Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 155 Stack Exchange communities.

What is meta?
Here's how it works:
  1. Any Stack Exchange user can ask a question
  2. The community provides support, votes on ideas, and reports bugs
  3. Your voice helps shape the way Stack Exchange operates

There is a controversy about a certain chat bot in some chat rooms. The one major disagreement is about the bot greeting new users in chat. Some users argue that this is helpful to explain the chat room rules to new users and make sure they actually notice them, while others argue that bots shouldn't annoy users at all.

I have a rather strong opinion on the subject, and I think a definite rule on bot interactions would be helpful. Should a chat bot be able to initiate an interaction with a user, or should any interaction be initated by the actual user?

What should be the limits be that chat bots are subject to and that are enforced across the SE network?

share|improve this question
Can you elaborate on that? – jball Nov 12 '13 at 23:02
@jball You might have to be more specific. I don't want to link the specific issue as I want to discuss the general case. – Mad Scientist Nov 12 '13 at 23:04
Why do you say that? – jball Nov 12 '13 at 23:11
In other words, you have a strong opinion on people happily using a bot in a room you don't visit and would like to impose your (opposite) views on them... – Lorem Ipsum Nov 12 '13 at 23:45
I have upvoted this question because I feel that it poses the subject in an impartial manner that allows for answers to provide their own viewpoints. NOTE: I am in the "auto greetings are good" camp. I encourage others also in this camp to upvote, because the question is good, there is really nothing at all wrong with it. If you think something is left out, please edit or suggest an edit. Let's hash this out right here. – allquixotic Nov 12 '13 at 23:50
@allquixotic I'd edit out the opinion from the question and post it as an answer. That way it's clear that votes are not indicating agreement with the stance but with the fact the question should be discussed. – Benjamin Gruenbaum Nov 12 '13 at 23:53
@yoda I have a strong opinion on people happily using a bot in a room and ignoring the unhappy users that they're driving a way. – Gilles Nov 13 '13 at 0:02
@Gilles I'm convinced there are none or next to none – Jan Dvorak Nov 13 '13 at 0:03
@Gilles Show me the unhappy users (meta posts, more than a handful comments, etc.)... otherwise, I have strong opinions as well on people spewing unsubstantiated claims willy nilly and driving away users. You don't see me asking for their commenting privileges to be revoked, do you? – Lorem Ipsum Nov 13 '13 at 0:34
@yoda I am such an unhappy user. So is balpha. – Gilles Nov 13 '13 at 0:50
@Gilles First, please don't put words in balpha's mouth. He was only asking about greeting and in the scope of that chatroom. Second you've chosen not to respond to what yoda has said. His complain was about making a site wide policy and discussing it in meta. He never had any beef with people in rooms who don't appreciate the bot interacting in some way telling it not to. He was just asking why impose an opinion on other rooms. – Benjamin Gruenbaum Nov 13 '13 at 0:53
@Gilles I think benjamin already clarified that enough: "He was only asking about greeting and in the scope of that chatroom." – Jan Dvorak Nov 13 '13 at 0:58
@Gilles so it has devolved to justifying out of context quoting (his quote is about that SU room), and ad hominem? Kind of disappointing... well - I guess I needed to go to sleep anyway. Good night and happy coding :) – Benjamin Gruenbaum Nov 13 '13 at 0:59
I second Benja's comment - your last comment is ad hominem, @gilles – Jan Dvorak Nov 13 '13 at 1:01
up vote 25 down vote accepted

Call me naive, but... Why do we need special rules for bots?

The rules for humans are pretty straightforward - in particular, the rules of etiquette boil down to, "don't annoy other people". If a chat-bot is annoying people (by posting lots of noisy messages or - just as an example - graphic videos of a sex-change operation) then it's violating the rules.

Perhaps the only extra limit needed here is one similar to our rule of thumb for sockpuppets: a chat-bot shouldn't allow you to do anything you couldn't do by yourself, with your own account.

share|improve this answer
Is the greeting annoying people, @Jan? Big difference between a polite one-liner and the (often pages) of boilerplate that some IRC systems are fond of, IMHO. Of course, annoyance is always subjective. – Shog9 Nov 13 '13 at 1:41
I find it annoying when I join a room and someone pounces on me. But at least if it's a human then it can be a preamble to starting a conversation with that person. That doesn't apply to a bot. I think your rule of thumb is incomplete: a chat bot should also be able to do everything you can do by yourself, to wit, participate in a conversation (pinging someone and then refusing to talk to them if they reply would be rude). That would kill chat bots, and I don't think this is right. So I disagree with that rule. – Gilles Nov 13 '13 at 1:41
The reason for this discussion is to ask the community, to what degree are once-per-user automatic greetings "annoying"? When you take any given stimulus and apply it to a large group of people, you will have widely variable responses, from approval, to disgust, to annoyance, to indifference, and so on. There is no objective definition of whether X behavior is annoying. Well, maybe, a mosquito flying around you would universally be annoying, but considering the number of upvotes and comments arguing that the auto-greetings are not annoying, it's hardly as clear-cut as that. – allquixotic Nov 13 '13 at 1:42
@Shog9 the entire javascript room is convinced that the greetings never offended anyone. Gilles is convinced they scared away a substantial amount of pontential chatters. – Jan Dvorak Nov 13 '13 at 1:43
You admit that what annoys people varies wildly, and still want to come up with a single, fixed definition for it? Seems like a pointless endeavor. – Shog9 Nov 13 '13 at 1:44
BTW: I fully agree with the part about "shouldn't allow you to do anything you couldn't do by yourself", but copy and pasting a message to every new user you see the first time is something that a room could do manually. Here's how: maintain 24/7 shifts of a chat monitor person, who is responsible for noticing when users enter who've never entered before, and @ them with the template greeting. Get enough users cooperating in this, and bam -- you now have a manual system doing what the bot does automatically. So that rule at least does not apply to this situation. – allquixotic Nov 13 '13 at 1:45
I'm intentionally not reading the ridiculously long back-story on this, @all. I'm proposing a very simple, easy to understand rule of thumb based on the question being asked here - if there's a reason why this wouldn't work, then let's hear it. – Shog9 Nov 13 '13 at 1:56
Are you saying you haven't seen this happen in chat, @Jan? I'd be surprised if you haven't. Heck, I've watched folks go into multiple rooms, one after another, and greet everyone there and everyone who joined. Repeatedly. I'm sure you've seen this too. How'd that work out? – Shog9 Nov 13 '13 at 1:59
@Shog9 I have no problem with the third paragraph of your answer. The second paragraph however, is problematic, because how many people have to be annoyed before it's a problem? And does it matter who is annoyed? If a nameless userXXXXXX is annoyed, does that matter as much as a SE employee being annoyed, or a diamond mod? If one user is annoyed and 10,000 find it acceptable, does that one user's annoyance outweigh the fact that the other 10,000 find it acceptable/useful? The 2nd para. of your answer seems to really punt this question to an endless parade of new meta questions about this. – allquixotic Nov 13 '13 at 2:00
Here's my question: Would you consider it annoying if a person (or group of persons) manually copy and paste a templated greeting with a @ping to new users (first-time visitors only!) of an official site chatroom, which gets on average 0 - 10 new uniques per day? Does that fall into the definition of "annoying"? If you ask me, I don't think it does. But my opinion was not solicited before I was asked to remove this feature from my bot. At this point I don't care about the past; I just want to understand what is kosher going forward, so anything you can do to clarify is great! – allquixotic Nov 13 '13 at 2:03
I would find that highly annoying in any chatroom I frequent, @all. Also, "brown". – Shog9 Nov 13 '13 at 2:04
@Shog9 unfortunately, the community moderator who came to enforce the rule didn't seem to care for "our situation", as far as I've observed. – Jan Dvorak Nov 13 '13 at 2:08
...In that case, I'm going to have to officially register my opinion dissenting with the second paragraph of this answer, on the grounds that, because I have personally observed what I perceive as bias by SE community managers regarding the interpretation of the word "annoying", I feel that this is a poor criteria for establishing a set of guidelines. Take that with a grain of salt; I'm just one in a large sea of users. Anybody else care to weigh in? – allquixotic Nov 13 '13 at 2:09
@Shog9 (still haven't read below this so sorry if it is out of context) re: "If only there was a way for room owners to create text that was visible to everyone entering the room, eh" - we tried pinning it for months on end, also putting it in the description and yet no other solution has worked as well as the initial ping when a user enters the room for the first time. well, at least in the JavaScript room that is. – rlemon Nov 13 '13 at 2:10
@Shog9 because they didn't read it outright. They saw the room name, joined, and asked to ask a question. Something we very much dislike in that room. The solution to give them a personal one line notification and leave it at that was made, and honestly it has worked in everyones favor (in the Javascript room). I have never seen anyone complain about it (today excused) and it is often replied to with a "Thanks!" from the user it pinged (again, in the Javascript room). – rlemon Nov 13 '13 at 2:18

I honestly believe that the people making decisions about the bot should be the community of the room.

Stack Overflow has always worked well because of community decisions. This has been working really well for us in the JavaScript chat and we find the bot extremely useful.

Other rooms who use bots need to reason about it and decide on what they like and what they don't. The room's community should be capable of handling that and deciding on usefulness pretty well.

Let's deal with the problem when it becomes a problem

Forming a formal policy on this gets us nothing for now. Rooms that have bots (like JS and Rebol) have utilized them successfully. Rooms who chose not to chose not to.

The system works, there is no abuse going on. I see no reason to fix something that isn't broken.

(flags on the other hand...)

share|improve this answer
It has become a problem, that's why Mad Scientist posted this question. – Gilles Nov 13 '13 at 0:01
@Gilles It has become a problem on once in one room in SuperUser, people have been using bots in multiple rooms for years. It has by the way - been working really well for us in the rooms that have bots like that JS chat, C# chat room, Rebol chat room and so on for years without any issues. It's not a site wide problem, there is no point in meta discussion about it and it was blown out of proportion in superuser too. – Benjamin Gruenbaum Nov 13 '13 at 0:03
@BenjaminGruenbaum Given that the system excludes feedback from dissenting people, your claim that “it has been working really well” is groundless. – Gilles Nov 13 '13 at 0:07
@Gilles given the fact I've been a participating member of a room where it has been working very well for years and where users expressed their opinions about it countless times (and are constantly encouraged to). Given the fact every single user in the JS room (for example) does not think its an issue (and we have a fairly active community). Given the fact you've not been a part of that community and had a chance to see it work well/not as well - I find your claims extremely groundless and somewhat rude. Let us moderate ourselves, stop inventing problems that don't exist for people. – Benjamin Gruenbaum Nov 13 '13 at 0:10
@gilles IIUC, the reason a moderator stepped in in the SU room was that there were too many greetings, not that people reacted badly to beeing greeted – Jan Dvorak Nov 13 '13 at 0:12
@BenjaminGruenbaum The fact that you scared away the people who do think it's an issue doesn't imply that it's not an issue. The problem demonstrably exists for at least three people (balpha, me and Mad Scientist). – Gilles Nov 13 '13 at 0:28
@Gilles did they actually say they don't like being greeted? I don't remember bAlpha having said that – Jan Dvorak Nov 13 '13 at 0:29
@JanDvorak “If someone knows about the bot and talks to it fine, but an auto-greeting is crossing the line.” Where do you get that bit about “too many greetings”? – Gilles Nov 13 '13 at 0:31
Reading the SU Meta post, wasn't it only a problem because it was the main chat room? Not that I have a horse in the race, I'm on a restricted network so I haven't visited chat anywhere for quite some time. – jball Nov 13 '13 at 0:32
@Gilles Over 5000 people joined the room since the feature was implemented. So your argument is that every single person out of the people who got welcomed and didn't like it was terrified and left the room immediately? (Hopefully, not losing too much sleep). You're making a lot of assumptions on a community you've elected not to be a part of. Like yoda said - you have a strong opinion on people happily using a bot in a room you don't visit and would like to impose your (opposite) views on them. Let rooms decide for themselves, it has been working well for years now, don't invent problems. – Benjamin Gruenbaum Nov 13 '13 at 0:36
@jball You are correct that the original problem raised by the StackExchange employee regarding the Root Access chat bot, was indeed related to the fact that it was a chat bot in an "official" chatroom doing the automatic greeting. That said, the decision seems to have later been expanded to include all rooms, or at least all popular rooms. I did read in balpha's answer that a room specifically designed for using the bot would be permitted, but presumably a room about JavaScript isn't about the chat bot. I have no horse in that race either, but I'm just stating what the reasoning process was – allquixotic Nov 13 '13 at 0:37
@allquixotic there was a suggestion to rename the javascript room to not sound official. Not sure how that went. To the drain, I think. – Jan Dvorak Nov 13 '13 at 0:39
If traffic and language popularity push a room into 'official' then we really need those limits defined. When does a room bump into this 'official' status. (This is not as a direct response to anything in this comment thread, this was stated in the chat and if it is going to hold weight I think it also needs to be addressed) – rlemon Nov 13 '13 at 0:42

Exactly the same as the limitations for a regular user (which any chatbot is). If your average Joe with 20 rep can ping newcomers, why can't a bot?

The initial ping (which happens to be the cause of the question) has conclusively been proven to actually reduce the amount of clutter that happens when a user joins. Pinned messages and room descriptions have been tried but did not result in an actual reduction.

Additionally, many new users have commented thanks or the like, expressing gratitude at the utility of such a message.

share|improve this answer
This implies Joe is allowed to ping every newcomer, however. – Jan Dvorak Nov 12 '13 at 23:18
This is exactly how I feel. The bot is not special in any way or form. It is not above being banned or flagged or ignored. If these systems are in place to control us, why can't it be trusted to control the additional tools we've created to help us. *And if Joe wanted to ping everyone and anyone had a problem with that, they can flag joe and have him delt with within the confines of the system we currently use. – rlemon Nov 12 '13 at 23:19
Which one was the control room? – random Nov 12 '13 at 23:23
@random All three options (as well as the control) were performed in the JS room. – SomeKittens Nov 12 '13 at 23:25
@random control room? The Javascript room has been using the bot successfully for months and is its birthplace. The SU room is the one that abused the bot. – Jan Dvorak Nov 12 '13 at 23:25
I wouldn't go as far as "abused" (although I do not know the entire story). However mishandled the bot can, yes, be a bit problematic.. Which is why (if you read the documentation Zirak wrote) room owners and mods have the ability to stop it from responding outright, as well as making it ignore certain users who abuse it on a case by case basis. In the Javascript room this works extremely well, and most everyone who interacts with her (our bot is a female) there enjoys her presence and abilities, including the welcome message. – rlemon Nov 12 '13 at 23:28
@rlemon As a SU chat regular, I wouldn't say it was being abused. Perhaps a bit more for humor/fun when compared to the purposes that the JS bot serves, but I don't think I ever saw it abused. I'm not lurking 24/7, though, just during the day. – Darth Android Nov 12 '13 at 23:45
@JanDvorak The Root Access room was not using the bot any differently than the JS room, except that we tweaked the wording of our greeting message slightly. I have stated on many occasions, and even asked a MSU question about this, I'm willing to change the wording to whatever people want me to change it to, if it's perceived as rude or unclear or scary or malicious or whatever. Other than that, the sheer existence of the auto-greeter is what the SE employees took beef with, not the content of the message, or the way the bot was being used otherwise. – allquixotic Nov 13 '13 at 0:14
@JanDvorak So if your claim is "Root Access abused the bot", and the consequence was "balpha made us get rid of the auto-greeter functionality", then by simple logic, the "abuse" would be the existence of the auto-greeter. The auto-greeter exists in the JS room. Thus, by deduction, JS room abuses the bot in exactly the same way. So, no, I don't think you want to argue along those lines. – allquixotic Nov 13 '13 at 0:15
@allquixotic I was pointing out the difference in size. Probably there's also a difference in purpose - What did the SU bot say? – Jan Dvorak Nov 13 '13 at 0:17
"If your average Joe with 20 rep can ping newcomers, why can't a bot?" for the very same reason that store staff that harass you "can I help you with anything?" as soon as you enter a store, are annoying and not welcoming at all. You may not mind it, but a lot of people do. – Sklivvz Nov 13 '13 at 1:55
That's a terrible example. WalMart, etc have those greeters because psychologically, they make people feel like they belong. – SomeKittens Nov 13 '13 at 1:59
Er, we are not talking about the same thing (not Walmart, where I've never been and neither a restaurant, where there is an expectation). In a chat room there's an expectation of not being called out or being forced to interact publicly. – Sklivvz Nov 13 '13 at 2:04
Wait, are you saying that whistling loudly at the nearest waitress to get her attention isn't standard etiquette where you live, @Jan? – Shog9 Nov 13 '13 at 2:17
@Shog9 negative. We sit patiently at the table until the waiter/waitress comes. If the waiter is willing to yell across the room, I always respond with head gestures. – Jan Dvorak Nov 13 '13 at 2:24

I have noticed that newcomers often do not have a good grasp on chat functionality. They see people pinging back and forth, pick a random person, and ask them a question. Sometimes that leads to users responding bitterly when really, all they were trying to do was see how pinging works.

Excessive pinging by newcomers has definitely been toned down, ever since we've had welcome messages. A welcome message actually shows them that putting @ in front of a name makes a sound. I think what we should really be focusing on, is that welcome messages can easily be changed. We really need a compromise, because tossing out welcome messages altogether is a bad idea.

I propose simply that, welcome messages should be required to be kept friendly and constructive.

share|improve this answer
How do you hope a generic bot to be constructive? It doesn't know anything about the user. – Gilles Nov 13 '13 at 0:08
@Gilles nor does anyone else. And we don't want to wait until they show they didn't read our rules – Jan Dvorak Nov 13 '13 at 0:10
@Gilles We've got 500 char to figure that out, the possibilities are limitless – Shea Nov 13 '13 at 0:22
If Gilles' comment is to be taken to its logical extent, then we should remove the help and FAQ pages from all sites as well, because these are "generic", all-purpose, one-size-fits-all pages that attempt to help users at great length. The sheer number of characters there vastly increases the chances that one phrase or sentence might be interpreted as offensive, condescending, or scary. The humanity! We should eliminate all help text from the site entirely! (No, I do not actually believe that, but reductio ad absurdum is quite nice...) – allquixotic Nov 13 '13 at 0:51
@allquixotic uh, no, the answers (except to canonical questions) should stay (but not the questions, as they're generic) – Jan Dvorak Nov 13 '13 at 0:54
@JanDvorak Did you miss the part where I said "No, I do not actually believe that"? – allquixotic Nov 13 '13 at 1:16
@allquixotic Nope. I did see that part (and I understood even without that part that your message was meant as sarcasm). I was just playing along. I don't actually recommend what I said either. – Jan Dvorak Nov 13 '13 at 1:17
@allquixotic Gille's comment only helps my suggestion. Not only does he have no doubt in the bot's ability to be friendly, but he has also reminded me to include an example. ie for JS: Welcome to JavaScript. I am a community bot (!!/help), I may be able to assist you. You can type !!/listcommands to see what I do. Please read our rules (!!/rules), linked to the left. Whether or not we should only list PG comments, with that command is another question. – Shea Nov 13 '13 at 1:19
@JanDvorak Oh, okay :D Well played, then! You got me! – allquixotic Nov 13 '13 at 1:19
@Shea Well, in balpha's (admittedly fairly dated now) canonical answer talking about bots and allowable behavior, he does state, "Finally, a bot will be held against abuse standards just like any other user." So that would imply that, in several rooms where swearing and offensive languages and/or pictures are common, we either need to (a) clean up the human users' language AND the bots' language; OR (b) allow bots AND human users to use foul language and so forth. To allow it for one and disallow for the other would contradict --- – allquixotic Nov 13 '13 at 1:23
... (continued) balpha's statement that bots will be held against abuse standards just like any other user. Personally, I am completely indifferent as to whether the community would prefer to allow or disallow foul language and other such nasty stuff. I would be perfectly content as a user either way. But I'm looking for consistency in the treatment of bots and users, especially since it was an SE employee who stated that they would be treated "just like any other user". If that is no longer true, then it needs to be clearly stated and clarified that the policy has changed. – allquixotic Nov 13 '13 at 1:23
And once again, I would not be bothered in the least about them deciding to make a policy change on this matter, even if the new policy places restrictions or sanctions on bots that are not present for human users. I just want them to pick a story; stick to it; and make it abundantly clear what the story is. I think much of the ado about this issue today has been chiefly because the people who operated these bots in these ways did not know that the rules had changed, or were changing, until after disciplinary action had already been taken against one bot. – allquixotic Nov 13 '13 at 1:27
@allquixotic This is about the SO community and the rules that it, as a whole, thinks should be set against chatroom bots. Bots are not humans, they are software and should be treated like software. – Shea Nov 13 '13 at 1:29
@Shea I am unable to argue against what you just said, because in the context of my previous comments, that is a valid point. I apologize: I was just clarifying what I understood as the current policy as of balpha's post in 2011 talking about allowable bot behavior. I have no problem with us deciding to single out bots and give them special rules they must follow. It's just that the rules seemed to have changed spontaneously without any prior notification or community discussion. But yeah, the "bots == humans" de facto rule was a bit simplistic; clear bot-specific rules would be better. – allquixotic Nov 13 '13 at 1:38
@Shea thus, this meta post – Jan Dvorak Nov 13 '13 at 2:10

What should be the limits for chat bots interacting with regular users?

Quite a premise isn't it, that the bots need to be limited in the first place? I don't think they do.

In the c# chat room we have multiple bots. They do not use the auto-greet feature, one of them is the Zirak creation CapricaSix.

It is used for utility, mostly as a thesaurus, and no one ever complains about it. How should we limit our thesaurus?

Perhaps it should only do this or that? Perhaps. It was configured to be in line with the culture of the room, just as almost every chat room is guided by its own unique culture. If there was an issue with the bot in the SU room, then the room should deal with it because we sure as heck didn't notice any exploits or annoyances while there was commotion there.

I am sure the SU room is a great place. They had some fun with the bot because bots can be fun when they are first used. Once the novelty of it wears off (and I am sure this post is contributing to that at a rapid pace) then the bot will more than likely calm down.

"Boys will be boys", which is to say, it was just a little bit of fun.

There are more than enough constructs in place to deal with abusive users. The bots are, after all, users. If they are abusive or offensive, then take action against them as you would a user. Click on one of their message's "message actions" arrow, select "flag for moderator" and then explain what is happening.

edit "greetings"

Chat is kind of a free for all in some ways. It is regulated by SE guidelines, but it can get terse at times because those guidelines are rather broad.

There really don't seem to be any direct rules dictating greeting messages. Sometimes I will greet someone who is new to the room, sometimes I will greet someone who is a regular, and sometimes I do nothing. My personal habits are not as predictable because I am a person and they depend on my mood and the other user.

However, a bot is predictable. It will execute as configured, and if it is set to greet every user (not just new users as in Javascript for example - which in my opinion does no harm) then it will greet every user joining the room. Even if they leave, then join, then leave, then join in quick succession.

The configuration of the bot in that scenario is the error, not greetings in general. And in this specific case, I think it was not intentional to spam, abuse the bot, or drive users away. It was just a side affect of product testing. Cool new feature! Turn on all the things!

If you can't handle one ping for joining a room then perhaps chat isn't for you. If your room is being spammed by greetings, then perhaps those greetings should be tampered down. But I do not think that greetings are bad, I personally enjoy joining to a hello, or seeing someone join the room with a greeting that they have arrived.


I had assumed when I first read this post we were talking about generalizations. I read through most of the content here but did not realize there was a lot more to this situation than this post or a broad statement.

With regards to the exact Root Access room, and a claim that it should be ruled with an iron fist because it is part of StackExchange and not a single exchange, I just don't buy it.

If the bot is not spamming, and is only welcoming new users to the chat room, then to me it does not seem to be egregious.

Further, Root Access gets a third of the Lounge's traffic per day, half of Javascript's, and on par with C#'s. That makes it account for 1/6th of the traffic of just those three rooms.

I respect balpha, and if he makes a decision about a room he has direct interest in then I wont argue with that. However, if that room is so important, it seems odd to have users be owners of it, and even more interesting that no moderators are owners.

To me, it is just another room, and they are just trying to help new users stay on topic as is defined in the faq.

Another point to make here is that users do not have to join a room in order to read its contents. Merely clicking on the transcript will allow inspection of the topicality of the room and what the recent messages have been. Joining implies participation.

share|improve this answer
It was the greeting that was the problem. Without that the bot was probably fine. – ChrisF Nov 13 '13 at 0:40
@ChrisF "probably"? How else could the bot be "possibly not fine"? – Jan Dvorak Nov 13 '13 at 0:41
@JanDvorak @ChrisF There was also mention of the !!f***able command, which was promptly removed, as our user base largely agreed that it was superfluous, puerile, and not needed. We inherited this command from Zirak's source tree and never thought to remove it. We cooperated fully with the requests of the SE employees with regards to modifying our bot to meet the standards of what they feel the bot's behavior should be. The bot was then swiftly un-banned. In that sense, the matter was nicely resolved in a civil manner. The bot is back in the room right now. It just doesn't greet you. – allquixotic Nov 13 '13 at 0:43
I can't speak for SE, hence the "probably". I don't want to put words into their mouth, so to speak. – ChrisF Nov 13 '13 at 0:44
@TravisJ Excellent post. I agree with your answer, but unfortunately, I think you are talking past the real issue here, which is the automatic greeting message. It would add to the discussion if you could weigh in (one way or the other, as you see fit) regarding whether these automatic greeting messages should be permitted. – allquixotic Nov 13 '13 at 0:44
@allquixotic - Thanks :) Please see my edit. In my opinion, all forms of spam should be addressed (even if they are in the form of a bot's automatic message). How they are addressed depend on the tools available and the extent of the spam. In this case, it was localized to just one chat room of the entire exchange. In this case, it seems simple to just adjust the bot's configuration. – Travis J Nov 13 '13 at 0:59
@TravisJ I'm very sorry, but I think you are misinformed. First, the chat bot (which I operate) in Root Access (the SU chatroom) was greeting new users exactly once, not greeting them over and over if they leave and join the room. Second, the total number of greetings per day, depending on our popularity, varied from approximately 0 to 10 in any 24 hour period, with occasional spikes when people linked to our chat from popular questions. Please read my question here and balpha's answer, to obtain the factual perspective on the SU bot situation. – allquixotic Nov 13 '13 at 1:04
@allquixotic which is, by no means, excessive – Jan Dvorak Nov 13 '13 at 1:05
Furthermore, I would implicitly interpret a StackExchange employee -- the one who designed the chat system, at that -- having suspended my bot until I removed the automatic greeting, as indicative that there is a "direct rule dictating greeting messages". Well, there may not have been yesterday, but there is now. The rule is, they are not allowed. This entire MSO question is about what users feel should be the rules. What the rules are, is now quite clear from balpha's answer to my MSU post. – allquixotic Nov 13 '13 at 1:05
@allquixotic he directly said the suspension was only because you weren't at hand to chat. Either kill the bot, or don't. – Jan Dvorak Nov 13 '13 at 1:07
@JanDvorak OK, but what would have happened if I were at hand to chat, and then flat-out refused to change the bot's code? The rule is quite clear, regardless of whether or not the bot's owner is at the keys when they notice the bot's behavior: if your bot is automatically greeting users, and the SE community managers are unable to persuade the bot's owner to stop automatically greeting users, then the bot will be suspended until it stops automatically greeting users. The rule, therefore, is fairly unambiguous, no? – allquixotic Nov 13 '13 at 1:11
Also to make an official statement on the command that caused controversy (the age one). As the person who introduced it as a user taught command, which makes me ultimately responsible for it being in the code base in a way.. I am sorry and it is removed from the codebase once Zirak is available to accept the pull request pending on it! It was a mistake and I regret ever introducing it for our room in the first place. However this discussion shouldn't be tainted by a single misguided command. Just remember that everyone. – rlemon Nov 13 '13 at 1:12
@allquixotic I'd be careful about claiming this unambiguous. You don't know what would actually happen. Unfortunately, I have to leave now, but I will rejoin in the morning (Europe). – Jan Dvorak Nov 13 '13 at 1:14
@rlemon Don't worry, I really don't think that SE or anyone here is arguing that the existence of that inappropriate command is a reason for doing away with bots entirely, or not having them broadcast auto-greetings. That was merely an issue that they brought up in passing as a "we're going to have to make a decision on this later if you don't make it easy for us and disable it" type of thing. But nobody wants to leave that command in, so it's really, really a moot point. – allquixotic Nov 13 '13 at 1:14
@allquixotic - You are right, I was not aware of the entire context regarding the bot's actions. I expanded my answer as a result. – Travis J Nov 13 '13 at 2:48

A chat bot must never initiate conversation with an unsuspecting user. When a user joins a chatroom, the expectation is to chat with humans, not to be assaulted by a bot. If the bot takes the initiative, it isn't interaction, it's spam. Interaction with a bot must be opt-in.

It's fine for a chat bot to post spontaneously as long as the rate is reasonable.

It's fine for a chat bot to reply to posts addressed to it. Preferably the bot should only reply when addressed a standard ping like @botuser, but bots reacting to other syntax are ok as long as it's not something that people are likely to type unaware. A bot should generally not reply with more than one message without good reason.

Are auto-greeting bots useful to punch users in the face with chatroom rules? Yes, sure. That doesn't make it appropriate to punch users in the face. (No, not even on Server Fault.) Consider that for every user who is enlightened by the bot (yes, I'm sure you can find one), there are also users who run away in terror. If your chatroom rules don't fit in the room description, consider simplifying them. If your problem is that users don't read the rules, welcome to the real world.

This also applies to humans to some extent, by the way. Being pinged seconds after joining a room is creepy. Sure, joining a room is public and triggers a visual effect for the participants. That doesn't mean that you should gang on the new guy. Greeting regulars who you know don't mind is fine, but please don't scare away newcomers.

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.)

share|improve this answer
"there are also users who run away in terror." -- citation needed – Jan Dvorak Nov 12 '13 at 23:13
@JanDvorak They rarely come back to testify, but it's a common enough sight — user joins, lurks for a minute, then goes away. I have been the one to run away, though fortunately this didn't happen to me in my early days on chat.SE and I've now learned to ignore it when I join a new room. – Gilles Nov 12 '13 at 23:15
So if this is really going to be the case, there should be a limit put in place before I can interact with a new user. I probably also should need a request system to be able to view their public profile. Sarcasm aside, do you not see what restrictions you are leading into. Also, if saying "welcome! <quick tips so you don't get ignored>" is scary, why are they joining the chats in the first place? – rlemon Nov 12 '13 at 23:35
@rlemon It's not what the bot is saying that matters, it's being pounced on the minute you join the room. – Gilles Nov 13 '13 at 0:03
@Gilles it is literally what our bot is saying. Not sure about the SU bot, though – Jan Dvorak Nov 13 '13 at 0:04
As for when should the message be posted - when they ask "may I ask?" it's already too late. – Jan Dvorak Nov 13 '13 at 0:04
@JanDvorak Again: the content of the message from the bot is mostly irrelevant (I've disambiguated my previous comment). – Gilles Nov 13 '13 at 0:05
@JanDvorak I fail to see what's so bad about “may I ask”. It isn't particularly disruptive. This is a chatroom, after all. Phatic expressions are to be expected. – Gilles Nov 13 '13 at 0:06
@Gilles for regular users of a very popular room being asked "can I ask" a few dozen times a day gets pretty frustrating pretty quickly. This way they know to just ask and we don't get annoyed with them. This also means they get a much better chance to have their question respected and taken seriously. – rlemon Nov 13 '13 at 0:08
@Gilles Our room finds it non-constructive. It's tiring to say "yes, you may ask" when they drop silent over and over again. That is why the house rules came to be in the first place - and not only in the Javascript room. – Jan Dvorak Nov 13 '13 at 0:08
@JanDvorak Don't get too lost in the point about asking to ask, though: the fact remains that Gilles (and SE) contend that the message itself, regardless of its contents, should not be there. The text "Please don't ask to ask" (which our bot also said in Root Access before I removed it) is not what people have specifically taken exception to. It's being pinged at all. If we are going to expend our energy debating this, we should debate that point, not the tangential point of the contents of the message. We can always hash that out later on, if we regain permission to auto-greet. – allquixotic Nov 13 '13 at 0:56
@allquixotic I bringed that out because I was defending the point that these messages are actually useful. – Jan Dvorak Nov 13 '13 at 1:03
I second this point. Saying that the message is irrelevant is not really constructive in my opinion. The message carries a lot of weight in whether or not the message (in general) is useful. If the room has an issue and the message helps with that issue, or if the room would like to guide people so that they get better results out of using the chat, what is the issue? Now if we want to discuss what is not acceptable for a bot to say in a welcome message... – rlemon Nov 13 '13 at 1:07
The edit that was just made to this post has suddenly made it extremely constructive, and I've upvoted it in turn. I still don't like the majority of the opinion on this post, but because it's trying to be constructive at the end, it deserves my vote. Thanks Gilles, that's very useful. – allquixotic Nov 13 '13 at 1:30
@allquixotic It's always bothered me a bit that clicking on the big fat link in a room description would be an action that's visible to others. So I've proposed a change to the chat interface so that joining a chat room would be a truly active action. If that change was implemented, I wouldn't object to a greeting bot as a matter of principle. – Gilles Nov 13 '13 at 1:31

A while ago I came up with a list of basic rules for Chatbots operating within the Mos Eisley Chatroom. They seemed to be well received by the community.

  • Bots must be clearly labelled as bots in their chat profile and main site profile (as well as indicating who their owner is)
  • Bot accounts may not be used for upvoting/casting close votes, etc. (e.g. they cannot be used as sockpuppets)
  • Bot accounts that go beserk (spamming or spouting nonsense) will be subject to immediate suspension
  • Bot handlers/owners will be held accountable for their bot's actions.
  • Intentionally interfering with someone else's bot will be frowned upon.
  • Main chatrooms are not a sandbox. If you want to test your bot, go do it somewhere quiet.
  • Automated actions are limited to the chatrooms. Any bot found to be active on the main site will be destroyed.
share|improve this answer
Non of your points address the question Should a chat bot be able to initiate an interaction with a user, or should any interaction be initated by the actual user?. Your list might be a better fit for the CHAT FAQ posts – rene Apr 21 at 17:37
@rene - I would class an unasked-for contact as spamming. – Richard Apr 21 at 17:42
Sure, see how well this list works. I define spamming waaaaaay differently – rene Apr 21 at 17:44
@rene - That is, of course, your privilege. – Richard Apr 21 at 17:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .