Some issues with a recent action that was taken regarding the Root Access chat bot

Question

I would like to discuss with some civility a recent action that has occurred in the Root Access chatroom. First, allow me to catch everyone up on the facts of the situation:

Overview

1. Q: What happened?
   A: Please see this message in Root Access chat. This decision is the primary reason for asking this question on meta.SU. But before you make any judgments or conclusions, please read the other statements of fact which follow.
2. Q: What is the chat bot's account name?
   A: ChatBot John Cavil.
3. Q: Where is the source code for the chat bot located, and what is its source code lineage?
   A: This chat bot is based on the JavaScript ChatBot implemented by Zirak. Its code is located on GitHub. There are three repositories of interest: My Driver written in Java, which provides login functionality only; Zirak's base ChatBot; and my minimal changes as a friendly fork of Zirak's bot. The primary purpose of my changes is to adapt the functionality of the ChatBot to the Root Access chatroom, insofar as the existing code is specifically tuned for the JavaScript chatroom, which is the original home of Zirak's ChatBot.
4. Q: Where else is a similar ChatBot (based on Zirak's code) hosted in chat rooms on the SE network?
   A: In the aforementioned JavaScript chatroom, as well as the Android chatroom, and perhaps others that I am not aware of. Some users in other rooms have expressed interest in having one, but have not yet put the hosting infrastructure in place to host one.
5. Q: Are bots allowed on the SE network in general?
   A: It would appear that they are. I completely understand the proviso that SE has the final say regarding what they consider allowable and what is not considered allowable. I am eager to work with SE and the other ChatBot maintainers to ensure that these bots meet community standards and can continue to be a part of the flavor of the chat rooms for a long time.

The Problem

1. Q: Having read balpha's message explaining the reason for suspension, what is your (allquixotic's) interpretation of the reasons for the suspension of ChatBot John Cavil?
   A: I see the reason as being three-fold:
   1. The bot greets every user who enters the room for the first time, if it has never seen them before.
   2. The bot contains commands which contain foul language, or might be viewed as objectionable and not rated "PG".
   3. The bot's message is written in a way that could be misinterpreted as indicating that the bot is part of the official chatroom functionality, or that it is officially endorsed by StackExchange, Inc.

Observations

1. It is observed that other chatbots based on Zirak's code have existed on the StackExchange chat room network long before John Cavil came into existence.
2. It is observed that every other chatbot based on Zirak's code...
   1. ...contains the same built-in command which balpha specifically found objectionable.
   2. ...automatically greets users joining the chat for the first time with some sort of welcome message.
   3. ...allows users to define their own commands, which can contain arbitrary text, links, or oneboxes, including foul language or inappropriate material.
3. It is observed that regular users in Root Access, The Comms Room, The DMZ, and other popular chatrooms on the SE network, very frequently use foul language and post objectionable content, which is not flagged or removed from the chat history, even in the presence of several site moderators. It would appear that this content, in general, is considered to be acceptable by the majority of the users of these rooms, except in cases where it crosses a line in the opinion of the users, is flagged, and is taken care of. The efficacy of the chat flag system is very clear, because flags are usually handled within seconds, and valid flagged messages are almost always nuked within seconds. It would seem that this system should be able to take care of any foul language or objectionable content being output by the bot (either built-in, or induced by a user) in cases where chat users find it to be objectionable, and suspensions are unwarranted.
4. It is observed that, to the best of my knowledge, ChatBot John Cavil does not have any chat flags which have been deemed valid, based on a user supplying non-objectionable input commands and receiving objectionable output. (It may be the case that it may have a flag when a user typed something that contained objectionable input, in which case that user, not John Cavil, should be held responsible for the output. It is impossible to write a filter for a chat bot that will absolutely prevent any objectionable material from being printed in chat or linked to, regardless of user input, without severely limiting its functionality.)
5. It is observed that, as of this writing, and to the best of my knowledge, ChatBot John Cavil is the only chatbot which has been suspended from chat, despite containing the exact same feature set as other chatbots on the SE network, which on the whole have been met with chat user approval and have seen continual maintenance and improvement in cases where their behavior has proven annoying or undesirable.

My Questions

1. Given the observations above, why is ChatBot John Cavil, specifically, suspended, and not the other chat bots?
2. Without doing away with the ChatBot entirely -- which, thankfully, was not requested by the StackExchange employee who suspended it -- what are ways that I can improve its functionality in order to make it more acceptable? I will take a shot at this right now; if you have any additional suggestions, please post them as an answer or a comment:
   1. I would very much like to keep the automatic greeting in some form or another. The wording can easily be tweaked to indicate that the ChatBot is not built-in, and is not officially endorsed by StackExchange.
   2. I completely agree with the sentiment that removing the command which was found specifically objectionable by balpha is sensible. I will, additionally, submit my patch as a pull request on GitHub for Zirak, in the hopes that he will "clean up" his bot as well. I will then do a search through the command list and attempt to remove any other built-in or user-learned commands which could be construed as objectionable.
   3. Lastly, if it is the only way that StackExchange will agree to allow the bot to continue to exist, then I will reluctantly remove the automatic greeter.

The Case For The Auto-Greeter

(Big Kudos to @Skliwz for making me aware that I needed to add this section!) Here are my responses to some of the common objections which have been made regarding the auto-greeting functionality:

1. Objection: The auto-greeting messages are obtrusive or noisy for users who are already acquainted with the room.
   Response:
   1. Relative to some more popular rooms, there really are not all that many new users who visit the Root Access chat, so the frequency of the messages is pretty low. Maybe one or two a day at most. There were a few spikes when popular questions linked to chat, but those are anomalies. We may be able to add code to the bot that stops greeting users if more than X number of users enter the room within a Y second window.
   2. Some room regulars find it useful to know when a new user has entered the room, because often we will immediately stop some off-topic conversation and prepare to help the user with their question.
   3. When a user joins the room, there is currently no way for regulars to know whether that user is completely new to Root Access; completely new to chat; or even whether they are able to post messages. Generally, users that the bot has never seen before are going to be the most impressionable, and we want to make a good impression, so we will make a greater effort to eliminate side conversations to allow the new user an opportunity to interject with a question about the site or seeking live support. Experienced users will not feel put-off by a continual stream of conversation, and are more likely to interject even if we don't make an opening for them specifically.
   4. Users who do find the bot's messages to be annoying can choose to ignore the bot using a feature built into the chat client.
2. Objection: New users who wish to lurk in chat and not participate in the discussion are "pinged", which can annoy them, or bring unwanted attention to their profile or user name.
   Response:
   1. SE is a public site; user profiles and display names are public; and users should have no expectation of privacy when creating an account and electing to provide information in their profile. Users who wish to remain anonymous may do so: if you do not create an account (or even if you do have an account, but aren't logged in), you do not appear on the user list, and will not be "pinged" by the bot, or any other user.
   2. Everyone can already see your presence in the chat room in the user list if you're logged in. We know you're lurking, and we're totally fine with it. I have never seen a regular in Root Access ping a lurker just because the ChatBot pointed them out. We tend to only speak to users who have participated in the chat at one point or another, and are acquainted with at least a few people in the room.
3. Objection: New users may mistakenly think that, due to the auto-greeting, the chat bot somehow represents the official points of view or stances of StackExchange (the company).
   Response:
   1. I am willing to modify the greeting message in any way that would help to clear up this misunderstanding, either by supplying text dictated by StackExchange themselves, or a community-approved string, that makes it unambiguous that this is a community, unofficial bot that is simply trying to provide useful and/or entertaining commands and information.
   2. It is already a common misconception among new users that diamond moderators represent the views of StackExchange, Inc. Clearly we know that they do not, as veteran users of the site, but it is very hard to avoid this initial impression. The best we can do is attempt to inform the user to the contrary, which some moderators explicitly do in their profiles. Maybe I can also add verbiage to this effect in ChatBot John Cavil's profile to help the situation?

Answer 1

This isn't actually an answer proper, but there's a few points where I suspect the process of well, all this could have run better. Hindsight is 20/20 , and this is not really a reflection on SE's moderation and administration processes. I'm posting this as a root access regular, and someone intimately familiar with the channel and its culture, and having heard one side of the issue. To mangle a quote from b5, understanding is a three edged sword, your side, their side and the truth, and I'm hoping this gives another perspective of this.

• There wasn't any indication that an automatically triggered script was an issue before action was taken here. In general, the rules seem to be bot can't do anything a regular user can't. I suspect that's the real issue here, since it felt like a sudden decision, rather than a wider sitewide policy. We'd have had a lot less of a mess if we weren't working out what the lines were after action was taken. I do understand that the dev team explicitly didn't want this behaviour, but there was no way that the upstream or RA bot devs would have known of this.

• Neither the regulars in the room, nor newbies had mentioned any issue with the bot. If we had felt so, we could have probably had a word with @allquixotic. The robogreeting was actually tweaked in consultation with the users, and for most part, the bot hasn't been a annoyance.

• The first we knew of this was the bot getting suspended - which is a tiny bit like getting a telegram delivered through the roof by a swat team.

A little less shock and awe, might have been nice. @allquixotic has been very nice about tweaking his version of the chatbot (as time permits!), and this could have been probably handled with a little less drama (of which there was a little).

Answer 2

This may not be the proper place to post this, but because of traction, we were pretty much ordered to disable the bot's auto-welcome in the javascript chat, which makes it kind of a big deal to me. Since most participants in this discussion are seeing the auto-welcome as one finished unit, I'd like to say how it came to existence.

But first I'd like to talk about margarine and my great-grandfather.

My grandmother used to tell me how she worked during vacations. Her father (let's call him Bob) was an engineer (or something of the sorts) in a company which made oil-based produce. One of their products was margarine. I think she told me one of her duties was packaging the margarine slabs. You know, putting them into these neat rectangular packages. It was a pretty annoying job, but it quite a few decades ago, and they didn't have something better.

And then they did. Bob, my great-grandfather, added a component in the production line which packaged the margarine. It was (and is) very not-trivial, since you have a moving product and you have to fold it just right with fixed mechanical movements. But it was a success, and all was well, and happiness was spread through all (except those who ate the margarine).

Since we're mostly programmers here (and those of you who aren't can still relate to the point), we know that any advancement, any introduction of a product is made out of some simple steps:

1. The world is in a state where problem A exists (A can be margarine packaging, communicating with your friends, eating ice cream, anything).
2. We identify A as a problem which we want to fix.
3. We build a solution (this may take a while).
4. A is no longer a problem for us.

This is a self-feeding loop, since introducing a solution can spawn many other problems, known as "bugs". Our solution isn't necessarily perfect.

Let me introduce a problem. We're in the javascript chat room. We want and like to help people (we really do!). We dedicate hours of our lives to educate and solve problems without asking for anything in return, because of our own reasons (some of us are bored, some of us are interested in the problems, some feel that education is the best way to advance the js community and the programmer community in general, etc).

And we notice things which annoy us. One of them is when new people show up, they have the same problem StackOverflow has - they're used to a certain format, but we work in another format. In some chat rooms (especially MMO chats for some reason), it's common practice to say things like "anybody here?" or "can anybody help me with X?" or "can anybody help me?", or maybe they just randomly select someone and ask. But we don't quite like it. We're here to answer questions. We can't answer your question...if we don't know what it is. Answering these questions produces 0 value in general, and just aggravates us because we see them over and over and over...

We notice a pattern, we notice a problem. We're now at the second step to solving it: https://github.com/Zirak/SO-ChatBot/issues/39 And we have the technology.

So we make a system. It had bugs, the community complained. They were so severe, we shut the system down, only two days after it was launched. But we didn't give up. We improved the system, we fixed bugs, we considered edge cases. 9 days later, it was live again. A month later, the system got improved again.

The system solved our problem. It didn't just do that, it did a lot more - new users really liked it (1, 2, 3, 4, 5, ...). It made them feel welcome (maybe because, well, the message does welcome you). So not only did it mostly solve our problem, it improved the overall feel of new users. Unlike IRC welcome messages, which are actually giant walls of text, this was a 2 liner asking something simple of you.

Can the system be better? Certainly! It still has bugs, it can still be improved, it can be more meticulous about who it welcomes and who it doesn't using better metrics. Heck, it might even have cross-site problems, since I really just made it for the SO chat, not SE in general. However, the important fact here is that it solves a real problem we had. Before it, we had to say "please don't ask to ask, just ask" about four dozen times a week. Now, it's a lot less (because some people either manage to forget, or need a reminder). Just today, when I was catching up to the whole story, someone came in and showed us the problem again.

I'd like to point out another important fact: Just because the system works for us doesn't mean it'll work for everybody. Some rooms don't need this kind of message. Some might just not like the idea. Some might not even have the problem to begin with - because we're a heavily invested technical room this problem began. Horse Grooming chat rooms may have 24/7 long discussions about...horse grooming, without ever having to face this.

And another important point, which may have been lost in the flow, the automation of it. In several points, we were asked to just do it manually, that is if we see a user acting in a problematic way, we welcome him. But that's just going back to how thing were before the system, which as described, was not ideal, which is why the solution sprouted in the first place.

To conclude, I'd like to say how community driven the bot is. Benjamin Gruenbaum said it before me, and I'd like to repeat: The bot is shaped by the community. We have an active and healthy discussion going around for every suggestion. If we see something isn't going well, we talk about it and improve. Features were added and removed, features were thrown away at planning stage, and it's all community driven. We also ask moderators (one of SO's mods is also a regular in the room) in some gray areas, ask for their approval. We do all this because we're programmers. Our job, our source of fun, is identifying problems, and solving them. So I'd like to ask the moderators to reconsider making this site-wide. Since they're the people who make the rules, my next commit is going to respect that and remove the auto-welcome. But it won't feel right, and it won't feel fair.

Answer 3

I suspended the bot because of the greeting, and only because of that, which is also what I told you in my chat message.

Root Access is the general chat room for users of Super User, and any new user coming in being curious about the chat will have that bot greeting them. That is a functionality that the Stack Exchange chat system specifically and intentionally does not offer, and a user-written bot shouldn't either.

Given that the user who is in control of the bot – you – wasn't available in the room at that time, I disabled the bot by "suspending" the user you created for it and left you a message, in particular telling you

I have suspended the bot for now, let me (or another moderator) know when the greeting is off.

I put the word "suspending" in quotes because it's not like I put a real user into the penalty bin – I just disabled a bot.

As to the question of why it's allowed in other rooms – it's not. I just wasn't aware there were more. Tim, one of our community managers, will check out the other rooms you mention and enforce the same thing there.

If a room is specifically created for a bot, or the room's purpose is very specific to a well-defined group of people who consider the bot fitting and its greeting appropriate, we don't have any issues with it.

But in this case we're talking about a general for-everybody chat room, the first point of contact of new users with the chat, and in such a room, we have to ask bots to tread lightly.

As to the /fuckable command, I merely asked you to reconsider. Personally, I find it extremely immature an inappropriate, but I haven't made up my mind about outright disallowing it, and if you do reconsider and remove it, we wouldn't have to make a decision about that. If you don't, then we'd have to talk about it and decide. From your point 2.2 it looks like this won't be necessary in this case, but if you say the same bot runs in other rooms, there may be other cases.

Finally for completeness and reference, our general stand on chat bots is stated in this answer of mine.

Answer 4

So you are spamming anyone who joins the chatroom.

I'm puzzled how you could even think that assaulting anyone who joins the chatroom could be appropriate. It's bad enough when people do that manually, but automatically‽

Any auto-greet message is spam and that makes the auto-greeting bot a spammer. Banning that bot was appropriate. Letting the bot run was not appropriate.

A bot should never ping a user who didn't opt in.

---

I've proposed a change to the chat interface that would make this issue moot. If joining a chat room was an affirmative step, rather than the automatic and potentially accidental thing that it is now, I would have no objection to users joining a chat room being greeted. (Not that it would necessarily be a good idea, but it wouldn't be spamming undemanding users.)