Wikipedia talk:CheckUser/Archive 5

I found J. delanoy's initial explanation on the arbitration committee noticeboard talk to be a well thought out, detailed explanation of their actions consistent with maintaining the privacy policies of the project. While I have some reservations about the use of checkuser, these originate much more from the documented fail in blocking Int21h than the MF / George incident. I choose not to raise my concerns on AC/N as I consider it secondary to the primary issues in that case.

I was surprised to come to the realization from Risker's comment on WTT's talk page that CU routinely conduct off wiki operations; my first thought is she was wrong on policy as Handling suspected sock puppets encourages users to request investigations on-wiki. A more careful review of applicable policy pages had me come to the realization I was wrong; secret CU operations are within policy. That doesn't mean it's a good idea.

The follow up nitpicking posted in reply to J.delanoy's explanation was admittedly excessive, but J.delanoy comparing themself and/or the checkuser community to quantum physicists is insulting, arrogant, and shows a real of lack of understanding quantum physicists. To wit:

• We don't believe quantum physics because the physicists have degrees or make convincing speeches, but because repeated and reproducible experimental results validate the findings.
• Quantum physicists don't keep secrets, they publish results.
• Quantum physicists are wrong but welcome review and correction:
  • Einstein famously added Cosmological constant to his relativity theory, later called in "his greatest blunder," but more recent physicists have revived it.
  • Nature published a note announcing tachyon neutrinos -- later scrutiny found the result erroneous.

The checkuser mystique appears to be based on:

• We know what we're doing because we've done it thousands of times and learned from other checkusers.
• If we decide on behavior evidence to do a check user and get the same IP's (and possibly http protocol strings from client browsers) that proves we were right.
• If we decide on behavior evidence to do a check user and don't get any technical evidence, that doesn't mean we were wrong.
• We don't act alone, we have other check users verify our results.

If I'm reading ListUsers correctly there are only 39 CUs out of a population of 136,201 ({{NUMBEROFACTIVEUSERS}}); that's 0.0286 % of the wikipedia community. If there was ever a recipe for Confirmation bias this seems to be it. Given that years of fulltime scholarship have unable to resolve the Shakespeare authorship question, what makes ya'll think you can reliably tell if two users are the same?

Some concrete questions:

• Has there ever been any rigorous studies on the accuracy of CU? (You know, controlled experiments, control groups, that sort of thing?)
• How many CUs were performed based on behavioral evidence in 2012 (or other reasonably large time period). Of those checks, how many found reasonable technical confirmation the two accounts were the same editor? NE Ent

NE Ent, I'm not the party to which you aim your questions, but I'd like to weigh in here. First, to answer directly your first question: I'm unware of any rigorous studies on the accuracy of CU, though I'd very much like to change that.

Second, as to the issue of confirmation bias, remember that there are some of us who come from outside that group (for instance, my staff has access to the checkuser tool and a limited assignment for use of it) who routinely are asked to weigh in for second opinions or to audit results. The stewards also have access to the tool, though they don't use it on the English Wikipedia. The Ombudsmen commission exists to investigate misuse or misinterpretation of the tool as well. As an outside set of eyes and speaking only anecdotally, I can tell you that I very rarely disagree with the team that's here as to the results of their checks. When I do, it's usually a situation where people of good faith can disagree because the evidence is vague. So I don't think there are any really widespread situations of confirmation bias here, but - as I mentioned - I'd love to see a controlled, blind study conducted. That would be a monster to set up, and not something that I have the time or the budget to undertake, but I would absolutely devote some limited resources towards helping a suitable outside party do that. (By way of explanation, my previous job was working with a survey/research design company that did this sort of thing. Obviously, we wouldn't use them for this - they have no technical expertise, and there are huge COI issues - but I mention that so that you know I'm speaking from a place where I actually have some experience designing and implementing that type of study).

As for your second question, I'm in no position to answer it, and I rather doubt that full enough records were kept to answer it in a retrospective question. But I could be wrong. :) Philippe Beaudette, Wikimedia Foundation (talk) 23:54, 16 March 2013 (UTC)

It would take a lot of manpower to come up with this statistic, but you can get an overall sense by going through the WP:SPI archives. --Rschen7754 23:57, 16 March 2013 (UTC)

The quote from Risker indicates SPI initiated checks are a minority so that doesn't seem the SPI archives are a valid representative sample. NE Ent 02:24, 17 March 2013 (UTC)

The two problems with public checks are a) they tell all the sockers "oh look, this is what we're checking, so you just need to avoid x and you can get away with anything!" and b) it would provide an awful lot of overhead. One case that comes to mind is Wikipedia:Sockpuppet investigations/D62943/Archive. When all was said and done there were over 200 accounts and it's much quicker to just block them all. (And no, there is absolutely no reason why someone needs 200 accounts). As a SPI clerk, we come across such farms on a regular basis, and we don't tag those accounts unless we are feeling a lot of editcountitis. I can think of one known crosswiki socker whose name I am reluctant to mention because I know that it will trigger another rash of socks which will need to be found and locked, and which have hit another wiki that I hold adminship on, and which have started to be locked by stewards due to the serious abuse. That is why public checks are bad. --Rschen7754 02:35, 17 March 2013 (UTC)

I've never seen the need to tag accounts with badges of shame. Any wiki-saavy editor reviewing the contributions of another editor will use the "User contributions" option, which clearly shows the block reason of currently blocked editors. Private checks are bad because it's contrary to the transparency necessary for self governing communities.^[1] NE Ent 15:24, 19 March 2013 (UTC)

(unindent) SPI initiated checks are not a minority. This is a small sample size, so the inference one can make from it is limited, but let's look at the last 50 checks. 9 of the last 50 checks were related to account creation requests from the account creation tool. 3 were related to the handling of an unblock request made on a user's talk page. Almost all of the other 38 were either directly or indirectly related to SPI cases; some were direct investigations into currently open cases, others were follow-ups of older cases (for example, to check for new sleeper accounts). There were also one or two checks that were clearly run by a second checkuser who was double-checking the results of the first checkuser (by request of the first). A more detailed study into this would be very time consuming, but this hopefully gives some indication. --(ʞɿɐʇ) ɐuɐʞsǝp 14:54, 17 March 2013 (UTC)

Thank you for the information. The aggregate response of Risker, Rschen, and Deskana are not forming a coherent picture of the relative quantity of privately vs publicly initiated checks and, unfortunately, it appears no one is tracking the rate at which user checks find evidence of sockpuppetry. This is disappointing if this is true -- it appears not only is the practice of scanning accounts based on behavioral unvalidated in any formal manner, but the practice does not appear to be scrutinized in any informal manner. My understanding is that if I want to pursue this further I'd have to assemble my own data by reviewing public SPI activity, which is not something I currently have wikitime to do. NE Ent 15:24, 19 March 2013 (UTC)

The Arbitration Committee is seeking to appoint three non-arbitrator members to the Audit Subcommittee ("AUSC"). The Committee is comprised of six members and is tasked with investigations concerning the use of CheckUser and Oversight privileges on the English Wikipedia. The AUSC also monitors CheckUser and Oversight activity and use of the applicable tools. The current non-arbitrator members are Avraham, MBisanz, and Ponyo, whose terms were to expire on February 28 but were extended with their agreement until April 30 by the Committee.

Matters brought before the subcommittee may be time-sensitive and subcommittee members should be prepared and available to discuss cases promptly so they may be resolved in a timely manner. Sitting subcommittee members are expected to actively participate in AUSC proceedings and may be replaced should they become inactive. All subcommittee members are given both CheckUser and Oversight access but are expected to not make regular use of them unless needed. They are subject to the relevant local and global policies and guidelines concerning CheckUser and Oversight.

If you think you may be suitably qualified, please email arbcom-en-clists.wikimedia.org to start the application procedure for an appointment ending 30 June 2014. The application period will close at 23:59, 1 April 2013 (UTC). Further information is also available here.

For the Arbitration Committee,
NW (Talk) 18:16, 16 March 2013 (UTC)

Discuss this

(Discussion arose on AN under "User:Evangp unblock request" , moved here)

Evangp (talk · contribs) was blocked for socking in September 2010. He used socks; the last was 6 months ago, Buttchunker (talk · contribs). There's no indication of any activity from him since then.

DeltaQuad performed a checkuser .

In the procedural policy, WP:FISHING is defined as, "to check an account where there is no credible evidence to suspect sockpuppetry. Checks are inappropriate unless there is evidence suggesting abusive sock-puppetry."

As CU can only see back a few months, and there is no indication of any activity since 6 months ago, this seems to fit the definition of fishing. 88.104.27.2 (talk) 16:04, 20 March 2013 (UTC)

• Checkuser policy is open to interpretation, and deliberately so. Him having used sockpuppets in the past is certainly credible evidence of sockpuppetry. It is current practice to occasionally check blocked sockmasters to see if they have created any sleeper accounts to vandalise with. Checking a former sockmaster to see if he's still socking, when his unblock request depends pivotally on him not having used sockpuppets recently, is not fishing. Additionally, see the quote from the global checkuser policy below.

The tool is to be used to fight vandalism, to check for sockpuppet abuse, and to limit disruption of the project. It must be used only to prevent damage to any of Wikimedia projects.

— m:CheckUser policy

This check definitely falls under the clause of preventing damage to Wikimedia projects, as unblocking someone who is actively using sockpuppets could be very damaging to this project. If my response does not satisfy you, then I recommend contacting the Audit Subcommittee so they can verify this for you. --(ʞɿɐʇ) ɐuɐʞsǝp 16:42, 20 March 2013 (UTC)

I guess it comes down to interpretation of "no credible evidence to suspect sockpuppetry" then. Maybe the policy should be clearer on that point, or maybe it's just me. Possibly the 'fishing' section should specifically exclude people with a history of socking. 88.104.27.2 (talk) 16:47, 20 March 2013 (UTC)

It's pretty vague on meta, too - "There must be a valid reason to check a user". To you, someone socking 6 months ago is enough; I wouldn't have thought it was. I'm sure some could interpret just 'using an IP address and knowing what they're doing on the wiki", ie me, now is a valid reason to check a user too? The current policy, and interpretations of it, means a CU can pretty much check anyone, any time? 88.104.27.2 (talk) 16:51, 20 March 2013 (UTC)

I actually pointed out that specific sentence to the legal team recently, as it's tautological and totally useless. Regarding the checks, I'm pretty sure you know that is a misrepresentation of the way things are done. If you do know this, then your argument is a straw man and not worth responding to. If you do not know this, then you should definitely be reading up more on our practices and standards before launching discussions here. --(ʞɿɐʇ) ɐuɐʞsǝp 16:59, 20 March 2013 (UTC)

And also, it's actually a bit pointless commenting here too, as nobody here actually has the authority to change anything, or punish people for misusing the tool. The appropriate avenue for you to pursue is contacting the Audit Subcommittee. --(ʞɿɐʇ) ɐuɐʞsǝp 17:02, 20 March 2013 (UTC)

I asked here, because I wanted to try to understand if it fitted the policy as 'fishing' - it seems to to me, but I wondered if others did. 88.104.27.2 (talk) 20:03, 22 March 2013 (UTC)

You're more than welcome to do so. My apologies, I should have clarified; our debate was pointless, but your enquiry was not. --(ʞɿɐʇ) ɐuɐʞsǝp 17:14, 25 March 2013 (UTC)

This is a reminder that the application period for the three non-arbitrator seats on the Audit Subcommittee will close at 23:59, 1 April 2013 (UTC), less than 36 hours from now.

The Audit Subcommittee ("AUSC") is comprised of six members and is tasked with investigations concerning the use of CheckUser and Oversight privileges on the English Wikipedia. The AUSC also monitors CheckUser and Oversight activity and use of the applicable tools. The current non-arbitrator members are Avraham, MBisanz, and Ponyo, whose terms were to expire on February 28 but were extended with their agreement until April 30 by the Committee.

Matters brought before the subcommittee may be time-sensitive and subcommittee members should be prepared and available to discuss cases promptly so they may be resolved in a timely manner. Sitting subcommittee members are expected to actively participate in AUSC proceedings and may be replaced should they become inactive. All subcommittee members are given both CheckUser and Oversight access but are expected to not make regular use of them unless needed. They are subject to the relevant local and global policies and guidelines concerning CheckUser and Oversight.

Please note that due to Wikimedia Foundation rules governing access to deleted material, only applications from administrators will be accepted.

If you think you may be suitably qualified, please email arbcom-en-clists.wikimedia.org to start the application procedure for an appointment ending 30 June 2014. Once again, the application period will close at 23:59, 1 April 2013 (UTC). Further information is also available here.

For the Arbitration Committee, T. Canens (talk) 14:57, 31 March 2013 (UTC)

Discuss this

The Arbitration Committee is seeking to appoint at least three non-arbitrator members to the Audit Subcommittee, and is now seeking comments from the community regarding the candidates who have volunteered for this role.

Interested parties are invited to review the appointments page containing the nomination statements supplied by the candidates and their answers to a few standard questions. Community members may also pose additional questions and submit comments about the candidates on the individual nomination subpages or privately via email to arbcom-en-clists.wikimedia.org.

Following the consultation phase, the committee will take into account the answers provided by the candidates to the questions and the comments offered by the community (both publicly and privately) along with any other relevant factors before making a final decision regarding appointments.

The consultation phase is scheduled to end 23:59, 17 April 2013 (UTC), and the appointments are scheduled to be announced by 28 April 2013.

For the Arbitration Committee,
Risker (talk) 04:42, 9 April 2013 (UTC)

Shouldn't users who have access to the CheckUser tool really be known as "CheckUserers" or "CheckUserUsers"? ;) FrigidNinja 01:08, 21 April 2013 (UTC)

Welcome to Wikipedia (WP). That would be logical, I guess, but WP is not a strictly logical place by any means. NE Ent 01:18, 21 April 2013 (UTC)

I was joking, but thanks. Being logical is boring! --FrigidNinja 02:16, 21 April 2013 (UTC)

The Ombudsman Commission is currently holding a request for comment. Currently, the Commission only hears complaints regarding the privacy policy. We propose to change the scope of the Commission to also include hearing complaints about the global Checkuser and m:Oversight policy policies.

For more information please visit the RFC, which can be found at m:Requests for comment/Scope of Ombudsman Commission. Please direct all questions and comments there.

For the Ombudsman Commission,

--(ʞɿɐʇ) ɐuɐʞsǝp 21:24, 6 May 2013 (UTC)

I'm not exactly active in the areas of administration that would require me to use CheckUser services that frequently, so I apologize in advance if this is a naive suggestion—but why don't we do a procedural checkuser of anyone who is issued a formal ban (i.e. by the community or by ArbCom)? It would mean we could find the existence of any other sleeper accounts the person might use to circumvent the ban, and if it happens to be a "bad hand" account of another user, we would be able to find the master. The only objection I can think to this offhand is an argument that this constitutes fishing, but if they have violated other policies so grievously that they have been asked to permanently leave the project, I would argue that they are in a position of not being trusted to have followed the sockpuppet policy anyway. —Scott5114↗ [EXACT CHANGE ONLY] 18:05, 24 May 2013 (UTC)

If they haven't logged in for 90 days, it would be useless because they wouldn't have any stored ip adresses to compare other users against. At least, that's my understanding.--Gilderien Chat|List of good deeds 22:32, 24 May 2013 (UTC)

I wasn't necessarily thinking of users that are banned now, but rather in the future, whenever a ban is issued, that a checkuser is done as part of the banning procedure. Of course, you're right, if they hadn't used another account in the 90 days prior to the ban, the check would show nothing. But on the other hand, site bans are (or should be) an exceptional enough occurrence that it shouldn't be too labor-intensive to run a procedural check as a preventative measure. —Scott5114↗ [EXACT CHANGE ONLY] 23:14, 24 May 2013 (UTC)

The Arbitration Committee is seeking to appoint additional users to the CheckUser and Oversight teams. Experienced editors are invited to apply for either or both of the permissions, and current holders of either permission are also invited to apply for the other.

Successful candidates are likely to be regularly available and already familiar with local and global processes, policies, and guidelines especially those concerning CheckUser and Oversight. CheckUser candidates are expected to be technically proficient, and previous experience with OTRS is beneficial for Oversight candidates. Trusted users who frequent IRC are also encouraged to apply for either permission. All candidates must at least 18 years of age; have attained legal majority in their jurisdiction of residence; and be willing to identify to the Wikimedia Foundation prior to receiving permissions.

If you think you may be suitably qualified, please see the appointments page for further information. The application period is scheduled to close 22 July 2013.

For the Arbitration Committee, — ΛΧΣ²¹ 22:03, 8 July 2013 (UTC)

Discuss this

The Arbitration Committee is seeking to appoint additional users to the CheckUser and Oversight teams, and is now seeking comments from the community regarding the candidates who have volunteered for this role.

Interested parties are invited to review the appointments page containing the nomination statements supplied by the candidates and their answers to a few standard questions. Community members may also pose additional questions and submit comments about the candidates on the individual nomination subpages or privately via email to arbcom-en-clists.wikimedia.org.

Following the consultation phase, the committee will take into account the answers provided by the candidates to the questions and the comments offered by the community (both publicly and privately) along with all other relevant factors before making a final decision regarding appointments.

The consultation phase is scheduled to end 23:59, 16 August 2013 (UTC), and the appointments are scheduled to be announced by 24 August 2013.

For the Arbitration Committee, Callanecc (talk • contribs • logs) 05:06, 5 August 2013 (UTC)

Discuss this announcement

A motion has been proposed regarding activity levels for holders of both CU and OS tools. If you wish to comment, please join the discussion at the motion on the motions page. Thanks. Carcharoth (talk) 01:55, 30 November 2013 (UTC)

Please review the indef IP CU blocks at Wikipedia talk:WikiProject on open proxies/IP indefblock review 2014/CU blocks. Thanks. -- zzuuzz ^(talk) 08:40, 23 December 2013 (UTC)