Wikipedia talk:Sockpuppetry/Proposal

I fully support this proposal. There are some cases where the checkuser facilty or other technical methods are unable to give an answer one way or the other, or gives an answer that is strongly suggestive but not conclusive. To deal with these instances it might be worth noting the established principles used by the Arbitration committee:

For the purpose of dispute resolution when there is uncertainty whether a party is one user with sockpuppets or several users with similar editing habits they may be treated as one user with sockpuppets. (Wikipedia:Arbitration policy/Precedents#Sockpuppets (and related principles)

Accounts designed to impersonate other contributors are not permitted (see Wikipedia:Blocking policy#Impersonation). Accounts designed to impersonate may be immediately blocked indefinitely by any administrator. (Wikipedia:Arbitration policy/Precedents#Impersonation_accounts)

[…]Abuse of sockpuppet accounts, such as using them to evade blocks and bans, make personal attacks or reverts, or vandalize, is strictly forbidden. (Wikipedia:Arbitration policy/Precedents#Sockpuppets (and related principles))

Other useful reading:

Nice idea, but fairly hopeless IMO. If I were a developer, continuously checking whether accounts X and Y were connected because the Wikipedia community wants to root out sock puppeteers would not be high on my priority list, and moreover, it would get annoying. There is occasionally a way to conclusively prove the sock puppet link, but more often there is not, and we're pissing in the wind. The damage that could be done to innocents is likewise considerable—I vividly recall the ABCD adminship fiasco. Keep sock puppet checks reserved for ArbCom cases and serious suspicion; don't bother trying to weed out the childish variant at the root. It's like vandalism: there's just no way to get rid of it permanently, and you're probably better off just undoing it over and over again than to think of clever solutions that will limit the damage, because clever solutions usually aren't, or trivial to subvert. The majority of these accounts is annoying but no threat whatsoever: accounts like User:Smell Etitis don't live long, and don't get to do any permanent damage. Even imposters are quickly caught on behaviour. I also don't think we have many registered users who are perfect little angels except for using sock puppets to abuse people. True colors and all that. JRM · Talk 16:05, 2005 May 11 (UTC)

Three things... first, developers aren't the only people who can sockcheck; in fact, it is usually done by David Gerard and/or Tim Starling, as the devs have other things on their minds.

Second, the proposal doesn't state that every sock must be checked, it merely allows for socks to be checked.

And third, the ABCD thing was eventually resolved, and would have been resolved a lot faster if sock checking for certain reasons had been allowed in the first place.

Oh, and finally, if as you say we don't have many users who do sockpuppet abuse, then that means it's only two or three people, and this would allow us to get rid of them without disturbing the pretty little angles. HTH! R adiant _* 17:41, May 11, 2005 (UTC)

Then I don't get the proposal, really. Why don't we contact Tim Starling or David Gerard when appropriate, then? Is this proposal just "lift the restriction that only the ArbCom can do sockpuppet checks"? Are sockchecks verboten right now? JRM · Talk 17:45, 2005 May 11 (UTC)

The proposal is twofold. First, present policy is unclear on when sockchecking is or is not appropriate, and thus understandably Tim and David don't do it all that often, except in ArbCom cases for instance. Second, the aim of the proposal is to make hold a user responsible for the actions of his socks. Thus, if User:Foo is angry at User:Bar, and creates User:Barr to harass the latter, then I feel that it should be possible for User:Foo to be admonished and/or temporarily blocked for vandalism. R adiant _* 18:18, May 11, 2005 (UTC)
Ah, I see. I'm being slow on the uptake here because this all seems perfectly obvious to me. Wikipedia works according to the "one person" principle, so a user is responsible for all their accounts. In that sense, I should think that if account X is blocked, and X belongs to a set of accounts operated by one person (and we know this for certain), then all accounts from that set are blocked likewise— or more accurately, all abusive sockpuppet accounts are banned indefinitely regardless of severity of the action, and the non-abusive accounts (usually just one) will be RFAr'ed, blocked, or banned, according to severity. I see this isn't quite instruction creep, because it would take some people by surprise, and others prefer a needlepoint definition of the blocking policy even in cases of clear bad faith.

I'm still sceptical about the workload this might involve, but as long as Tim/David/whoever think they can hack it, and people really want to bother with this sort of check as opposed to just "fire and forget", I say power to 'em. I'm afraid we're just going to breed more resilient socks, but that's just my opinion. JRM · Talk 18:32, 2005 May 11 (UTC)

I'm generally in favor of this proposal. My first reaction was: "Support. I thought this was already official policy. --~~~~". I'm less worried about minor vandalism and abusive sockpuppets, which are relatively easy to deal with. What I do consider very problematic, though, are deceptive sockpuppet accounts that have been used for astroturfing, ballot stuffing, and other dishonest purposes.

A recent example is the person behind User:Amerinese (see WP:ANI#Amerinese/BlueSunRed/160.39.195.88 sockpuppetry again), who voted at least three times in several recent polls (e.g. as User:Amerinese, User:BlueSunRed, and User:160.39.195.88 in the poll on the scope of the "Taiwan" article). It is conceivable, perhaps even likely, that none of these accounts are the main account of this person: invariably, the first edits by these accounts are all fairly recent, occurring after the start of the debate that led to the poll. Which leads one to suspect that an otherwise reasonable contributor may have created them specifically in order to influence this debate while keeping his or her "main" account clean. As was pointed out above, policies generally apply per person, not per account (this is made very explicit for WP:3RR and ArbCom rulings), and in the case of deceptive sockpuppetry (or persistent abusive sockpuppetry) an RfAr seems appropriate.

On the other hand, as JRM points out, identity paranoia should be considered harmful. If a decent contributor opens a secondary account to go on a vandalism spree on one occasion, that sockpuppet account will be dealt with like any other vandal and the whole matter should be considered too trivial to warrant any further attention. --MarkSweep 23:57, 11 May 2005 (UTC)[reply]

Feedback[edit]

This is not a vote. Instead, please discuss and give feedback, and indicate if you agree that this is a problem, and what should be done to solve it. Amendments, wording clarifications and alternate proposals are most welcome. Any proposal should be worded such that good-faith Wikipedians have nothing to fear even if anonymous or having multiple accounts; only vandals or impersonators should be dealt with.

R adiant _* 13:12, May 11, 2005 (UTC)

I think this is a great idea. However, you have missed a class of sockpuppets I have often come into contact with; sockpuppets created for the purpose of reverting/edit-warring. Shouldn't they be added? Jayjg ^(talk) 15:55, 11 May 2005 (UTC)[reply]

Good idea. Please add some examples. R adiant _* 17:36, May 11, 2005 (UTC)

Sorry for the delay in responding. Please note, for example, these recent blockings of sockuppets by [1]. If you look at the edit histories for these sockpuppets, you will see that they were created for the purpose of edit-warring, abuse, or vote subverting. Jayjg ^(talk) 15:10, 16 May 2005 (UTC)[reply]

I agree. See also my comments on the talk page. Thryduulf 16:10, 11 May 2005 (UTC)[reply]

In practice, obvious sockpuppets used for revert warring are usually treated as being the same as the original account. This would simply formalize the practice. --Carnildo 18:43, 11 May 2005 (UTC)[reply]

Yes, but in practice, people who create a sock like Smell Etitis walk away unpunished, because they can hide behind anonymity. Even if they do it dozens of times. Is that desirable? R adiant _* 19:20, May 11, 2005 (UTC)

Is it possible to reliably tell exactly who's a sockpuppet? IP checks aren't always reliable, and there are plenty of gray areas in similar behavior (for example, many people thought User:Klonimus was one, but he was not). Meelar (talk) 18:48, May 11, 2005 (UTC)

To my knowledge, a true sockcheck was never run on Klonimus. It was merely established that he was not a sock of GRider. Anyway, this isn't about him - even if he were a sock, he's not disruptive or impersonating or anything.
To answer your question, in some cases it is reliable to establish sockpuppetry. In those cases where it is not, we should assume good faith and drop the accusation. R adiant _* 19:20, May 11, 2005 (UTC)

I broadly agree with this proposal. It could be phrased a little better. What exactly does it mean to verify that an account belongs to the same person? I know that there are data (such as IP addresses of logged-in user traffic) accessible to some administrators but not to ordinary users. What other criteria are eligible? --FOo 02:23, 13 May 2005 (UTC)[reply]

You'd have to ask David and/or Tim exactly how this process works. There is an existing process that in a significant amount of cases can prove-beyond-reasonable-doubt that a certain person is or is not a sockpuppet. There exists a gray area in which proof is uncertain - and people are obviously to be assumed innocent if not proven guilty. R adiant _* 07:38, May 13, 2005 (UTC)

A rephrasing possibility:

Whenever an account is indefinitely blocked for vandalism, impersonation, or other offenses in the blocking policy, editors affected by the misbehavior may request that it be investigated as a possible sock puppet of an existing user. [Insert role name here] shall do so and report on the evidence. The administrator who blocked the sock-puppet account may then apply the same penalty to the original account.

It seems important to lay out who can call for a sock puppet investigation (should it be any editor? The administrator doing the blocking? Only the people who are capable of investigating?), who shall get to see the evidence, and who shall make the judgment of what to do with the original account. I'm not purporting to answer these questions -- just to raise them. --FOo 16:55, 13 May 2005 (UTC)[reply]

1. It seems only fair that anyone can ask for a sockcheck (after all, what's to stop them from asking?) However, the Sockcheckers are not obliged to grant such requests; this is intended as an possible measure against severe offenders, not for spurious name-calling and wild accusations.
2. By current MediaWiki implementation, sockchecks are logged, so people can see if they're being investigated (which seems only fair).
3. Finally, who shall make the judgment - well, considering this is a form of vandalism, we already have some rules for that, and I suppose any admin could block the offender for a couple of days for vandalism. Heavy or long-time punishment should only be decided by the ArbCom. That's my $2, comments welcome. R adiant _* 17:34, May 13, 2005 (UTC)

Although sockchecks are logged, that log is only available to people who have access to the CheckUser feature in the first place (currently Tim Starling and David Gerard), so it is not true that users would know they are being investigated. Please see meta:CheckUser for more details on this feature. I'd rather the ArbCom decided when to use it rather than letting any affected user request such a check. Angela . 04:10, May 16, 2005 (UTC)

Thanks for the pointers. The feature could be modified to place a message on the user's talk page -- of course, people with access could easily remove that message, but they shouldn't have any motivation to do so. On the other hand, if one point of checking for sock puppets is precisely to deny abusers further opportunities to abuse, then alerting them could set them off and lead to greater disruption.

I don't think that the ArbCom needs to be involved in requesting a check on a suspected sock puppet. The ArbCom is not involved in dealing with pure vandalism today -- administrators are fully empowered to block those responsible for it. The decision to investigate needs to be reviewable but it doesn't need to be delayed for review. --FOo 04:34, 16 May 2005 (UTC)[reply]

Looks good in the current incarnation. --Carnildo 20:21, 19 May 2005 (UTC)[reply]

Why is this proposal necessary?[edit]

It is my understanding that users and not accounts *are* held responsible for their actions. The only place I've ever seen the contrary assumed is in this proposal. I'm sure if you take a look at arbcom decisions, administrator and editor actions in response to known or suspected sock puppets, there is never even the breath of suggestion that (to give a concrete example) someone using a sock should be permitted to perform more than three reverts in a twenty-four hour period because the sock reverts count as a different account being used.

As a practical matter where a known sock is involved in abuse I normally block all the socks for an offence by any one of them. --Tony Sidaway|Talk 12:37, 19 May 2005 (UTC)[reply]

Socks are generally blocked, but their owner is not generally held responsible. Check the 20+ socks in the table above; all of them are blocked, but for none of them was an effort made to find out which existing user was responsible. If User:X creates "User:Y is a jerk", then proceeds to vandalize Y's pages from that sock, I believe X should be warned or blocked for vandalism. That is not presently the case. R adiant _* 13:10, May 19, 2005 (UTC)

What if it does turn out that the IP address of an abusive sockpuppet is identical to the account of another contributor? That does not neccesarily mean that they are the same person. There are a number of shared IPs out there. Can this be a problem? Sjakkalle 13:46, 19 May 2005 (UTC)[reply]

Fortunately IP addresses alone aren't the only evidence used in determining whether an account is a sock puppet. Shared proxies should be identified in the normal way. There will undoubtedly be cases where a parent is blocked because he's using NAT to share a single internet connection and his kids go vandalizing Wikipedia, but that already happens. We tend to deal with such cases individually using email. --Tony Sidaway|Talk 13:53, 19 May 2005 (UTC)[reply]

I think this policy is common-sense, assuming issues of verifiability are addressed. I'm not sure how much of an impact it'll have--Willy on Wheels is (I hope) not a regular Wikipedian. I think this policy pretty much exists already--people are expected to take responsibility for all their actions, including those done through socks. The problem you're claiming seems to be that there's no way to enforce this. I would say the arbcom can handle these cases. I think the idea behind this policy already exists. If it doesn't, I would support it. Meelar (talk) 13:57, May 19, 2005 (UTC)

This policy is pointless. If we have any policy that currently (mistakenly) implies that accounts rather than users are the basic unit of responsibility, then the wording of those policies should be amended. No need for a separate policy on the matter. — Chameleon 14:04, 19 May 2005 (UTC)[reply]

It is very well that people think that this is already policy, but the point is that it isn't. Presently, people are NOT held accountable for the actions of their sockpuppets. Even if you think they are. If you think they SHOULD be, I wholeheartedly agree, that is the point here. R adiant _* 14:15, May 19, 2005 (UTC)

I've given you a concrete example of where abusive sock puppeting results in penalization of the owner. The list above seems to be a set of sock puppets that may have otherwise unsuspected "respectable" Wikipedia fronts. If and when these relationships are discovered the owners are also blocked as long as the abuse continues. It is impossible and even if possible it would be intrusive, to test all abusive sock puppets against every single editor. We rely on commonsense to identify possible socks. --Tony Sidaway|Talk 14:24, 19 May 2005 (UTC)[reply]

Absolutely true. But nobody's calling for a check on all sockpuppets. It should merely be possible. If someone were to create User:Radiant is an idiot I for one would like to see the offender admonished. Radiant_* 14:42, May 19, 2005 (UTC)
- They would be, through existing methods of dispute resolution. -- Netoholic @ 14:50, 2005 May 19 (UTC)
  - And what is done if a banned user is found using sockpuppets? As far as I know, their ban is not reset, unless the ArbCom decision specifically mentioned this as an outcome. Jayjg ^(talk) 21:38, 19 May 2005 (UTC)[reply]

Let's get to the debate about the formulation of this policy later. As I see it, the first question is: Does anyone object to the intent of the proposed policy? Note that if you think that the intent to hold users accountable is already captured by existing policies, then the only disagreement is about whether a separate policy is needed, or the existing policy should be clarified, or nothing needs to be done at all. That's what I mean by formulation. But regarding intent, does anyone think the proposed policy is actually a bad idea and should not become policy and/or be removed from existing policies? I for one agree with the policy's intent. --MarkSweep 19:57, 20 May 2005 (UTC)[reply]

I agree in principle. Let me see if I understand: If a user account is also used as a sock-puppet in the above methods, they would be forwarded to the two authorities for some kind of punishment. What could be wrong with that? - Tεx τ urε 22:41, 24 May 2005 (UTC)[reply]

I definitely have no quarrel with teh intent of hte proposal. I strongly prefer community solutions over technical ones, however. Tedernst 08:28, 27 October 2005 (UTC)[reply]