Meta talk:Rewriting/Oversight policy

Issues

• Appointment: When? Which requirements?
• Oversighters deal with sensitive information. Privacy policy applies?
• Minimum of oversighters per project
• Mailinglist
• IRC channel?

Wikis with Arb Coms

Looking at this I see Wiki's with Arb Com's should have two users where wiki's without must have two users. This strikes me a wrong? Surely it should be "must" in both cases? I changed the CU policy page to the same effect earlier today (Obviously if I am wrong feel free to revert me). Thanks --Herby ^{talk thyme} 12:24, 10 November 2007 (UTC)

of course you are right, wording is wrong :) Not sure though if people should be allowed to edit policies... (to the stewards it is clear btw) Effeietsanders 12:28, 10 November 2007 (UTC)

"Inactivity"

As with the CU one this needs to be defined. Oversight inactivity or project inactivity? --Herby ^{talk thyme} 14:26, 14 January 2008 (UTC)

I would prefer project rather than oversight... if there is no need for it during a particular time period, yaay! But then turning it off is silly, the world is not going to never ever do oversightable things again. Here is my suggestion: [1] ++Lar: t/c 03:32, 15 January 2008 (UTC)

Yeah, agree with that, makes sense. Effeietsanders 20:41, 15 January 2008 (UTC)

I agree - in the end oversight may well be rarely used (other than en wp) so overall activity is probably best (not my view on CU) --Herby ^{talk thyme} 12:10, 4 February 2008 (UTC)

voting margins

I reworded policy to conform to CU practice, it has been 80% and 25 or more supports, not the bands, in all the CU elections I've seen so far. This is the diff for CU: [2] I did the same to Oversight. [3] ... comments? ++Lar: t/c 03:32, 15 January 2008 (UTC)

I think the idea here was that a project could define or use it's own definition of "consensus". For instance, on nlwiki they use always 75% for "consensus". (However, there is no CU vote, but if there would be one, it would be 75%), so I think it should indeed be somewhat clearer, but I would prefer to leave it to the local projects to use their own consensus-definition. Effeietsanders 20:43, 15 January 2008 (UTC)

Can Oversighter A see what Oversighter B removed?

If someone oversights an edit, can another oversighter see what the content was, or just that the edit was oversighted on from this article at this time on this date? The reason I ask is that there seems to be a discrepancy between what I see at Meta:Rewriting/Oversight_policy#Logging ("The log lists who made the removal, when, from which page, and a provided comment. The contents may be reviewed by those with oversight rights, though the originally stated intention was that this will only be available for a limited time period.") and this recent edit at Commons. (Check the edit summary.) Cowardly Lion 20:30, 31 January 2008 (UTC)

Other oversighters cannot see the content, only the log. Majorly (talk) 20:41, 31 January 2008 (UTC)

The log is, however, sensitive, because considerable information can be derived from what exactly was oversighted. Enough so that disclosure of the log is cause for loss of oversight privs and the log page explicitly warns you of that... Note that on Commons there is an active discussion about whether a much more sanitized/innocuous log be made public, showing only that oversighter A did something on day Y. See commons:Commons:Administrators'_noticeboard#Herby_.26_Oversight_on_Commons and commons:User:Herbythyme/Oversight_log ++Lar: t/c 01:36, 1 February 2008 (UTC)

Interesting - having looked a little more an oversighter can see (via the log) what was removed by another oversighter. It cannot be seen on the page itself (history or the like) only via the log --Herby ^{talk thyme} 14:14, 2 February 2008 (UTC)

Can you see *exactly*, or is it something like "phone number"? Majorly (talk) 15:00, 2 February 2008 (UTC)

The log has a "narrative" (edit summary style) which would contain "phone number removed" or similar but when I looked closer there is a "details" link which shows what was oversighted (the whole of the edit that was removed). I'd not had time to look closely till now. I hope that clarifies things --Herby ^{talk thyme} 15:05, 2 February 2008 (UTC)

Just confirmed this with another oversighter. I'm not sure where I got the idea they couldn't see the contents - otherwise how could they monitor one another. Hmm... Majorly (talk) 16:14, 2 February 2008 (UTC)

libellous at request

"Removal of potentially libellous information either: a) on the advice of Wikimedia Foundation counsel or b) when the subject has specifically asked for the information to be removed from the history, the case is clear, and there is no editorial reason to keep the revision." is not really correct as written. It's rather common for a manufacturer to think any criticism of his products as libel, sourced or unsourced, and this would permit removal of it. On the other hand, if unsourced libel is entered in an article, we can ask for its oversight without needing the WMF counsel or the subject to authorize it. and I dont think we should be removing potential libel--just actual libel, or what is likely to be considered such. "Potentially" opens a very wide door...71.247.18.120 20:02, 4 February 2008 (UTC)

Suggestions regarding who can do what where.

• Stewards:

May oversight on any wiki, at any time, regardless of the presence of local oversights and ArbCom.

Why? Oversight is the functional opposite of checkuser. Whereas checkuser is used sparsely to protect our identities,

oversight protects us more when trusted individuals are in abundance to remove the information.

• Local Arbcom:

May select local oversights, preferably with community approval (or vice versa).

Some communities may develop oversights who do not have the full trust of the community, but may have the trust of Arbcom.

Communities with Arbcom should hold confirmations every X months (no greater than 12) to ensure that the users are still

trusted with the tools by that community.

Why? ArbCom is (imo) slowly devolving into an authority with far more power than mere arbitration. The communities should

have final say about who they trust. Lose the trust, lose your privileges.

On communities with ArbCom but no selected Oversights, Stewards handle all requests.

It is strongly suggested that there be users with Oversight rights who are trusted members of the community, but not

members of the local ArbCom. This encourages community trust in the process and prevents any group (even a trusted one

such as ArbCom) from having too much "power" in a community.

It is suggested that communities large enough to support a local ArbCom avoid having the same users hold both the

CheckUser and Oversight permissions to avoid consolidation of power.

• Local Oversights (no Arbcom)

Chosen by local consensus only.

Communities may select oversight even when not large enough to support Bureaucrats (Those with at least 5 admins)

Questions? Comments? ~Kylu (u|t) 01:14, 5 February 2008 (UTC)

Issue responses

• Appointment: When? Which requirements?

18+ and be identified to the office (not to Legal directly, it's not CU).

Be subscribed to the maillist once approved.

Be reasonably active on the project.

CU and O should both be subject to local and central recall in the event of abuse.

• Oversighters deal with sensitive information. Privacy policy applies?

Yes, if only to reinforce that oversighters do not release information that they have oversighted.

In the event that the Foundation or the (currently checkuser only) ombudsman committee requires the edits,

it is encouraged for them to do investigations directly, not by requesting the other local oversights to

perform these functions.

• Minimum of oversighters per project

Two, to keep an eye on eachother similar to checkuser, as otherwise this would encourage abuse.

• Mailinglist

We already have one.

• IRC channel?

No, use Stewards. An IRC channel would encourage people to discuss oversighting and oversighted edits.

I suggest registering an oversight channel and just redirecting it to #wikimedia-stewards

~Kylu (u|t) 01:55, 5 February 2008 (UTC)