Question: Uploading with minimal tags.

I also think that a short test wouldn't hurt, as it will clear out doubts about the bot's efficiency and functions:

-If it works, we can always use it as a support. I am still unsure whether using it to replace fast uploaders would work (I think it wouldn't, and would have more negative impact than anything).
-If it doesn't, we can close the discussion about the bot, for now, and try to look for another solution, and perhaps one that stirs less controversy.

Personally, I think motivating users to upload/tag better would be more effective through a positive approach (encouraging them through rewards, even those with little meaning. Something as simple as a 'Thank you' can have a great effect) rather than a negative one (driving them away from 'easy posts).

I'm suprised no one (except glasnost maybe) considered the fact, that if DakuTree was able to create such bot then unless he's some kind of computer programming superstar there are others who can do exactly the same? In fact we're not sure if this hasn't been done already. We are only sure of DakuTree because he did tell us about it. But blaming him for the following shitstorm... looks like typical shooting the messenger stance for me.

Anyway, we can leave things as are now, but I'm afraid that will produce another rounds of speed posting drama in the future, and this time extra spiced with accusations of using unauthorized bots. So maybe, just maybe, we should try to create a special user-bot account with list of "excellent, danbooru quality pixiv artists" maintained by community (under the enlightened leadership of our mod team) - making such pixiv artwork clearly off limits for every uploader and thus limiting flow of bad blood in the future.

Before this is locked, I'd like to point out the woefully inadequate discussion on the implications of an escalating bot-war. Would the community be happy with that? And I must wonder, would DakuTree himself be happy with that? I consciously elect to leave this to the rest of you to discuss. When speculating on the effects of the bot, please keep in mind that DakuTree isn't the only person who can make one.

I'll address something else while I'm here.

The lapses in logic used to make the claim that a Pixiv camping bot will positively reform Danbooru are inexcusable to the point where I can only imagine it's an attempt at deceit.

• It will not encourage people to look for pictures elsewhere. When other people snipe during the Pixiv rush, "good uploaders" supposedly quit, but when DakuTree does it, they instead search the internet for new sources. Someone needs to make up his mind.
• "Good uploaders" aren't being victimized. The window in which an uploader can be sniped during the rush is narrow to the point where it's safe to assume the uploader is also camping. Both parties are aware of the game they're participating in; if you don't like it, don't play.
• There is no wasted effort. One's tags still get merged in even when sniped. What's wasted? A character tag was added twice? Does this imply that the fewer tags used in a snipe, the less waste there is? Regardless, how the other party spends his effort is his own business, and everyone camping is guilty. See the above points.
• Being first is important. With the above points considered, for what other reason would someone complain about sniping besides not being first? People should just be honest with themselves and admit they want their efforts rewarded.

Incentivising and rewarding desired behavior could ideally be better, but the proposed bot doesn't help.

I'm actually thinking it may be better to try out what Shinji suggested first. At least from what I've seen since I brought up the issue again, I've seen the odd person I don't usually see actually uploading from more "useful" sources, aswell as the odd user that I consistantly saw doing the whole minimal tagging nonsense not doing it. (There is still users doing this though..)
Would be nice to see how this actually pans out over the next few days.
This being said, I still think the something along the idea of what MyrMindservant said would probably be the best route to take. It could be tested or something to see how it goes.

The bot is kind of a "final option". As for the whole "catching images before they get lost" side of things, I might have something that can already do that better to a certain extent. Shall see what I tweak it a bit so anyone can use it.

Dbx said:
...
Incentivising and rewarding desired behavior could ideally be better, but the proposed bot doesn't help.

I've already rambled about this entire thing earlier in the thread. As for incentivising people, how exactly? You can't exactly use records for such a thing, as it's essentially a one time thing.

About the records being one-time thing, I think you can always ask another staff member to add another one for you. Whether it's positive or negative, having more people updating an user's record should also have a greater effect as well.

Dbx said:
The lapses in logic used to make the claim that a Pixiv camping bot will positively reform Danbooru are inexcusable to the point where I can only imagine it's an attempt at deceit.

I agree entirely with this post and glasnot's posts earlier. All of these issues DakuTree raises are either very minor, solved easily in other ways, or utter nonsense.

As for his bot itself, to everyone but him arguing about it in this thread it's merely a hypothetical. We have no substantial data about how it works in practice or if it'd even be distinguishable from the current status quo on our end. I strongly doubt it would be. It's kind of a non-issue until it becomes an issue sort of thing.

If he wants to semi-automate his camping of Pixiv so he can do it even easier that's his prerogative really. But I definitely don't see any reason for these proposed changes to the system just to appease his bad attitude.

Deelles said:
About the records being one-time thing, I think you can always ask another staff member to add another one for you. Whether it's positive or negative, having more people updating an user's record should also have a greater effect as well.

Not exactly what I meant. You can give the same user multiple records. The point I was making is more often than not you won't see multiple records for the same thing.
Say a user fixes x problem, makes x script, maybe does a huge amount of effort in editing the wiki or tagging, more often than not somebody will give them a positive record for it, and other users will hold off on doing so simply because they already got a record for it.
Same thing applies to looking for content. Even then you have to have someone actually notice you actually bringing said content.

Hinacle said:
All of these issues DakuTree raises are either very minor, solved easily in other ways, or utter nonsense.

As I've said before, yes this is easily solved in other ways, most problems are. Ways that have already been completely ignored.
I complained about the entire issue months ago and nothing was done about it. Any "easy" suggestion was essentially shot down as "there is better things to implement" and "this is a non-issue".

It was a non-issue then, and it's still a non-issue now. So why are you trying to make a mountain out of a molehill?

DakuTree said:

Not exactly what I meant. You can give the same user multiple records. The point I was making is more often than not you won't see multiple records for the same thing.
Say a user fixes x problem, makes x script, maybe does a huge amount of effort in editing the wiki or tagging, more often than not somebody will give them a positive record for it, and other users will hold off on doing so simply because they already got a record for it.
Same thing applies to looking for content. Even then you have to have someone actually notice you actually bringing said content.

Really? I know plenty of members who regularly gets rewarded for the same thing (above all else, translators), and I don't think there is a rule that forbids users to give more than one record for this, especially if the user continues to do so. Maybe there is some sort of mentality here that makes it awkward? I personally think that such a mentality is more harmful than anything.

DakuTree said:
As I've said before, yes this is easily solved in other ways, most problems are. Ways that have already been completely ignored.
I complained about the entire issue months ago and nothing was done about it. Any "easy" suggestion was essentially shot down as "there is better things to implement" and "this is a non-issue".

I'm going to give you the benefit of the doubt and assume "the entire issue" is people uploading with minimal tags and not this other silly stuff about people not using their time in the ways you desire.

You've given one person a record for this supposedly pervasive issue in the past year. A cursory glance at their last few months of posts shows an improvement after that. The one time you tried the normal solution to this problem it worked.

In the terrifying world where they don't improve after being told to you can just refer them to one of the active mods and they can try blocking them for a bit, which is what ultimately happened to Mr_GT and eventually solved that problem.

What more do you want exactly?

HNTI said:
@Shinjidude: So basically you want to leave things as they're right now ?

That's not exactly true. I do think there is an "issue", namely the fact that the "first" game causes bad behavior as a side-effect (uploading without tagging, leaving posts under-tagged, down-voting good posts to jockey for daily ranking position, etc). While I think the game is silly; it in itself isn't a problem.

If I had to pick a side, I think it's pretty clear from my previous posts that I'd lean towards DakuTree's camp. I actually like the idea of semi-automated or supervised automatic systems that enrich the site with no real ongoing effort on the part of the users. I really wouldn't care who the posts were credited to (the bot's creator, various people vetting the system, or no-one at all). That said, eradicating the game isn't a real benefit in itself, it's correcting the misbehavior that is what needs to be done.

Any solution that results in curtailing members' privileges further than they already are, or one that drives away large numbers of active users is not a true solution in my mind. Also solutions that require Albert make sweeping changes are not likely to be viable. For one thing, many changes have unforeseen detrimental side-effects (as we've seen in the past), and for another Albert is busy as it is and it's not reasonable for him to change things willy-nilly especially with major pending updates yet to be completed. Any augmentation or additonal features are much more likely to stick if they come from the outside and supplement the existing system (as happened with IQDB, DanbooruUp, the various UserScripts, and DakuTree's previous scripts).

On the "first" camp's side it doesn't really bother me that you like to compete, or that you like to boost your ego with increasing metrics associated with your account. I think it's silly to consider such a competition more important than the well-being of the site as a whole, but by itself it's not harmful. If somehow people would chose to follow ground rules, thoroughly tag posts, not maliciously down-vote; or even better yet if they decided to shift the "game" to boost and compete on other metrics that need boosting, (non-Pixiv posting, tag gardening, translation, etc.) I'd consider the issue solved. The problem is that there is no motivation to get people to do that at the moment.

Back to the original point. Things aren't exactly "broken" at the moment, just sub-optimal. Any solution that will cause issues worse than what we have now (such as curtailing features and driving away users) is no better than the status quo. Enhancing the site, allowing us to get more good quality posts with more comprehensive and accurate tags, retaining users while encouraging them to act in ways that help the site, and discouraging them from acting badly should be the real solution. I haven't seen such a solution here yet, so the status quo remains an viable option. I think there have been some good ideas in this thread, but unfortunately more of it has been dedicated to bickering and ad hominem attacks with little effort to try to see the opposing side's point of view.

Just wondering: do we know what Albert thinks of a solution such as a uploading bot in the first place?

And I thought I had a migraine before I started reading this.

To start: The way this was presented was all wrong. "I'm fed up with this and am doing something about it even if people hate me for it" should instead have been a new thread saying "Hey Danbooru community, I have an idea for addressing what I feel is an issue. Here is exactly what the bot will do, and how it will do it. Please discuss whether this is appropriate, and if it is, how it can be made better."

After reading this - admittedly I've probably already forgotten half of what I read - I still do not understand very basic questions about this bot:

1. What quality control is in effect to address Pixiv's notoriously incomplete and non-standard tagging. I only occasionally see a Pixiv image I'd consider tagged to danbooru standards.

2. How the content rating problem will be addressed, considering most non-explicit images are actually safe, not questionable, thus defaulting to questionable will be wrong more often than not.

3. What safeguards are in place regarding "high priority" tags, by which I mean tags critical to blacklisting: loli, shota, sex tags, violence tags.

If the answer is that "it might be able to do a somewhat better job than the more irresponsible "first!"-obsessed uploaders (and almost by definition a worse job than responsible uploaders), and good tag gardeners will just fill in behind it", then I cannot conclude that the benefits outweigh the costs.

It seems like a lot of support you got earlier in the thread turned out to be based on good tagging and good uploads you'd done in the past few months without the bot, thus they don't inform us as to how the bot will perform.

My other issue with this is that your fervor, while based on good intentions, is disproportionate to the actual problem, which was hardly as catastrophic as claimed. If you held off on talking about this because you thought others wouldn't share your belief that it's a gigantic problem, that in itself should tell you something.

There have been a few suggestions in the thread about finding ways to quantitatively identify sub-par taggers and surface their names (whether just to mods or more publicly) and punish (or educate! not everyone who tags poorly is being a dick) as appropriate. These strike me as far more effective solutions than simply making it impossible to upload popular artists.

(Let's face it, there's probably not enough reasonably accessible non-Pixiv material to go around. I partly stopped uploading because I did believe strongly in following individual blogs, just as you do, but all those artists moved to Pixiv and it was pointless to try to manually monitor 200 individual websites when everything was on pixiv - often faster, in better resolution, and with more amendments.)

After all, the problem isn't uploading quickly. It's tagging badly. If someone camps Pixiv but exhibits good tagging, there's no problem.

To be clear, I don't really care about uploaders' feelings. However, Danbooru is still a community. Automation can help in some ways, but if you get hit by a truck or your script breaks beyond repair, human beings still need to be around to pick up the slack. The role of "incentive", which has been debated, is not zero. I believe the incentive should be primarily driven by the individual's desire to contribute usefully (not by concrete rewards parceled out by the site); uploading a good image and tagging it well needs to be its own reward. Users who sacrifice quality for speed would need to be scrutinized - though it won't necessarily be as harshly as you or others might want.

The last thing I'd say is that I would prefer never to see an image on this site that didn't pass through a human filter at some point (obvious exceptions like correcting for Pixiv thumbnails aside).

I could go on but this is enough for now. Other mods, please leave the thread open, since this is obviously not resolved.

Has a maximum upload per day threshold ever been considered?

Clamping people at 30-40 uploads per day. It seems the only people really hurt by this are the "first" folks.

Pyrolight said:
Has a maximum upload per day threshold ever been considered?

Clamping people at 30-40 uploads per day. It seems the only people really hurt by this are the "first" folks.

It already exists for members below contributer, and members above that level are trusted to post as they will. 30-40 is probably higher than "first" people would need anyway, so it would just prove an annoyance more than anything.

jxh2154 said:
To start: The way this was presented was all wrong. "I'm fed up with this and am doing something about it even if people hate me for it" should instead have been a new thread saying "Hey Danbooru community, I have an idea for addressing what I feel is an issue. Here is exactly what the bot will do, and how it will do it. Please discuss whether this is appropriate, and if it is, how it can be made better."

I agree with both of these points.

jxh2154 said:
The last thing I'd say is that I would prefer never to see an image on this site that didn't pass through a human filter at some point (obvious exceptions like correcting for Pixiv thumbnails aside).

I also agree with this. Unless the bot is as good as or better than the average human (which is hard to automate), it should be assisted by a set of human eyes. This is why I was suggesting the bot be made to be semi-automatic or push through it's own queue for users to vet. I think a bot or script or whatever you want to call it has a place in picking and pre-tagging things to make things easier, but mistagging and undertagging is as much a problem if a script does it as if a human does it.

jxh2154 said:
Other mods, please leave the thread open, since this is obviously not resolved.

Rodger. The only reason I even suggested locking it before was because the whole conversation seemed to be devolving into bickering and name calling and people ignoring what each other had said. Things seems to have gotten back on track a bit, and with the bot out of service for now the discussion can continue as it should.

I've been considering hiding the uploader for posts from public (non-moderator) view. Meaning you wouldn't even be able to search for them and there would be no reliable way of comparing upload counts between users. In my opinion this gets rid of the perverse incentive of minimal tag first uploads, but would let mods still track and promote prolific uploaders.

Thinking long term, if users are so interested in inflating their upload count, then maybe further gamifying the site with a generic point system (wherein things like tag updates are treated as equal to uploads) would level the playing field more.

Shinjidude said:
It already exists for members below contributer, and members above that level are trusted to post as they will. 30-40 is probably higher than "first" people would need anyway, so it would just prove an annoyance more than anything.

Fair enough, just thinking of limiting contrib camping pixiv without totally messing everyone up.

albert said:
I've been considering hiding the uploader for posts from public (non-moderator) view. Meaning you wouldn't even be able to search for them and there would be no reliable way of comparing upload counts between users. In my opinion this gets rid of the perverse incentive of minimal tag first uploads, but would let mods still track and promote prolific uploaders.

It would definitely shake things up.

Thinking long term, if users are so interested in inflating their upload count, then maybe further gamifying the site with a generic point system (wherein things like tag updates are treated as equal to uploads) would level the playing field more.

I would love to see a title/level for those that tag a ton.

Since you can't exactly give them caret-blanche as far as uploads go but letting them know they are as much a contributor as an major uploader would be nice.

It might be going overboard but a level that involves both good uploads and good tagging above contrib.

Even if it didn't really offer anything extra I think the acknowledgement would be pretty nifty for those folks.

albert said:
I've been considering hiding the uploader for posts from public (non-moderator) view. Meaning you wouldn't even be able to search for them and there would be no reliable way of comparing upload counts between users.

I strongly disagree with not being able to search a user's posts. Sometimes I like to follow a user's posts because they show good taste in uploads, and that shouldn't be something restricted to mods. It's a bit too extreme of a solution to a problem which I don't think has been particularly serious at all.

Each upload adds a line to the tag history, so they're already linked in some way.

Uploader names have the potential to annoy artists (finding their art on an unknown site, attributed to someone else) so removing or obfuscating them might help. Maybe refer to the user's id number instead of the username?

albert said:
I've been considering hiding the uploader for posts from public (non-moderator) view. Meaning you wouldn't even be able to search for them and there would be no reliable way of comparing upload counts between users. In my opinion this gets rid of the perverse incentive of minimal tag first uploads, but would let mods still track and promote prolific uploaders.

I feel the same as EB about this.

parasol said:
Uploader names have the potential to annoy artists (finding their art on an unknown site, attributed to someone else) so removing or obfuscating them might help. Maybe refer to the user's id number instead of the username?

A better answer if you're doing something like that would be to restrict the "by [uploader]" bit (and maybe the user metatag, too) to Member+, the same way tag history already is.