Donmai

Changing touhou tag to full name

Posted under Tags

Mysterious_Uploader said:

I think "the software might fuck up" shouldn't be an argument on whether the change to the full name is correct or not.

Actually, no, that is a huge consideration. I don't think you are considering the gigantic amount of churn or database load that would generate in the post version records. That table is already several gigabytes and slow as hell.

In worst case scenario, i think Danbooru can rollback a BUR, but i'd like to ask for confirmation.

You can't just "rollback a BUR". That would essentially double the amount of post versions created which would bloat the table even more.

But, since you asked, nowadays we always add full names to copyright tags and alias nicknames/shortened versions, so i think it should be done for the touhou tag too.

This might have been doable several years ago, but the boat has already passed the point of correcting it the way you want to go. It's one of those tags which is now too big to change.

I gave this BUR a -1, but would support the reverse alias if needed, i.e. "touhou_project -> touhou".

As a very general statement not restricted to this website or topic, the observation that "we generally, but not always, do something in one specific way" brings with it the necessity of understanding why that course of action is taken if one wants to use it as a guiding principle. It isn't a course of action which is always taken, so I expect everyone will agree that either:
1. it should always be taken
2. there is more to consider than just yes/no

In this specific case, this is relevant to observation that danbooru generally, but not always, uses the full, official names of copyrights. To simplify what I've said so far, pointing to danbooru's history of usually doing something a certain way is not a valid argument. If it truly should be done that way in this case, you'll be able to explain why.

Danbooru has autocomplete, so people who start to type in "touhou_project" will see "touhou" pop up. No human will be confused about the lack of an exact string "touhou_project" when the tag "touhou" exists. Therefore, concerns of consistency in this case are more relevant to machine interactions. Other image databases are the most important other sources to be consistent with for two reasons.
1. Danbooru itself is an image database, so it has considerable userbase overlap with these other sites. Users writing programs which make API calls to danbooru may well be making calls to several other databases as well in the same program.
2. There are numerous databases which mirror danbooru across the web. It is best for the data received from danbooru to be consistent with posts of other origins (mirrored from other sites or uploaded directly by users). These sites have greatly varying levels of maintenance (and presumably hardware capabilities too). If the BUR goes through, some of them may choose to follow danbooru and convert the tag on their own site. However, it is also inevitable that some of them will not and others will not be able to.

A BUR on danbooru will keep the tag consistent on danbooru itself, but won't be applied to existing posts on other sites. With this in mind, I looked at other image database websites that I knew of which deal with similar content to danbooru and checked how they each tag the series. The following sites tag the series as "touhou":

  • gelbooru.com
  • zerochan.net
  • yande.re
  • konachan.com
  • rule34.xxx
  • rule34.paheal.net
  • e621.net
  • e-shuushuu.net
  • booru.allthefallen.moe
  • drawfriends.booru.org
  • booru.vineshroom.net

It's worth noting that yande.re, booru.allthefallen.moe, and e621.net already have touhou_project aliased to touhou. Doing the opposite of the suggested BUR (as BrokenEagle also mentioned) could be a decent idea.

The following sites use both the tag "touhou" and "touhou_project" on different posts with at least 10 posts under each tag. Note that this is exactly what you don't want in image databases like these:

  • chan.sankakucomplex.com
  • xbooru.com

The following sites exclusively use the tag "touhou_project". This is the list of sites danbooru would be in convention with if this BUR were approved:
 

And that's all of the sites that I checked that had at least 50 posts between the two tags combined. It should include all of the most major sites already, although if you notice that I've overlooked a major site please let me know. I've neglected sites like cure.ninja/booru which purely mirror multiple other databases, though such a change would affect them as well.

It's true that other copyrights have been renamed in BURs before, but other series are not the single most popular (by post count) copyright for fanart of all time. Touhou is. Pointing to the fact that a copyright with 5,000 posts has been renamed before and expecting that to serve as proof of a precedent is, of course, insufficient.

nonamethanks said (in regards to "touhou_project"):

It's the official name, it's also how any site refers to it.

See above.

nonamethanks said:

There's as much merit in this as any other alias to the main name.

For the sake of time, let's pretend for now that this is true. However, benefits are weighed against drawbacks. That the amount of benefit is the same as in other cases is only half of the story.

nonamethanks said:

We don't use "fgo" or "kancolle" as main tags.

FGO is an acronym. KanColle is a contraction. Strictly speaking, these are not relevant. As DanielCM pointed out, abbreviations are still in use.

There are several more reasons not to go through with this BUR, but because the currently-standing arguments in favor don't even compare to the ones against, I'll stop for now.
Because one of the people arguing in favor of this BUR has expressed in danbooru's official discord (but not in this thread) that one of their main reasons for doing so is the entertainment value of watching people argue, one could reasonably expect that this discussion may not end even when an official decision is made. In light of this, I think this thread should be locked once an admin makes a final decision one way or the other.

Zurreak said:
Danbooru has autocomplete, so people who start to type in "touhou_project" will see "touhou" pop up. No human will be confused about the lack of an exact string "touhou_project" when the tag "touhou" exists.

[...]

FGO is an acronym. KanColle is a contraction. Strictly speaking, these are not relevant. As DanielCM pointed out, abbreviations are still in use.

In the interest of the argument, we still don't tag, say, jojo_no_kimyou_na_bouken as jojo or jojo's, even if autocomplete would pop up the full tag without an alias, and even if everyone just calls the series JoJo's or JoJo, and that's neither a contraction nor an acronym either. Functionally and generally I see little difference there with the current touhou discussion.

That said, I think BrokenEagle98's comment regarding the weight on the database renders most discussion moot anyway, since the technical consequences of such a change would probably be too negative to really offset the positives.

BrokenEagle98 said:

Actually, no, that is a huge consideration. I don't think you are considering the gigantic amount of churn or database load that would generate in the post version records. That table is already several gigabytes and slow as hell.

You can't just "rollback a BUR". That would essentially double the amount of post versions created which would bloat the table even more.

This might have been doable several years ago, but the boat has already passed the point of correcting it the way you want to go. It's one of those tags which is now too big to change.

The post versions table has 25M entries, this bulk request would add around 2% to the size. I don't see how that could slow it down any more than it already is.

Just by comparison, in the last couple of weeks I have populated the photoshop (medium) tag, creating around 500k versions. The database is not noticeably slower because of it. Discarding a BUR just because of technical concerns is not valid in this case.

Updated

Shinjidude said:

You often see "_______ Project" or "Project _______" attached to franchises, simply as an LLC to encapsulate the copyright legally as a separate entity from whatever other companies or inividuals might be involved in its creation or development.

For example if you go to the main Idolmaster site, you see at the bottom that it's officially "Project Im@s" and "Project Cinderella" legally speaking, yet no one would call them that.

Everything attached to Evangelion gets attached to / labeled with "Project Eva".

I think for similar reason, it's fine to leave the "Project" off "Touhou Project" as being similarly redundant. It's also such a large and established tag, I think it'd be smarter to just leave it alone.

The difference here is that "Project Cinderella" and "Project Eva" can be considered as side/uncommon names, while the only name for Touhou is "Touhou Project".

In this specific case, this is relevant to observation that danbooru generally, but not always, uses the full, official names of copyrights. To simplify what I've said so far, pointing to danbooru's history of usually doing something a certain way is not a valid argument. If it truly should be done that way in this case, you'll be able to explain why.

Danbooru has autocomplete, so people who start to type in "touhou_project" will see "touhou" pop up. No human will be confused about the lack of an exact string "touhou_project" when the tag "touhou" exists.

"Touhou Project" would be a more correct tag.

nonamethanks said (in regards to "touhou_project"):

It's the official name, it's also how any site refers to it.

Zurreak said:

See above.

Other imageboards usually follow Danbooru. You shouldn't use those as example.

To strengthen my first point: the english Wikipedia article calls the series Touhou Project, while idolmaster is called The Idolmaster, and NGE is called Neon Genesis Evangelion (franchise).
The english Touhou Wiki calls it Touhou Project too, and the biggest Discord server of the series is called Touhou Project.
If we want to give the full picture, there have been instances of ZUN calling the series just "Touhou" for convenience, but again, since we prefer Xenoblade_(series) over Xenoblade_chronicles_(series), i feel like we should prefer Touhou Project over Touhou, since that's the original name and not just a formality/legal thing.

Regarding your worry on other image boards: a Gelbooru admin in the Danbooru Discord Server has said he's already looking into changing the tags from Touhou to Touhou Project, and as nnt said, it's not particularly heavy on the servers.
I don't think we should be reliant on other sites, they can also alias touhou_project to touhou if they don't want to change the tag, because as you said:

No human will be confused about the lack of an exact string "touhou_project" when the tag "touhou" exists.

Touhou Project is a more correct tag, and I'm mainly interested in Danbooru being correct.
Sites that aren't well-maintained are probably on the brink of death/don't have much of an userbase, and have a lot other tags that are double/triple. Changing Touhou to Touhou Project wouldn't change much on those sites.

Delaying this matter would only mean that it would get even more deep-rooted. We have the chance, so we may as well do it.

EDIT: Adding another argument:
If we can consider that an image board, Exhentai and it's "sub-sites" (Nhentai, hitomi, etc.) use "Touhou Project" as tag.

Updated

Mysterious_Uploader said:

Zurreak said:

In this specific case, this is relevant to observation that danbooru generally, but not always, uses the full, official names of copyrights. To simplify what I've said so far, pointing to danbooru's history of usually doing something a certain way is not a valid argument. If it truly should be done that way in this case, you'll be able to explain why.

Danbooru has autocomplete, so people who start to type in "touhou_project" will see "touhou" pop up. No human will be confused about the lack of an exact string "touhou_project" when the tag "touhou" exists.

"Touhou Project" would be a more correct tag.

You had already stated this before I made my previous post. It also is not at all a response to what I said. My point still stands.

Mysterious_Uploader said:

Other imageboards usually follow Danbooru. You shouldn't use those as example.

How does that follow? Of course they still matter and they arguably matter more than any other type of site. You could argue that this isn't the case, but you'd need to provide actual reasoning.

Mysterious_Uploader said:

I don't think we should be reliant on other sites, they can also alias touhou_project to touhou if they don't want to change the tag

Whether they can do an alias or not depends on the website in question. Not everyone's site is designed the same way and not everyone has that functionality. Of course, danbooru as well can make the alias touhou_project -> touhou.

Mysterious_Uploader said:

Touhou Project is a more correct tag, and I'm mainly interested in Danbooru being correct.

Leaving your interests aside, touhou_project is a more official tag, not a more correct one. Don't discard the distinction. On that topic, touhou_hisoutensoku. You mentioned that that's the only proper copyright name of any of the touhou individual games because it's the only one in Japanese. However, by your own reasoning this is incorrect. The proper tag would have to be Touhou_Hisoutensoku_-_Choudokyuu_Ginyoru_no_Nazo_wo_Oe. Incidentally, there's another touhou game tag on danbooru which is also in Japanese, Touhou Gouyoku Ibun which would have to become Touhou_Gouyoku_Ibun_-_Suibotsushita_Chinshuu_Jigoku. Of course these tags aren't preferable over the current ones. However, they are more official and the current tags are only part of the games' real names.

Mysterious_Uploader said:

Sites that aren't well-maintained are probably on the brink of death/don't have much of an userbase, and have a lot other tags that are double/triple.

Sankaku Complex is very badly maintained and has many double/triple/quadruple tags. They're very, very much alive. I don't have exact numbers, but their amount of user activity may even be higher than danbooru's.

Mysterious_Uploader said:

Changing Touhou to Touhou Project wouldn't change much on those sites.

Nor would it change much on this site. It's not about whether it would change much, it's about whether it would do more good than harm (and in the case of danbooru, any good at all). On those other (badly maintained) sites, it would do only harm.

Mysterious_Uploader said:

Delaying this matter would only mean that it would get even more deep-rooted. We have the chance, so we may as well do it.

"May as well do it" dismisses the fact that there is any reason not to do it. By that token, we could also say "We may as well not do it". In fact, "may as well not do it" would be much more accurate, seeing as how the only valid argument in favor of it which still stands is that some people think "touhou project" sounds cooler (Source 1, Source 2). While this is true, it is also the case that some people prefer just "touhou".
I'm not forgetting about the observation that touhou_project is a more official name. However, as I explained before, this observation is not, on its own, an argument in favor of going through with this BUR.

Updated

No:

  • Too much churn for too little reason.
  • Previously rejected in topic #12488.
  • Breaks links in many places, including comments, forums, wiki pages, and offsite. I'm not fixing 800 wiki pages by hand for a cosmetic change.
  • Breaks compatibility with other boorus.
  • Requires fixing 100+ other Touhou tags at the same time, plus everything that links to them.
  • Very unpopular given the number of downvotes.

We have this attitude sometimes that "technically correct is the best kind of correct". That's not something I agree with. The "use full names" principle just means "don't use first names only for characters" not "be as pedantic as possible about naming".

1 2