The Problem with Prolific Uploaders

kuuderes_shadow said:

Am I the only one who doesn't want to see their name prominently displayed on an image that they actively dislike (or worse) just because noone else could be bothered to tag it up properly?

I'm with you on this one. Not everyone who takes part in tag gardening is doing it for glory. I just like for things to be neatly organized, even posts that I'd prefer not to be associated with. The third solution presented doesn't provide much of an incentive for contributing unless you're the sort of person who just wants your name in the credits, and it may even encourage deleterious behavior (like loading up posts with irrelevant tags) by individuals less interested in improving search than in stroking their egos.

Wow, this a pretty interesting topic, and I can see there are a lot of things that are related. I'll try to give my two cents so first things first, my own motivation for uploading:

albert said:
...is to share good art I find and to contribute to a database where I can easily search for it in one place. This means I don't care who gets credit for the upload.

Pretty much the same, with just a bit of:

Mikaeri said:
...have their name on the best content.

Now, regarding prolific uploaders, im with:

BrokenEagle98 said:
I'm happy enough if others upload more of my favorite artists, copyrights and characters...curating art can be hard work.

But I get the idea.

IMO there is no instant solution. The fact that there are more things related to this issue, like mintagging, padding tags, upload cap, and so, is a signal that the current overall system is no longer working like it should. Is not possible to suggest something that will solve this problem without opening the doors to other problems in other areas. That's why a broader solution should be planed. However I agree with some suggestions mentioned before that I too think will help and that can represent the beginning of this broader solution that is needed.
Is to be expected that not everyone will agree, and that's ok. The goal is to have something that works better and that appeals to the majority of users, especially to those with higher levels (lets face it).

Mysterio006 said:
...uploaders not considering the quality of the content and going straight to upload to make sure they get it under their name. But that can re-enter the moderation queue, so that's not too bad.

Sorry if I dont know the statistic but I get the feeling that most of the times the content from a Builder never gets revised unless submited to. If someone sees a "bad" upload from a Builder, automatically go to the "trust" policy thinking "If a Builder uploaded it, it must be good for Danbooru" when there might be cases when it's not. The fact that they can re-enter the moderation queue sounds like more job for Mods.

wuv u said:
If people can’t find anything to post, then I think that it’s their problem. There are thousands of artists waiting to be discovered or cleaned. If people can’t take a few minutes to find them, then it’s shouldn’t be everyone else’s problem.

Agree. The internet is vast ocean of content, with a lot of treasures in it. Problem is that everyone seems to be looking in the same spot and that has nothing to do with danbooru. Plus, there are tons of old series/artist with very good art that everyone seem to forget.

I dont think its about throwing everything to Danbooru, that's not being a Builder. Uploading just for the sake of it and to get a number rising is like carrying a bunch of bricks, rocks and crap, toss it in a spot hoping somebody else will sort the things and build a wall, and then say you are a Builder. In other words: users that upload the most not always do so in terms of quality. For Danbooru purposes I don't think that's contribution.

Real contribution is about search, bring the best and tag it rigth. Real contribution is to help with the copyrights request, artist request, translation request, Tag gardening and so on. But somehow we ended up giving more importance to uploads.

People want credit, but they don't want responsabilities or punishment. IMO, everything that's encouraged should be equaly punished on fault. We already have some of this covered but I think we can go further.

Now, into suggestions and agreements:
- Highlight the tagger, this task is equaly important and is actually what makes Danbooru different from other sites, among other things.
- Padding tags should be treated like tag vandalism.
- Hide the uploader temporaly so no one gets the instant credit. Make it a day.
- Lock the tags for the uploader a day or two. This way taggers can do their job and mintaggers get easily spoted. Punish accordingly. Pay attention to the on-upload tags.
- Good tagging on-upload should be encouraged. This also means punishment when not done right.
- Limited reservation slots sounds good for those post you really really really want your name next to it. Same conditions for on-upload-tags apply.
- Power votes on post where uploader=tagger could be a sign of a good quality post and a good contributor.
- Votes have value, lets use it where apply. Could be used to affect the upload cap, grant more reservation slots, spot good contributors, and so.

I hope Im not spining wheels here.
Sorry if too long and excuse my french.

If we really want to highlight the tagger, then one should take this into consideration:
-Uploading revision
-Uploading Pixiv version of Twitter posts
-Better scan of already existing scan
And probably some more situations.

Should we really want to highlight the user as the "Tagger" when the user of the inferior post did most, if not all tagging work? The child post gets forgotten more easily, so if we are doing this for "credits" then we should also think about these situations.

What are "padding" tags? Something that is not notable to some user but the others it is? Then the users would get punished who notice more tags and the other party simply can't see it. Punishing a tagger for this is a horrible idea. This will lead inevitably to tagging wars if punishing gets enforced.

Provence said:

Get a room together, please.

Anyway, what about making everything nameless?
Be it tagging or uploading: The idea is always that something is degenerating into competition. More and more tags are getting added where we go into really insignificant details or tagging wars because user keep disagreeing about something (tags have blurry borders at borderline cases, meaning their definitions aren't fix despite a wiki and people will be argumenting in their favor of course).
And uploading competition leads to sniping.

So...make everything nameless?

Taggers should be named. Uploading should be nameless. That's my opinion through and through. I will elaborate a bit more (once again), but I've read all of the current discussion and I've seen a number of other ideas I've already thought of (and perhaps why they'd fail, as maybe others would've pointed out).

Since I'm sorta done gaming my heart out with some of the stuff I bought during the steam sale now, I have the time to write a more elaborate response to forum #141371. I'll probably be doing another writeup in markdown regarding this subject, but this is something I feel deeply about, and I think it's contrary to the goal of the site to just allow the users that do this sort of thing rampantly to continue doing it, edging out other potential new uploaders to the scene given how discouraging it is to them.

So, let me just say this for consideration first. It is already extremely difficult to attract new uploaders. There are plenty of people who browse art not only on the mainstream sites but also all the other little niche sources too -- they are capable (and perhaps even better at us) than finding that sort of content. Edging them out, or making it hard for them to break in and contribute is something that is already a constant in the current scene. Whether we like it or not, that's just plain fact as most of us aren't nearly that altruistic in the first place.

Then, I suppose the question becomes less about uploads (and upload habits) and definitely more about the current method of accreditation, and what responsibilities are attributed to users. How do we make it such that contribution attracts less of a "I want my name on this" to "I want everyone to be able to find this work"? Because that's how it should be, and that's how everyone starts. Nobody (or at least, nobody we want) goes in at first thinking the former thought.

But again, watch for what I'll write soon. I have a lot more to say.

Unbreakable said:

Well it doesn't really bother me since no one is forcing you tag the post in question, but I can understand what you mean.

There are dozens of such posts that would get my name on immediately, though, and in many cases forever. The only way to get rid of it would be to delete my account and start again.

And do we really want to encourage people to just leave stuff chronically undertagged, or to not do mass tagging updates, in order to avoid having their name accidentally get on images that they don't want them on?

I'll put in my two cents and say that I oppose the removal of usernames from uploads. It allows me to see a user's uploading habits and see what artist's work they're uploading so I can upload work from that same artist as well.

As far as my uploading habits go, I usually just do it when I find content that the lightning sweepers "miss" or something unusual I don't see often on danbooru. It's not high-volume but I think it's a decent contribution. At times I can see how it can make users feel "powerless" or unable to gain recognition though.

Now, a little bias on my part, tagging and gardening is something I'd like to see recognized more. Partially because it's the way I enjoy contributing to the site. Also since because I do that, I see how much work there is to be done in that area and a little influx of users into that area might be helpful.

Regarding Prolific uploaders, I don't think those kind of changes are going to discourage them. In any social environment there's going to be competition and even if the uploader isn't displayed there's going to be that minority that does it just to see the number on their profile go up and compare to others to see who is "better".

I don't really interact with the uploader sections of the site enough to confidently suggest an answer to the validity of the options.

(I apologize if this thread was meant for builders+ only.)

Elfenus said:

- Lock the tags for the uploader a day or two. This way taggers can do their job and mintaggers get easily spoted. Punish accordingly. Pay attention to the on-upload tags.

This would be disastrous for people like me that constantly find new tags to add even after having uploaded a post with 50 gentags. It'd also mean not being able to fix typoes or autocomplete mistakes.

nonamethanks said:

This would be disastrous for people like me that constantly find new tags to add even after having uploaded a post with 50 gentags. It'd also mean not being able to fix typoes or autocomplete mistakes.

I'm with nonamethanks on this one since I have this problem too, small things that you oversee or common tags you forget like looking at viewer or copyright/artist tag.

Generally: Enforcing punishment is a horrible idea. You shouldn't get exposed or punished because of small things that are easy to fix.
That way, we will lose even more users since constant fear isn't a good environment.
That is not only with locking the tags to expose "mintagging" but also "padding tags" because who decides what a "padding tag" is? If it's only between two users, like most tagging wars, then the one removing the tag shouldn't be doing the "punishment" and call it "vandalism".

Tagging on upload isn't always good. There are many everyone images and animated ugoira/gifs and to pretag them is pretty horrifying. Also, and especially for older posts, it's much more pleasant to tag the post after the upload.

tapnek said:

I'll put in my two cents and say that I oppose the removal of usernames from uploads. It allows me to see a user's uploading habits and see what artist's work they're uploading so I can upload work from that same artist as well.

I use the tag search on users of good taste similarly to favorites quite often. One of the key features of this site is the many different ways you can do tag searches. Crippling that feature is undoubtedly going to be a net negative for the experience of the site.

EB said:

I use the tag search on users of good taste similarly to favorites quite often. One of the key features of this site is the many different ways you can do tag searches. Crippling that feature is undoubtedly going to be a net negative for the experience of the site.

Speaking of uploaders with taste or better said the lack of taste:
How should quality checking on users be done then? There are some users that have not the best upload quality and while images should be judged on their own merits: If you can't see the users name you also can't evaluate how this user is performing and that also means you can't complain about that.
It will be a free ride for anyone to upload even more crap then.

I think one way of alleviating the tag padding problem is to weight the value of tags. A tag with hundreds of thousands of posts isn't as valuable as one with a few dozen. Then the usefulness of things like color tags decreases over time, while artist tags will tend to be valuable. The only problem is this turns something simple (count the number of tags added) into an abstract formula. Honestly, maybe this is a premature concern and it will turn out not to be an issue in practice.

A similar sort of logic could be applied to displaying the uploader name. If the uploader only has a few recent uploads, then show them. If they have thousands, then well, the value of knowing the uploader isn't as useful. You could figure as much from probability.

Honestly I find some tags with smaller tag counts to be useful in a way too, getting to find posts with rarer concepts, and they also don't take as much time to fully browse.

albert said:

I think one way of alleviating the tag padding problem is to weight the value of tags. A tag with hundreds of thousands of posts isn't as valuable as one with a few dozen. Then the usefulness of things like color tags decreases over time, while artist tags will tend to be valuable. The only problem is this turns something simple (count the number of tags added) into an abstract formula. Honestly, maybe this is a premature concern and it will turn out not to be an issue in practice.

A similar sort of logic could be applied to displaying the uploader name. If the uploader only has a few recent uploads, then show them. If they have thousands, then well, the value of knowing the uploader isn't as useful. You could figure as much from probability.

That is interesting since tags like 1girl, solo and the hair/eye color tags are something that I consider most important since these help you the most to identify unknown characters from copyrights you aren't that aware of. I exclusivle work with these tags and maybe some other very specific item that is associated with a character. I wouldn't make a hierarchy of tags here
But I think that's going off topic here.

Ok, I have just written a mail to Albert but I also want to share some of the keypoints relevant to this topic.
I think we can agree that Danbooru is currently stagnating on image content The growth in uploads in the last few years is there but it also isn't all that relevant.
Anyway, I want to stress strongly that the problem is not the uploader name. This is not the case.
The far bigger problem is that there is a score next t the posts, because if you are saying that uploading is for the credits then the score next to the post plays a major role in it. I don't really care for the uploader's name but I look at the score and sometimes I get this feeling of jealousy because I see that another post has score 14 while mine post has score 12 despite being posted later.
The score that is blatantly visible inflicts uneccessary conflict between users as it says that the posts of user A are better than the ones of user B and then user A and B are getting into an argument. This has nothing to do with the uploader's name but the score is creating this sense of evilish competition between users.
If we cut this out instead of the user name then the user who brought the image to Danbooru still has their name on it and feels rewarded by this. They can tag it and everything is nice and cool.
Stuff like the popular search went hidden probably because of that. Order:rank can still exist under "Hot", though. Hiding the score doesn't mean we can't search anymore for posts that are "loved". Order:Rank can still exist with that but we don't get into this "If I upvote post a and downvote post b then post a will be better than post b" thinking. Or probably not t this extent.

It also help when users are getting "1upped". That means when another is uploading a Pixiv version of a Twitter version (Parent-Child relationship). The Pixiv post is generally better and the uploader is taking all the glory from the Twitter post and the tags. That means the user who uploaded the Pixiv version gets more spotlight but the user uploading the Twitter version did all the tagging. If we would erase the score then you can't say that easily that the other post gets more spotlight. One might also think if a parent-child relationship is really this good then since it basically means that the Parent is better but that's not always the case.

I think the other big box that keeps new users away is the upload limit. Danbooru has always been high quality, even with the old system in place and with some old Approvers that are now demoted.
I think that the current moderating system has one big flaw and that is that one approval doesn't generate a new upload slot instantly but only after 23 hours.
This is especially frustrating in the first 24(!) days (You have to subtract the first week where nothing is possible, that will be important later, too).
With the current system you can't fall below 10 uploads per day. Now imagine two users: C and D.
User C's uploads are getting approved instantly everyday in the first 24 days and he uses all 10 slots everytime.
User D is also using all 10 uploads each day but nearly all of their uploads are getting deleted.
Both users will have the same amount of uploads after 24 days with the curret system.
If we keep everything the same but change that every approved upload returns one upload slot back then user C is rewarded for their good uploads while user D still lurks around with 10 uploads each day.
This is only the first time but also regarding users with 50 upload cap: While they probably won't ever reach that amount of uploads that day, they still see that their approved uploads are generating one free slot again. I think that this will be a motivation boost for most users and it will speed up the time a user needs to get promoted.

The last thing is probably minor, but I said before that the first week of registration is basically pretty shitty.
If you haven't signed up for Danbooru and click on "Upload" then get redirected to the "Sign up" page. Then you sign up but if click now "Upload" then you will get slapped into your face since this is the first time you get to know that you can't upload during the first week.
This is probably pretty frustrating. I'd like to suggest that a warning is displayed before signing up. Something that tells you that you can't upload during the first week. If you are told this after registration then you feel dumbfounded, if you are told this before registration then you know what to expect.
(And don't come me with "it's in the wiki". New users aren't aware of that.

Anyway, these are some ideas that I think could improve the whole situation, but I don't think hiding the uploader's name will achieve anything. I guess it's the only thing right now that helps Danbooru to stagnate at least.
I mean, in case of Touhou people aren't said that the copyright belongs to one user. They say that one user only takes the best (-> score) posts.

Can we try to pinpoint the exact problem here? Is it people getting angry simply about losing the upload competition to someone else, and not getting their username (and all associated statistics) next to it? Or is it that people are angry about wasting their time tagging a post that someone else has already uploaded by the time they did, possibly with minimal tags? Or is it something else after all?

Also, here's a crazy idea. I was going to say that we absolutely need uploader names for statistics. Then I kinda realized that we only need those statistics when looking up users for promotions to unrestricted uploads. So how about creating an artificial, hivemind-like user, and attributing all uploads from users with unrestricted uploads to that user? Much like all formerly-anonymous changed were attributed to user named Anonymous during transition to D2. In a way, you either give up your unrestricted uploads, or become a part of a faceless uploader cartel. Naturally, it should still be possible to look up the original uploader by looking at the tag history (and it SHOULD be preserved to avoid abuse of anonymity), but that would strip people of bragging rights without much change in the code.

Type-kun said:

Or is it that people are angry about wasting their time tagging a post that someone else has already uploaded by the time they did, possibly with minimal tags?

This, mostly. I've mentioned it before. Also seeing newly-uploaded images that should have been filtered by my blacklist but aren't. There are some things I really don't care to see (and not necessarily the things you'd expect) that often pop up on the first page of /posts for ten minutes or longer while the uploader defers tagging to rush more posts.

I guess it really is mostly that people are afraid of not getting a high scoring post. The problem with Touhou for example is that apparantly one user is taking all good posts instantly and leaves the rather mediocre stuff in the dust for others. If we take away the scores then we don't have the seperation between good, mediocre and low ranking posts.

The sniping problem is really only happening during rush to this extent because there are many people competing for stuff. The tagging is lost, yes, but you can still copy your tags to the other upload. But please with correct tags. There is nothing more frustrating than cleaning these tags up because someone else was too impatient to wait.

Long things short: I think the score next to a post is getting pretty toxic in the end since people get annoyed mostly when this happens to high scoring posts or posts they think will score high.

I'm far from a prolific uploader and my perspective on this is pretty distant, but I think we should be looking at ways to make it easy and worthwhile to contribute so that a broader base of users can be more effective. I'm coming at this from a private bittorrent tracker perspective where there's also a very strong sense of meritocracy through contribution. On the best trackers, there's always a healthy balance between requirement of user initiative and varying opportunities to creatively contribute. Basically, almost every facet of the user's interaction with the site is measured, and any time a user does anything productive, it adds to an overall rank.

Establishing accessible and sequential milestones could go a long way to encourage participation. Right now, site perks are largely tied to donation, which I can understand (you gotta pay the bills), but it could be worthwhile to consider ways to show new uploaders that if they get started uploading and do a good job, rewards will come.

I don't see any healthy reason to anonymize the work that people are doing here (unless those people choose to be private). The resentment angle is totally wrong. People should always be given credit for their contributions, and top uploaders (and people who do other helpful things) should be seen as leaders who can help others improve the site similarly.

Weighting tags and rewarding tagging definitely sound good. There was concern about people adding pointless tags for credit, and also just going for the popular tags, but I think that could be alleviated by grading on a bell curve (by total tag count) and having the centre of the curve weigh the most. That way, the extreme highs (1girl) and extreme lows (card_on_necklace) will weigh less than tags that make up the middle and therefore hit the sweetspot of descriptive variety and useful organization.

If stagnation is really a concern, maybe it's worthwhile looking into expanding some social features to promote a sense of community for new people, even if it just means linking users with similar taste or making toplists of the users who have contributed the most in a variety of ways each day/week/month/year and then giving them a _temporary_ prize (more search tags, more uploads, +1 voting power -who knows what else- which might also encourage users to upgrade to keep those perks permanently or to keep contributing to maintain them temporarily). Prizes could even be separated by user class so that the builders don't automatically clean house.

Basically, I think that more roads to improvement and more opportunities for community engagement will help heal the resentment and negativity. Back to the private tracker analogy, when someone complains about the system it's always shot down with, "you can find another way," and that's because those opportunities do exist for those who make the effort, and the community is there to help.