tag "Captain John Smith" would be moved up one level and synned to "John Smith".
Okay, I think that's it.
Arguments for using meta-tagging instead of ambiguity tags:
1) It is more in line with how users tag their works. I don't have statistics on this, but I would guess that most users don't intuitively tag their work with "John Smith (Fandom X)" when they have already tagged for fandom X a few lines above. They think that it's enough to have "John Smith" or "John" when you already know what fandom the work is in.
2) It spares work for the wranglers. They wouldn't have to research whether or not a character name is unique or not, and then disambiguate it if it's not. Besides, it's a subjective decision whether a name is unique enough, and people make that decision quite differently. As the archive grows, every character name could potentially need to be disambiguated. That's a lot of work, and not actually necessary.
3) It makes for shorter canonical tags. (Shorter tags might also be good for performance, is that right?)
4) It saves work for the coders—they wouldn't need to write code for ambiguity tags and disambiguation pages, etc. Meta-tagging is something we want anyway.
5) It makes canonical tags more stable. For example, if I were to write RPF for John Sheppard the 16th century composer, then the tag "John Sheppard" would suddenly need to be an ambiguity tag, and the 1292 (and counting) fics with that tag would suddenly be stuck with a tag that's not canonical any more. "John Sheppard (SGA)" would have to be created, but it wouldn't solve that problem. If "John Sheppard" is a shared canonical between all fandoms from the start, there is no problem.
6) It eliminates the gray zone between metatags and ambiguity tags. In the wiki, "The Doctor" is used as an example of a tag which would be a metatag, linked to all the different characters in Doctor Who (and potentially other doctors, as well). But "The Doctor" could just as well be an ambiguity tag, which led to a list of the different doctors. How would you decide if it was to be a metatag or an ambiguity?
7) It makes sure that character names are not implicitly assumed to belong to a fandom just because there's no disambiguation. By contrast, in the current situation John Sheppard is automatically assumed by most people to be the SGA character, just because he's the most popular character with that name and there's no disambiguation.
8) It makes for better filtering. I don't think there's actually a way to click on a tag and bring up all fics tagged with the character you want in one step, either with disambiguation or with metatags. But metatags definitely do the job better.
So, say you're looking for John Smith in fandom X. With metatags, you click on "John". Then you get everything tagged "John", "John Smith", and "John Smith (Fandom X)". A drawback is that you also get fic tagged with "John Sheppard", but you can always filter for Fandom X afterward. Also, you don't get the fic that's tagged with just "Smith". So you also bring up the tag "Smith", then filter for fandom, and you're done. That's two different tabs.
In the scenario with ambiguity tags, you can start at the tag "John", and go to the disambiguated tag "John Smith (Fandom X)". But then you only get the fics tagged with that specific tag, so you also need to bring up all fics tagged with just "John" (which I understand would be an option on the disambiguation page), and filter for fandom. But you're not done, because there's also the tag "John Smith", which I can think of three different options for here. It's either
- unwrangled, which would be unfortunate since it's probably the most popular tag for the character (it can't be canonical or synned to a canonical, since it's ambiguous).
- synned to either of the ambiguity tags "John" or "Smith" (how would you choose which one, though?)
- an ambiguity tag at an intermediate level, so that from "John" you could disambiguate to "John Smith" and from there to "John Smith (Fandom X)".
In any case, you would need to bring up all the fics tagged with "John Smith" and "Smith" and then filter by fandom. In total, that's four different tabs, as compared to two using metatags (okay, it would be three tabs if "John Smith" were synned to either "John" or "Smith"). But I think another serious issue here is that by offering a disambiguation option, we are making the user think that they'll get all fics with John Smith in Fandom X if they click on that tag. That is very far from being the case. (In contrast, if you click a link on a disambiguation page in Wikipedia, you are reaching all the information about the disambiguated subject, because there is only one central page for it.)
I assume advanced search will eventually let you search for (("John" OR "Smith") AND "Fandom X"), so that might solve the searching issue, but people are still going to click on tags.
9) It would make tags like "John" and "John Smith" immediately useful and filterable. Even if we don't have metatags yet, we can still make "John" a canonical tag, which would let users filter on it, and then connect it up with the various Johns when we get metatags.
Potential objections to using meta-tagging instead of ambiguity tags:
1) What would we do with the disambiguated tags like "John Smith (Fandom X)"?
We would connect them to the canonical meta-tag "John Smith" (or let the rake kill them if they're unused). There are reasons why a user might want to use a disambiguated tag. Also, we would transfer pairing associations and synonyms to "John Smith" if they're non-fandomspecific (like "John Smith/Jane Doe") or let them stay with the disambiguated tag if they're also disambiguated.
2) Won't that be a lot of work?
Maybe? But I think it would be more work in the long run to keep disambiguating. Also, I think we're hoping it'll get easier to rename canonicals soon, so that we won't have to unhook and rehook all the associated tags. Besides, if we got ambiguity tags instead, it would also be a lot of to hook them up.
3) What if there's an ambiguous character within a fandom? Then we wouldn't be able to filter by fandom to differentiate them.
The users in that fandom probably already have an established way to differentiate between them. Say, "John Smith Jr." and "John Smith Sr.". These could be made into canonicals and connected to the meta-tag "John Smith". Or they might want to use "John Smith" to refer to their main character, and "John Smith (18th episode)" to refer to the John Smith who only appears briefly in the 18th episode.
4) What if someone wrote a crossover featuring two different John Smiths from two different fandoms?
Then the user would instinctively disambiguate between them by means of fandom, and the disambiguated forms would become canonicals linked to the metatag "John Smith".
5) Wouldn't it be confusing for wranglers to share tags over fandoms?
Yes, maybe? OTOH, ambiguity tags would also be shared between fandoms.
6) I don't like having characters in different fandoms share the same tag.
We already do this for the book- and movieverse versions of the same character, if they're considered similar enough (and the same for other media), so it's not really something new. We also do this for historical and mythical characters who are shared between many fandoms. Having all characters with the same name share a tag would eliminate the judgement call as to which characters are similar enough to not need disambiguating.
Okay, that's it! Please point out the issues I didn't think about. : )
