As someone involved in this, I'll state the complaints I see and exactly what I propose to improve the situation. My goal is not to create a perfect world without sexism or anything dramatic like that, but simply to get things to a point where any user on the main site feels comfortable participating in chat. We can debate feminist theory and the effect of anime/manga on women in Japan another time, but frankly, that's only marginally relevant here and our goals are much more down-to-earth.
First, I feel I should clarify the situation. There are two connected components that together determine the effect the bot has on the chatroom. The bot itself is a piece of software written by Hakase which accepts links to images and posts them randomly according to rules which can be modified by verified users (room owners and moderators last I checked) and on-command. In principle, any user who wants to can add any image they like to the bot. The goal of this was to create something which would allow users here to upload their favorite anime/manga-themed images images and save them for times when the chat has downtime. This would create more discussion. I will refer to this as the bot.
In practice, I'm the only user who has done so, so the collection is rather slanted to my own interests. Those interests are somewhat limited. I have a collection of approximately 25000 images which I collected over a period of several months. I've uploaded them to Google Drive and fed them to the bot about 1-2k at a time when it gets low on images. I'll refer to this component as my library.
It's important that these two components are not conflated. The bot itself was always intended to be a community project. The fact is that there isn't much community involvement in it, but that's really the problem here. Because of that, it is drawing exclusively from my library, and gives off the impression that the entire community's interests are the same as mine, which is not exactly true.
Now that this clarification is out of the way, I've seen complaints that basically fall into 2 categories. The solution to both of these is more community involvement, but it's useful to differentiate the two since they're conceptually very different. This post is rather long, so I've made a tl;dr version for the solution to each of these problems.
1. The bot posts some inappropriate images
I've heard this from a relatively small number of people, but it seems to come up repeatedly. It's rare that people can point to a specific image and say exactly what makes it inappropriate, but there is a general sense among some users that some of the images aren't as unobjectionable individually as one might like.
To be clear, this is a complaint about my library, not the bot itself. As such, I'll detail the procedure I go through to check the images and my goals in doing so.
Initially, when downloading images, I used various image boards. Pixiv was probably the most common, followed by Danbooru, Sankaku channel, e-shuushuu, yande.re, and konachan. Most of these sites are not work-safe, but all the ones which aren't safe rate images, and in searching, I stuck to those images rated as "safe". This already removes anything sexually explicit, most nudity, etc. The downloads were done by hand.
In reviewing the images, I noticed that a small number of them were still perhaps debatable, even though they had been classified as "safe" rather than "questionable". They were mostly no worse than things we've posted in chat before, but since the bot would be posting randomly on command and would be a regular user, upon talking to the other room regulars, we figured that the images it posts should be held to some higher scrutiny than what we'd hold the regulars to.
As such, when I upload images, I manually hand-check each one and see that it conforms to reasonable standards. Since users under the age of 13 aren't permitted on SE, it seems reasonable that the bot should be permitted to post roughly PG-13 level images, but in practice I aimed for mostly PG-level. For reference, the sorting is not solely to remove images which are not appropriate, but also to remove those which are not of particularly high quality.
Because of the sheer magnitude of images needed (the bot posts 20 images per day, which means we need over 1000 new ones every 2 months) these checks have to be done quickly. In practice I average 5-10 seconds per image, which means that going through about 1200 images (which will produce about 1000 to be posted in chat) takes about 2 1/2 hours by hand.
In checking, I have a small number of hard-and-fast rules for what to remove. Anything with nudity or visible underwear is removed. There are a small number of other less common rules I follow. Beyond that, anything I'd need to think twice about uploading from my account, whether or not I'd actually do it, is removed. The end result is that the collection is significantly more conservative than what I myself would post. I've seen many other users (including many of those who are now complaining) post images that would not meet these standards.
I say all this not because I think it means that I should get a free pass, but simply to say that this is something that I'm already putting a great deal of effort into. I think the large majority of the images posted are not objectionable individually. There may be a small number of seriously objectionable posts that I miss. This is inevitable; I can't spend several minutes on each image dissecting it, and even if I could, there would always be a small miss-click rate. This small number might still create an impression on the regulars; if it's only 2% of all images, that still comes out to 3 or 4 images per week at the current rate, which a user might see as a much more statistically significant trend than it actually is.
Beyond that, there are images that some users have complained about, but can't explain exactly what the issue is. Upon investigating such cases, I (and often some other users) can't see anything wrong. This illustrates an important principle that what is appropriate is subjective. If it were possible to do so, I'd read the minds of all the regulars and check that each image is acceptable to each one. But I don't have the ability to read minds, and I can't see what one would complain about in many of these cases. I'm happy to refine my methods further if people have concrete criticisms that lead to useful checks that can be performed quickly by eye. But I'm already incorporating every check I can think of at the moment, so simply saying that an image is "NSFW" or "suggestive" without qualifying what exactly in the image is the problem is not particularly helpful, because in all but the rare cases of a mistake on my part, the image is SFW by my standards. I encourage those who think there are systematic errors in my methodology to point these out so that I can improve it.
On the other hand, if you see a questionable image, but can't exactly explain what the problem is, I'd suggest you think a little bit more about what is actually wrong with it. If it isn't something you can explain after thinking about it, it's probably highly subjective and you may want to try to have a bit more open mind about it at least until you can figure out what the problem is. This isn't really any different from anything else in life; for example, while some people have opinions on others without having any solid reason or explanation for their opinions, it generally behooves open-minded and intelligent individuals to explore these opinions and discover the reasons behind them. But if you're sure the image is inappropriate, and still can't explain why, keep reading.
I'm also not saying that users should ignore these few inappropriate images. That doesn't solve the problem. The bot shouldn't be the reason an active user on the site doesn't want to go to chat, even if they only see inappropriate images rarely.
The solution I propose is that (other) users become more involved in moderating the bot to deal with these potential issues. Just like the main site, the way we can get this to work is by community moderation, which is only fitting given that the bot was always intended to be a community-run project.
I'm perfectly happy to allow any other chat users who want to do so to inspect the images in my library before they're fed to the bot (in fact, I've asked several times if there is any interest in doing so). but that doesn't solve the problem completely.
Preliminary inspection can only go so far, since even if my library is totally clean, there's little to stop other users from uploading their own images which don't conform to the same standards. What we need is for people to be moderating the bot's images. This is allegedly a community project. It was started with broad community support to provide the community a nice tool to keep conversations going and interesting. But now, it seems that everyone in the community other than me is no longer interested in contributing. If we want the bot to be running in the main room, it behooves regulars of the room to at least familiarize themselves with its internal moderation capabilities and apply these when appropriate.
First off, the bot can (and should) be turned off during long serious discussions, such as those related to meta policy. This can be done via the command
#post disable. You can also modify the parameters such as how often it will post or how many messages to wait between posts; for more details I suggest referring to this meta post about Maid Café chat room bots.
More importantly, when an image does appear that isn't appropriate, you have the ability to remove it within 2 minutes. The command
#undo will remove the last posted message by the bot. While I've seen many users complain about messages, few if any have bothered to actually remove them with this. I'm not sure if that's because it is not sufficiently well-known or for other reasons, but in any case, that's available. This should be done if the image is definitely not appropriate for the room.
Secondly, if you personally don't want a given image on your screen, but it isn't clearly inappropriate, the
#unbox command will edit the previous message to add an extra character after it within 2 minutes. This will allow other users to click the link to see the image, but you won't need to see it. The current functionality is slightly broken; it will still edit the message, but not correctly convert it to a link. This should be fixed shortly.
To be perfectly clear, these functions are available for any users, and everyone should feel welcome to use them. I ask that you reserve
#undo for clearly inappropriate cases and use
#unonebox in questionable cases. But no one should feel like they have to put up with a particular image on their screen which makes them uncomfortable from a bot whose intent is to make chat more enjoyable and comfortable for everyone who wants to join.
While these are pretty good and should make moderation easy enough, I think we can also do more on the software end of things to make the bot more flexible. 2 minutes is a fairly short amount of time, and it's possible to miss an image for that long that you want to get rid of. While the image will usually fall off the screen that quickly, it may still be visible to some users with certain user scripts installed on the star board or if you scroll the screen up.
A regular user (like the bot) can't delete or edit their own messages after that amount of time though. For now, you can message me or another chat mod to get them removed, but that's less than ideal, and chat flags are even worse. Ideally, I'd like for the bot to have room-owner privileges, which it could use to move messages after the 2-minute deadline to a different designated bin room. This functionality does not yet exist, but if it is implemented, that would allow users to deal with such messages after 2 minutes. This is something I'm only somewhat considering right now, and there are possible objections to having the bot be a room-owner, but if this complaint continues to arise, it may be the best option.
TL;DR: There are several functions already available to anyone to moderate the bot. We can't hope to solve this completely without community involvement, nor should we try to since it's always been a community project.
#post disable turns off the posting module.
#undo removes inappropriate messages within 2 minutes.
#unonebox converts the one-box to a link to remove questionable images from your screen while keeping them accessible to other users.
- If 2 minutes is not enough time, we can try adding a function to bin an image, which would have no time limit. Feedback is appreciated on this suggestion.
- In the interim, you can message a mod to deal with any other cases, or in particularly egregious cases use the chat flag system (though it's definitely preferable to use the bot's own internal moderation capabilities whenever possible). I can't speak for any of the other mods, but I at least will be fine removing or unoneboxing any image someone complains about, whether or not I agree with their complaint.
But, in any event, I don't believe that the seriously inappropriate images are actually extremely frequent, given the amount of work I've done in pruning my library. They may occasionally occur, but they're something of a red herring here. Still, feel free to liberally use the above moderation tools; they're there to be used.
2. The overall effect of the bot's sole focus on female characters has the aggregate effect of objectifying women.
Let's clarify this, since the use of "objectifying" here is not standard, at least in the feminist literature I'm familiar with. The current collection of images is essentially all female characters. The aggregate effect of a bot which only posts females gives off the impression (true or otherwise) that the community is only concerned with female characters. It would not be a great leap in inference to assume that the (mostly male) community here is primarily interested in female characters for the sake of their femininity as an object to be distilled and consumed regularly, and then relating that back to real life, may get the impression that this is a somewhat sexist community.
This is different from the sort of objectification you hear about in e.g. movies. In movies, the complaint is the reduction of a (typically female) character's personality and role to that of an object which is effectively owned by another (typically male) character; this is an in-universe phenomenon that the viewer has no part in beyond observing and internalizing. Here, since we have still images, the complaint has nothing to do with story or personality; rather, it's about the perception of this community that a stream of images of feminine characters creates. No individual image is objectifying in this way, as none of them are able to depict characteristics or story and exist only to be consumed. In the standard sense, this is really weird terminology which would seem to have us objectifying a character from an anime or illustration which is literally already an object to be consumed by us; it's hard to see how that is an issue if you don't look deeper.
But having many of these images together without any others may still have some effect on the perception of our community. Specifically, it may have the effect of associating us with the idea of objectifying women and leading to the inference that we are fans of such objectification (valid or not). So I'm not particularly a fan of the use of "objectification" in this way, but we'll stick with it since I don't think there's a significantly better standard term in the literature I've read.
To be perfectly clear, this was not my intent when compiling the images. I've explained my thought process thoroughly in this chat message; I don't think it bears repeating here simply because it isn't particularly important. One of the lessons of feminist literature (and more generally literature on all culturally disadvantaged groups) is that something which is intended as acceptable and reasonable can still be unintentionally perceived as discriminatory or aggressive. I think we may have fallen into a minor case of that here.
I don't think this would be such a problem if it were only me doing it, but a bot has an illusion (true or false) of being community-supported. The impression is that the only value we find in anime/manga is the abundance of cute girls, which is not true for most of the users here.
Let's be honest about this. The bot is basically a red herring when it comes to this complaint. It's an easy target since it's visible to regulars, but it isn't even close to solely responsible for the perception of our community.
Our main chatroom's name is "Maid Café (メイド喫茶)". That already probably turns off a lot of users including some women. The concept of a maid cafe isn't exactly gender-neutral. Furthermore, the description is written in a mixture of English and incredibly ugly italic Japanese text with a bunch of very technical terms that the average person would not understand or would potentially be turned off by.
Our dialogue isn't even close to gender-neutral either. Forget the bots; our users enjoy talking about things like fanservice scenes and praising lolicon. The images the bot tosses out aren't even close to the worst ones that get posted. And this isn't a new thing; chat has been that way since long before we had any bots. Indeed, the bot was initially created to automate a process I and other users were doing manually of posting images during downtime, predominantly of female characters.
Now, before we go any further, we need to ask ourselves to what degree this is actually a bad thing. "Objectification" is something of a feminist buzzword, but we have to question whether it's actually a problem here.
In some ways, it really isn't such a bad thing. Yes, it has the effect of excluding some people from chat. But exclusion isn't a bad thing; it's the principle on which Stack Exchange was built. For those confused by this, I highly suggest this great talk by Joel Spolsky. The relevant section begins about 14 minutes in, but the whole talk is interesting:
The Cultural Anthropology of Stack Exchange
The point is that the way Stack Exchange works best is by creating an atmosphere that is significantly more welcoming to experts than to those who aren't. In the context of anime, our goal should be to make an atmosphere that experts find comfortable, and users who don't know or care about anime will not have any interest in. Chat being the deepest level of the site, it should be the most biased in that way towards keeping the expert interested users around and active.
I don't think anyone will question that there's a definite correlation between people who appreciate the bot's output and people who have a strong interest in anime. So, in some sense, the bot is really doing a good thing here.
But that's only at a very surface level. While interest in the bot does correlate with anime expertise, it also correlates with all sorts of other spurious things that we don't really need. Female anime fans are probably less interested than male ones, and fans of genres like action or (heterosexual) romance may feel left-out too.
I don't know whether the overall positive effect is bigger or smaller than the overall negative effect, or even if they're comparable. But we can keep most of the positive effects and get rid of the negative effects. These are real issues that we should try to solve and can make some progress with here, unlike the huge goal of ending objectification of women (which we're simply in no position to deal with on any real scale).
Getting rid of the bot isn't going to fix sexism in the world, or even in our chatroom. It isn't going to change our culture. Nor is it going to make more women interested in anime and manga as a whole.
And this isn't the headquarters of some feminist organization, it's a chatroom about a Q&A site for Japanese cartoons and comic books. We need to accept that anime and manga aren't gender-neutral, and that nothing we do here is going to change that, nor should that be our goal. Anime is still primarily made for a male fanbase and our site's demographics are accordingly skewed more towards men than women (according to Quantcast; incidentally, we aren't anywhere near as skewed as a lot of other sites like many of the tech sites). There are some very positive effects anime and manga have on gender-roles in Japanese culture, but also some negative ones, and like it or not, we're bound to both of these here.
Our goal should just be to make a chatroom that main-site users feel comfortable and welcome in.
So with the more modest goal of representing anime fans equitably rather than trying to cure "objectification", there's some real progress we can make.
The solution to this, to the extent that we need to solve it, is in fact quite similar to the other problem. We need more people adding more images of their own interests to the bot. The bot is supposed to represent the community's interests as a whole. Instead, it's come to represent solely (a subset of) my interests.
In fact, any user already has the ability to add images to the bot. There's few real rules about what the images should be either. Anything you want to add is fair game, be it cute girls, handsome men, or artichokes. The only serious rule is that everything should be appropriate and on-topic (and the latter is using the broadest possible interpretation of on-topic).
The process is unfortunately a bit complicated. I intend to make another meta post detailing the steps I take so that other users can follow the same steps. However, anyone can do it who is sufficiently interested in doing so.
I encourage other users whose interests don't (or do, if you like) match mine to add your own images. I'm not going to go out of my way to look for a ton of images that I don't even care about (or rather, I can't, since I can't even judge them effectively), but everyone else should feel free to share their own interests through the bot. That's what it was originally intended for, and to the extent that people are having issues with it now, it's because there isn't enough participation on this front.
I don't know if this will cure the issue of "objectification" from a constant stream of images of female characters. But what it will do is make the stream of images representative of what we do care about and are interested in. More diversity in interests, so long as they remain anime- and manga-related, is surely a good thing in light of what impression it gives off of our community. And if we get a lot of fujoshi in chat who really want to share their love of BL or male idols or shotas, I'm perfectly happy with that so long as it stays on-topic and appropriate, and that might help counteract the image that all this community cares about is cute girls. Or we might end up with 35000 different images of artichokes if that suddenly becomes the most important thing for us to chat about. The possibilities are really endless.
TL;DR: We need users other than me uploading their own images to the bot to more equitably represent our broad interests. This was always intended to be the case, but only after this discussion did we realize that by failing to do this, there can be negative effects as well as the positive ones the bot creates in chat for some users
I've already had a couple users express interest in adding their own images in chat. I will upload a guide to meta shortly on how to do this. Other users interested in doing this are definitely encouraged to do so.
As much as I would like to solve this on my own, it simply isn't possible. The fundamental problem is that the bot isn't representing people's interests, and so those people whose interests differ from mine need to go to some effort as well to get them represented by the bot. I realize that means some work, but this is a community undertaking, and a single person is not a community.
In all of this, the unifying theme is that the bot really does have to be a community task. It can't be just me adding and managing images. It was originally created to be something for everyone in the community to both add to and post from, but somehow that didn't happen. There was a strong consensus among community members that we wanted a bot like this, but no one besides me really wanted to put much work into it.
If the community isn't willing to take this up any more, then maybe we really don't need the bot any more. Sure, I might leave chat if that happens since I'd have little more reason to stick around, and some other users might be sad. It may also kill a lot of discussion that we currently have about the stuff it posts. But all of those are potentially acceptable if the bot is creating an atmosphere where a lot of serious anime fans on the main site feel under-represented and not particularly welcome.
On the other hand, if the community is willing to step up and add images and moderate the bot, I think it will be one of the most unique and engaging features in our chat and keep everyone interested. Right now it's having a positive effect for some people like me, but not everyone, and that isn't good. But it really doesn't take any major changes to the bot itself to make things good for everyone; it just takes people putting in a bit more effort to make the bot their own.