-10

I'm a mod on Ask Ubuntu. I've been talking to somebody who wants to run a chat bot on all the SE chat networks. To do this they're aiming for 200 rep with us to unlock the 100-rep association bonus.

Chatbots are fairly well accepted things these days (as long as they behave) but there are a couple of obvious issues with how they're created:

  1. They have to create a new account and grind for rep.
  2. They end up with an account that can do far more than it needs for chatting.

The term "sockpuppet" is really a reflection of an accounts actions and the user here has promised not to vote for themselves, it just seems silly to force wouldbe chatbot operators to go through exactly the same steps. I've got a few mutually exclusive ideas that might make things better:

  • Allow people to have a chatbot subprofile which can only do the things the main account can. Most importantly, it require the somebody to put themselves in the position to sockpuppet.

  • Create a dummy chatbot.stackexchange.com site where profiles (linked to the main profile) have their own chat profile with automatic and appropriate chat-only privileges. I'm not sure if that would actually work though.

  • Allow real accounts to be limited to chat-only privs, whatever their reputation. This would annul moderator fears of bots going wild.

  • Or finally —something I've suggested before— cross-user limits to stop "accidental" sockpuppeting happening.


There are auxiliary conversations happening about whether bots should be allowed to vote on main sites at all. That isn't my main focus but it's certainly relevant in the long run.

One thing I will say on the matter: isn't the point of crowd sourcing to get humans doing things? Inviting AIs (or facsimiles thereof) to do the job seems to counter the system's design.

  • 6
    Why is the 100 rep needed, the owner could create the room? Can't an extra option be: Let SE grant those chatbot accounts the chat privilege and that is it (so no privilege on sites is needed). – rene May 10 '15 at 12:07
  • 10
    This request is a total mess. Title talk about chat bot, while the body talk about sock puppet ring. – Shadow May 10 '15 at 13:37
  • 2
    @ShadowWizard That's because to create a viable chatbot, you currently have to take the same first steps as somebody would if they were creating a sockpuppet. New account, quick grind for starter rep. The suggestions are based on separating the behaviour so you aren't in the position where you could "accidentally" double-vote. – Oli May 10 '15 at 13:41
  • 4
    But chat bot is the least of our concerns - pumping reputation and votes into the system, pushing spam into the Hot Network Questions list - those are the main risks. – Shadow May 10 '15 at 13:42
  • @Oli Feel free to mention my name. It's actually me and my bot ByteBOT this question is about... To clarify: I actually logged myself into my bot account for a while and answered and edited some times to gain those 200 rep (having 132 currently). While being logged in as bot, I have to admit I also did actions not necessary for reputation earning like (up-)voting and flagging. – Byte Commander May 10 '15 at 14:31
  • However, I took care that I did not do anything to posts of my respective other account and that I did not perform the same actions to one post from both accounts (voting, flagging). Could have happened that I answered with one account and commented with the other or things like that, but nothing that would affect any vote score! – Byte Commander May 10 '15 at 14:32
  • 2
    By the way I explicitly state that my two accounts are related and in which way. So why should I try to abuse the system while publicly showing everyone that I own both accounts. Furthermore, my program has no ability to interact with any main site by now. And at the moment, I don't even have an idea how I would have to implement even read-access to the main sites. – Byte Commander May 10 '15 at 15:02
  • Your closing comment has generated a question at Community Building. It is available here: communitybuilding.stackexchange.com/q/1143/78 – Andy May 12 '15 at 21:22
16

No.

It's perfectly fine for a user to have multiple accounts as long as they follow the no-sock-puppetry policy — No interaction between accounts whatsoever, and not using the extra privileges to achieve more than they could with one account (e.g. voting).

Like many users signing up just to post their first question, chat bots accounts earn 200 reputation points for the network-wide bonus. This is acceptable just like you wouldn't ban users from signing up only to ask a question without accepting answers, upvoting them, or giving any response.

And plus, chat bots are people too. This policy change discriminates them.

  • 6
    It's also worth noting that, as far as I know, multiple accounts are acceptable under official SE policy as long as there is no cross-voting involved. – AstroCB May 10 '15 at 13:37
  • lol, the bot conversation you linked is funny :D – nicael May 10 '15 at 16:18
  • 1
    @nicael They turn on the bots once in a while, so you can always hope to visit again later for another new conversation, and I'm honored to lend them my bot account because I support giving bots freedom. You can find their source codes maintained by the SO CVR: github.com/SO-Close-Vote-Reviewers/Gelidus – Unihedron May 10 '15 at 16:28
12

As a user, a moderator, and a chatbot owner: Heck no.

They've explicitly said they won't abuse it but I'd rather not need to have to trust their word.

I can't think of any user running a chatbot on the network that I wouldn't trust*. Also, I really don't need to trust them - the serial-voting mechanisms moderators have are fairly good at catching these cases. And the only ones that would have a decent idea of how to get around them - moderators - should know better. And they do.

If I, as a prospective bot owner, want to rep up one of my accounts to 200 on a site (which is fairly easy, really), awesome. That's good content for a site. I myself haven't done this - preferring to add them to the access list instead. It's a moderator privilege to add 1-rep users to the list, so for some people it's easier to just get the rep**.

They've already earnt the upvote priv with us and they've "accidentally" voted 28 times.

A few cases here:

  • If there's crossvoting happening:

    Handle it just like any other case of serial voting. Delete sock, suspend user.

  • If there's concentrated voting happening - I.e. the main account and chatbot account voting on the same thing:

    This is also not good, the policy is that you really shouldn't use an account to do more (more voting, more reviewing, etc) than you can with your main account. This is not a chatbot problem, in this case a chatbot is involved.

It's not worth complicating the system to allow for the dozens of chatbot accounts we have. Use your normal tools, if there's bad things happening address it. Whether there's a bot involved or not.

And hey, if I can make something to use those privileges for good, why not?


*well, except maybe me
**or ping me, if you'd rather

11

Much like Mad Scientist's answer, I see two separate issues here.

  • "Accidental voting" / Sock puppets
  • Privileges granted to the secondary account

Sock puppeting of the bot accounts

If a bot owner if using their bot to sock puppet - voting to their own answers, or 'doubling' their votes toward others - then this is a problem. But, it's a problem that already exists regardless of the second account being an automated bot. As a moderator, you can deal with voting fraud already. This situation is no different. Follow your standard process.

If, however, the bot account gained it's reputation legitimately there is no problem (in my opinion) of having those permissions. I don't believe the fact that it is a "bot" vs a person has any relevance. The accounts should be moderated the same way. If the "bot" does something a human isn't allowed to do, whack 'em with a mod stick. If the bot behaves and contributes a bit to gain the reputation the bot owner wants, what is the issue?

Chat privileges

It's fairly easy to build a chat bot on Stack Exchange, despite there not being an API associated with chat. You also require the chat privilege on the associated site, if you want to talk. If you want to chat anywhere, you need to earn this reputation in three locations - MSE, SO, and any other site in the network. This is a few upvotes.

The "Create Chatroom" privileges isn't really needed to talk. If the bot isn't actually going to create rooms, it may be that the owner wants the ability to upload pictures.

What other new privileges in chat do I get?

At 100 reputation you also get access to the "Upload image" button in chat.

If the owner does need a room for the bot, I think they (on their main account) should be the one creating the room.


I differ from Mad Scientist's answer in regard to one statement that was made in that answer:

I'm not sure if there is a rule, but I'd consider any voting done programmatically by a bot and not controlled by a human to be abusive and reason for removal of the bot account.

I disagree with this. The Stack Exchange API has routes to cast many types of votes on questions, answers and comments.

Answers

Questions

Comments

ALL of these require authorization. All of these are public methods in the API. Simply utilizing the API to cast a vote does not make the votes "abusive". Votes taken via the API should be treated as any other vote. If they become abusive/sock puppety, then the moderator needs to step in and deal with the situation as they would against any other account. Part of dealing with sock puppets also involves dealing with the puppet owners. The same should apply here.

  • 2
    Case point of an automated process flagging stuff Can a machine be taught to flag comments automatically? – Braiam May 10 '15 at 13:38
  • 1
    Full disclosure, that bot is mine @Braiam. It is not a chat bot, though. It runs under my account. That's why I didn't mention it. It's not a separate user in the system. – Andy May 10 '15 at 13:39
  • Well, is something programatically doing stuff in the system, stuff that was taugh by an human ;). – Braiam May 10 '15 at 13:40
  • I couldn't disagree more about bot voting. SE is all about crowd-sourcing. Bots remove human variance. If you had 50 copies of the same bot software mowing through the review queue, you might as well just replace the whole review process with a single bot running at SE HQ. – Oli May 10 '15 at 14:05
7

There are two completely different issues here, a bot having reputation and the associated privileges, and a bot voting in a problematic fashion.

I don't see any issue with the first part, the bot account earned the reputation the regular way, I assume. The owner used the second account to earn reputation, and as long as they didn't cheat in any way, this is perfectly acceptable. Giving a bot that access is not an issue either, though the user is still responsible for any actions the bot performs. If the bot messes up, the bot account can get suspended or deleted as a consequence.

In cases where any damage was caused accidentally, that's probably all that might happen. If a user is negligent or reckless with his bot, the main user might feel some consequences as well.

The voting issue is a separate concern. I'm not sure if there is a rule, but I'd consider any voting done programmatically by a bot and not controlled by a human to be abusive and reason for removal of the bot account.

The bot account itself is just a sock puppet, it's bound by all the rules around sock puppets. So it is not necessarily against the rules to use that account to vote (controlled by an actual human), but it is likely a bad idea and could easily violate the rules e.g. by voting on the same post with the bot and the main account.

I don't think the bot aspect changes anything here, those accounts are simply sock puppets. We already have rules for them, and tools to investigate and deal with them if they violate the rules.

  • "I'd consider any voting done programmatically by a bot to be abusive and reason for removal of the bot account. " - I'm going to have to disagree with this. There are explicit routes in the API to allow programmatic voting. Just because a bot casts a vote does not make it abusive. – Andy May 10 '15 at 12:54
  • 1
    @Andy That has to be in the API to allow e.g. users with the Android app to vote. It doesn't mean that a bot voting is allowed. – Mad Scientist May 10 '15 at 12:56
  • 1
    It's listed in the public documentation. Why shouldn't it be allowed? I don't think this is a "bot voting" is the problem thing. I think it's more of a "don't sock puppet" problem. Users can have multiple accounts, as long as those accounts don't interact with one another. If an account builds up reputation legitimately, why restrict how they vote? – Andy May 10 '15 at 13:00
  • 1
    Maybe I misread the question but I don't read that the bot or is puppeteers were involved in voting fraud. The issue at hand is: Is gaining 200 rep on one site (where the puppeteer has enough knowledge to get that rep for the bot) a problem if that gives the bot network-wide privileges. – rene May 10 '15 at 13:10
  • What about user -1 and all their votes? ;-) – quid May 10 '15 at 13:21
  • Case point of an automated process flagging comments Can a machine be taught to flag comments automatically? – Braiam May 10 '15 at 13:38
0

I, for one, think this may be a good idea.

If, and only if, the main account already has chat privileges.

You must log in to answer this question.

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