Wikipedia:Requests for rollback/Draft poll

This is an attempt to poll the community as to whether some form of rollback for non-admins be granted. Jimbo has indicated that the community should decide on a poll, then poll, then ask arbcom to discern the consensus. The community therefore needs to build a poll.

Questions
What questions should we ask?
 * ^ That one.  Equazcion •✗/C • 14:33, 11 Jan 2008 (UTC)
 * Do we need to also poll on how to implement? Hiding T 14:35, 11 January 2008 (UTC)

Suggested questions

 * 1) I support rollback being given by administrators, using the Requests for rollback format.
 * 2) I support rollback being given to autoconfirmed users. (do we mean autoconfirmed as such, or autoconfirmed as in, after a month or after a certain number of edits? —Random832 14:37, 11 January 2008 (UTC))
 * 3) I oppose rollback being given to non administrators.


 * Support this version. Lawrence Cohen  14:41, 11 January 2008 (UTC)
 * Support this version as well. Tiptoety  talk 15:15, 11 January 2008 (UTC)

We should include boilerplate text encouraging people to vote for more than one option if more than one option would be satisfactory to them.

Note: I'm opposed to this being a "First preference, second preference..." thing, because the aim is to get a consensus. If 90% of people are happy with option X, but 60% of people are happy with option Y, but those 60% make option Y their "first choice", option Y shouldn't be judged the "consensus" choice given 40% clearly object to it. It needs to be a straightforward "Which of these options can you live with?" thing. — Preceding unsigned comment added by Squiggleslash (talk • contribs)

Minor rewording

 * Should rollback be given to non-administrators?


 * Yes, rollback should be given to users automatically (After a given time/a number of edits/to all autoconfirmed users).


 * Yes, by using Requests for rollback.


 * No.

Just like above, basically. One question, three possible answers. --Conti|✉ 15:27, 11 January 2008 (UTC)
 * This keeps it simple. Prompt voter for criteria (edit count, etc.) but indicate policy will be worked out after the fact. -- Paleorthid (talk) 19:44, 11 January 2008 (UTC)

Why are all the proposals putting the restricted version first? Wouldn't it make more sense to have them in order of either increasing or decreasing access? Eg:


 * Yes, to everyone


 * Yes, to all registered users with more than X edits who have joined more than Y days ago


 * Yes, to registered users who apply for access, with administrators granting access only if the user is a good boy.


 * No.

Squiggleslash (talk) 16:59, 11 January 2008 (UTC)


 * I changed the order of options accordingly. I don't think that "Yes, to everyone" is a sensible option, tho, since that would give every vandal out there rollback, too. --Conti|✉ 17:24, 11 January 2008 (UTC)

Version A of extended poll
Could be done in stages instead of all at once. Proposed by Carcharoth (talk) 14:37, 11 January 2008 (UTC)
 * (A) Is the tool useful?
 * Possible answers: yes/no/other
 * (B) Is the tool a big deal?
 * Possible answers: yes/no/other
 * (C) Should the option to give people rollbacker rights be turned on on en-wiki? (For who gets the right, see next question)
 * Possible answers: yes/no/other
 * (D) What should have the tool? (for how many, see the next question)
 * Possible answers: (i) Both humans and bots (ii) Only humans (iii) Only bots
 * (E) How many should have the tool? (for the amount of limitation, see next question)
 * Possible answers: (i) Humans: Anyone/no-one/limited numbers; (ii) Bots: Anyone/no-one/only antivandalbots
 * (F) How limited should the numbers be?
 * Possible answers: low entry barrier (eg. autoconfirmed) or higher entry barrier (selection method to be determined)
 * (G) Who should be able to both grant rollback rights and take them away?
 * Possible answers: self-selection/anyone/admins/bureaucrats/other combination (eg. admins grant, bureaucrats take away)
 * (H) Should the request and selection process be determined before or after the function is switched on?
 * Possible answers: before/after
 * (I) Is another poll required to approve the policy and decide between various different selection methods, or should those be developed by normal means?
 * Possible answers: more polling/go ahead

-- – Gurch 15:51, 11 January 2008 (UTC)
 * Oppose this version, too much. Lawrence Cohen  14:41, 11 January 2008 (UTC)
 * Good questions? Bad questions? Carcharoth (talk) 14:48, 11 January 2008 (UTC)
 * I think these are just too many questions. What's the point of the first two, for example? --Conti|✉ 14:57, 11 January 2008 (UTC)
 * It was intended to be a logical breakdown of the stages of thinking about this. I fully expect some questions to get overwhelming support, but that doesn't make them pointless. Carcharoth (talk) 15:04, 11 January 2008 (UTC)
 * They're not pointless per se, but they contribute to the feeling that these are too much questions. It's more complicated than it needs to be, IMHO. --Conti|✉ 15:13, 11 January 2008 (UTC)
 * Just to many questions overall, this process needs to be made extremely easy so that it can be resolved without more disputes. Tiptoety  talk 15:17, 11 January 2008 (UTC)
 * (J) Are you sick of these polls yet?
 * Possible answers: yes

I think QA and QB could be eliminated as difficult to !vote in a single word. QC and QD should be combined, since I think the bot part of those questions would be best dealt with through the WP:BAG and WP:BN (most users don't know how a bot works). QE/QF I think could be eliminated since again it would be hard to judge 50 users with 50 different totals. QG might be phrased as several questions: and corresponding views for removal QH and QI seem like arbcom interpretation issues, not votable topics.  MBisanz  talk 16:38, 11 January 2008 (UTC)
 * Rollback rights should be granted by an administrator
 * Rollback rights should be granted by a crat
 * Rollback rights should be granted by a Steward
 * Rollback rights should be granted 2 admins working together
 * Rollback rights should only be gratned after an RfA type discussion
 * QI is unwise. Let's not vote on having more polls... The above simplification is a draft that we could work on. I personally would object to giving the worst process wonkers any ideas of having 2 admins involved to press one button... those people will have the last option anyway (the RfA style process). EconomicsGuy (talk) 16:46, 11 January 2008 (UTC)
 * It was from an old idea I remember hearing that 2 crats, working together, should be able to de-sysop a compromised account if no steward was around.  MBisanz  talk 16:55, 11 January 2008 (UTC)


 * QE/F are vital, as the whole point is that self-selection (an option in preferences) and autoconfirmed rollback (effectively rollback for everyone) should be among the options on the table. Not everyone supports the idea of processes where someone has to ask for this tool. Carcharoth (talk) 16:58, 11 January 2008 (UTC)
 * Right. As for E though does anyone really object to the antivandalbots having this to work faster and reduce server load? I think F is more the root of the problem. EconomicsGuy (talk) 17:05, 11 January 2008 (UTC)
 * My fear is that open ended questions lead to lengthy, tiered, or unreadable responses and debate. Ideally every question should have a limited number of exclusive choices.  How to implement that (SurveyMonkey?) I do not know.  MBisanz  talk 19:04, 11 January 2008 (UTC)

Time
How long should the poll be open?

Arbitrary two weeks. Plucked the number out my head for starters. Seems a long enough period that most people would see it. Suggest it could be closed early if no vote cast in a 24 hr period. Hiding T 14:38, 11 January 2008 (UTC)
 * Two weeks - agreed. Carcharoth (talk) 14:38, 11 January 2008 (UTC)
 * Two weeks. No reason to do this longer than we need to and those who care will vote early anyway with all the noise this has caused. A notice on the watchlist will help speed things up if we do it right this time. EconomicsGuy (talk) 14:43, 11 January 2008 (UTC)


 * One month. Not everyone logs in each day. Lawrence Cohen  14:38, 11 January 2008 (UTC)
 * One day in 14 seems as likely as one day in 30. Hiding T 14:39, 11 January 2008 (UTC)
 * Then whats the harm in 30? We're in no rush. Do it right rather than having to do it a fourth time. Lawrence Cohen  14:40, 11 January 2008 (UTC)
 * How long are arb-com elections open? Hiding T 14:43, 11 January 2008 (UTC)
 * Two weeks. Carcharoth (talk) 14:47, 11 January 2008 (UTC)
 * How long was the Main Page redesign poll open? GlassCobra 14:46, 11 January 2008 (UTC)
 * 18 days: "The final election ran from 1 March 2006, 00:01 (UTC) until 18 March 2006, 23:59 (UTC)." Carcharoth (talk) 14:47, 11 January 2008 (UTC)
 * Right, sorry, got EC'd by your reply. :P This should go at least 14. GlassCobra 14:48, 11 January 2008 (UTC)
 * Criteria for speedy deletion/Proposal ran two weeks. Hiding T 14:49, 11 January 2008 (UTC)
 * Support for the passed proposals varied between 74% and 96%. Support for the failed proposals varied from 16% to 69%. Carcharoth (talk) 14:52, 11 January 2008 (UTC)
 * Three revert rule enforcement ran for two weeks. 85% support. Main Page redesign was 76%. Carcharoth (talk) 14:50, 11 January 2008 (UTC)
 * Criteria for speedy deletion/Proposal was 70%. Hiding T 14:52, 11 January 2008 (UTC)

A related question is notification. There was lots of complaining early on in the first poll since the notification didn't get to enough places. Off the top of my head, this should go on: Watchlist notice, wikien-l, all village pumps, administrators noticeboard, IRC, and on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying "Beware of The Leopard". Anywhere else? —Random832 14:54, 11 January 2008 (UTC)
 * What do you think about talkpage spam, to everyone that participated in the last 2 polls, or, that's in the 'rollbacker' group? SQL Query me!  14:56, 11 January 2008 (UTC)


 * May I propose at least a week "cool-off" period before starting again? There's been too many polls and too much confusion in a too short space of time. I suggest that any poll runs for 2 weeks starting on January 19th 2008. -Halo (talk) 15:01, 11 January 2008 (UTC)
 * It will probably take that long to agree on the wording of the questions. Carcharoth (talk) 15:03, 11 January 2008 (UTC)
 * Yeah, at LEAST a week 'cool-off'. SQL Query me!  15:10, 11 January 2008 (UTC)


 * Three and a half nanofortnights – Gurch 15:50, 11 January 2008 (UTC)


 * Don't forget Template:Cent. Stifle (talk) 16:12, 11 January 2008 (UTC)

Suffrage
The poll/vote page/pages will be semi-protected, limiting suffrage to those editors with accounts for four days or more.

Voting method
Approval? Ranked choices (in which case, how should it be counted? Instant Runoff? Condorcet method?) ?


 * Straight vote. No one can bitch or complain about pure numbers on this scale. Lawrence Cohen  14:40, 11 January 2008 (UTC)
 * Theoretically arb-com are to discern consensus. Maybe we should just poll and let them work it out. :) Hiding T 14:41, 11 January 2008 (UTC)
 * I think we should have a straight up poll on whether to have non-admin usage of rollback. That would determine consensus on that.  The rest is process related, and can be achieved as much in practise as theory. Hiding T 14:42, 11 January 2008 (UTC)
 * the problem with that is that there are people who DO want non-admin rollback but would prefer not having it to having it by any means other than auto-confirmation. We need to let people specify what their second choice is, if nothing else. —Random832 14:49, 11 January 2008 (UTC)
 * Indeed, 2 maybe 3 choices sounds reasonable. SQL Query me!  14:54, 11 January 2008 (UTC)
 * That's not a problem. If there is consensus that it should happen in some form, then it should happen. That's how consensus tends to work. You don't tend to say, I'll vote for this guy if he does that. But that's why you need a two stage poll. One for is this in general a good idea, and one for which is the best implementation? Just run them concurrently. Hiding T 14:55, 11 January 2008 (UTC)
 * Has anyone thought of having a poll? *jumps off nearest tall building* – Gurch 15:46, 11 January 2008 (UTC)
 * Well, it came down from Jimmy. It may be the least stressful way to channel everything, but if you have a better idea that you think the community will swing behind, I'd say we're all ears. Hiding T 17:10, 11 January 2008 (UTC)

Who judges consensus
Should an independent arbiter be selected/asked to judge consensus? Bureaucrats, ArbCom, Jimbo? Carcharoth (talk) 14:45, 11 January 2008 (UTC)
 * Whomever it is, more than one please. Either the cats or the Ac. Or a mix, but more than one, since no one person can call this. Lawrence Cohen  14:45, 11 January 2008 (UTC)
 * Jimbo mentioned something about the WMF board being the final deciders. Is that really what's going to happen? GlassCobra 14:49, 11 January 2008 (UTC)
 * I sure hope not. The day the board needs to judge consensus here is the day we have truly failed. EconomicsGuy (talk) 14:52, 11 January 2008 (UTC)
 * An uninvolved unbiased group of arbs selected by ArbCom itself. They can do that while we concentrate on this. EconomicsGuy (talk) 14:51, 11 January 2008 (UTC)
 * (ec)Our crats do a really good job at gaging consensus. If they're willing, I think that'd be the best way to go. Maybe a random 3 active crats? SQL Query me!  14:52, 11 January 2008 (UTC)
 * I thought Jimmy had said arbcom. Hiding T 14:56, 11 January 2008 (UTC)
 * I think he may have suggested that. I'm just suggesting something I think might be better :) SQL Query me!  14:58, 11 January 2008 (UTC)
 * The crats are generally well respected and could do this as well. Or a combination just to satisfy everyone. EconomicsGuy (talk) 15:00, 11 January 2008 (UTC)


 * We need a poll on whether we should have a poll on whether we should have a poll on whether we should have a poll on whether we should be using a poll to judge consensus in the first place – Gurch 15:36, 11 January 2008 (UTC)
 * I don't agree... Dorfklatsch 15:42, January 11, 2008
 * mu. —Random832 15:45, 11 January 2008 (UTC)


 * Jimbo and the WMF are actually the final deciders of the outcome of this Request and though they will make the final decision, its actually up to the developers to clarify on where they stand because if the developers give the thumbs up on this [rollback] request then Jimbo and the board can do nothing about it and agree with the developers and basically as mentioned by Tim, Wikimedia's second top developer, this [rollback] will actually prolong the life of the Wikimedia servers since using scripts are slowly but surely killing the servers and and as we all know these servers are very very expensive and replacing one will really be a burden for the foundation, who already don't have enough money to run the foundation and if using [rollback] can save them money, then I don't really see any reason why it wouldn't be implemented by the foundation..-- Cometstyles 15:54, 11 January 2008 (UTC)
 * But wouldn't it be preferable if we were able to make that decision ourselves without asking for help? Are you saying that rollback for everyone would have happened regardless of the polls? EconomicsGuy (talk) 16:02, 11 January 2008 (UTC)
 * If this is a technical matter for operations and not the decision of the community, why are we even doing this? Just turn it and archive all these pages right now. Lawrence Cohen  16:05, 11 January 2008 (UTC)
 * That's what I'm starting to think would have been best. Developers give everyone rollback and talk vaguely about performance needs because scripts are overloading the servers. Carcharoth (talk) 17:03, 11 January 2008 (UTC)

Disenfranchisement
Whatever happens, if no one else does it, when the "new" vote (stop calling it a poll) goes live, I will leave a talk page notice for each user that voted on the last one, so that they are notified that they need to weigh in yet again. Hopefully no one will interfere with my making those postings, as it is not canvassing. Lawrence Cohen 16:04, 11 January 2008 (UTC)
 * You might also do that other discussion from some time in the past when there were also 400 some contribs. I know there is an overlap, but with AWB-type tools that should be easily fixable.  MBisanz  talk 16:42, 11 January 2008 (UTC)
 * I really don't think that's necessary. There'll be another watchlist advertisement when the time comes.  Equazcion •✗/C • 16:43, 11 Jan 2008 (UTC)
 * Do it if you want to, but it sounds like a lengthy and rather boring task. WaltonOne 19:26, 11 January 2008 (UTC)
 * As long as it doesn't happen before next wens, it shouldn't be a problem for me to fire up AWB and pound it out.  MBisanz  talk 20:20, 11 January 2008 (UTC)
 * So according to AWB there are 1119 concerned parties, I'm figuring a tag like "Previously you have commented on an issue relating to the rollback function of Wikipedia. Currently there is a discussion going on x that you may be interested in participating in." Anyone object or can improve this wording?  MBisanz  talk 20:41, 11 January 2008 (UTC)