Wikipedia talk:Flagged revisions/Sighted versions/Archive 4
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 | Archive 4 | Archive 5 |
Editing what is already edited
If I am reading this correctly, if an IP editor sees an edit they what to make they click the edit button and see a totally different version.
If an article has a Quality version does it also have a sighted version? Zginder 20:22, 23 August 2007 (UTC)
- The edit tab is removed when viewing the stable version. However, the tag at the top of the page links to the current rev/where to edit that rev and says that it can be modified. More obvious for IPs, perhaps, than just the "edit this page" tab actually. Voice-of-All 20:25, 23 August 2007 (UTC)
- Removing the edit tab? People won't even realize they can edit. Pyrospirit (talk · contribs) 01:34, 25 August 2007 (UTC)
- No. Look at the test wiki. The tag says "the current revision can be modified; X changes await review". It uses a message that admins can even further configure to make editing more apparent. Voice-of-All 22:27, 29 August 2007 (UTC)
- Removing the edit tab? People won't even realize they can edit. Pyrospirit (talk · contribs) 01:34, 25 August 2007 (UTC)
Etymology (Derivation) of sighted?
What etymology (derivation) is being used for the proposed term "sighted"? Will people confuse it with "cited"? Was the term "visible" considered and rejected? Ignoring auditory confusion, "sighted" generally means "which has/had been seen" (ignoring also "which is able to see"), but does not refer to the intention to be seen or not seen. Perhaps the usage proposed here corresponds to publishing jargon. If so, IMHO, it is jargon which it would be better not to inflict on the public at large. (I say this all independent of any opinions on whether the proposal should be implemented at all).--SportWagon 20:41, 23 August 2007 (UTC)
- In English, as I understand it, "sighted" generally means "[which is] able to see". I.e. not blind. To say "[which is] able to be seen", you would generally use "visible". To say "[which is] unable to be seen", you'd use "invisible", or perhaps "unseen". However, it is rare to use "seen" as an adjective; it tends to be the default. I.e. you might say "unseen forces", but, in contrast, you'd usually say just plain "forces", or, more likely than "seen forces", you'd say "visible forces". I don't know why. And "invisible forces" tends to imply something magical somehow.--SportWagon 22:15, 23 August 2007 (UTC)
- That's not to say "sighted" can't mean "[which is] able to be seen", but it only gets used in special circumstances, of which I don't think this is a case. In war, an enemy might be sighted by your scouts. But the word is then used because the enemy had been explicitly searched for. Even then, referring to the "sighted enemy", using "sighted" as an actual adjective, doesn't seem quite right. When you add the confusion with "cited", especially in literary contexts (such as wikipedia), "sighted" seems to be an adjective to avoid in English, except for the "not blind" meaning.--SportWagon 22:25, 23 August 2007 (UTC)
- I think in this context, it means "has been looked at". Rather a strange usage, really... How about "vandalism-free"? --Tango 22:34, 23 August 2007 (UTC)
- In the section I should have put this in (maybe) [1], a similar proposal was put forth and rejected.--SportWagon 22:40, 23 August 2007 (UTC)
- I suggest the use of the term inspected rather than sighted. It's clearer. Pyrospirit (talk · contribs) 01:35, 25 August 2007 (UTC)
I think sighted is the best option discussed so far. I don't see a significant problem with confusing it for "cited", since it is always spelled, not said, on wiki. It's "sighted" as in "it has been sighted", the same way scouts would sight an enemy. It's concise, and with the corresponding "surveyor" right, even a little bit poetic.--ragesoss 01:52, 25 August 2007 (UTC)
- "sight" is inherently a problematic word. While scouts might sight an enemy, and say the enemy "has been sighted", people would avoid talking abstractly about the "sighted enemy", because that would contrast with the "blind enemy". Further, "sight" appears to be related to "seek", and "sought", but isn't quite. And if we assume articles or revisions start out "unsighted", natural usage, such as it is, would suggest they become "sighted" when an appropriate entity merely sees them, and not subject to any approval process whatsoever. But the proposal instead does say there is a selection processed. (This suggests "selected" as an alternative). While "sighted", like many alternatives, could be said to be concise, these factors stop "sighted" from being appropriately concise.--SportWagon 14:55, 27 August 2007 (UTC)
Yet Another Alternative Proposal: article endorsements
It seems to me that having a single account being able to flag or unflag a sighted article is an unstable position, especially on controversial or high-traffic articles which this policy is ostensibly targeting. My idea would place an "endorse" tab next to the "edit" tab. It would be labeled with the number of endorsements given to it by editors ("22 endorsements"). When clicked, a log page would be presented that identifies the users who endorse the given version. Users who are logged in (with or without some other condition, like edit count, age of account, etc.) would be presented with a button labeled "endorse this version" or "unendorse this version" to add or remove their endorsement. The count of endorsing users becomes a metric that can be used to sort a page history, to find a "most trusted" version. Furthermore, the list of endorsing users is available to any reader who wants to audit the trustworthiness of those who have chosen to endorse the version. It places a burden on readers to determine for themselves the quality of an article, but I think this is actually a good thing. Cmprince 20:49, 23 August 2007 (UTC)
- I endorse this version! :D Furthermore, I suggest that in the proposal brought forth, instead of the "Sighted" article being shown to unregistered users, the version with the most number of endorsements be shown instead. This way, we can still remain as a wiki and provide improved QA at the same time. speaks rohith 20:56, 23 August 2007 (UTC)
- No, that has major problems, IMO, because it doesn't nessesarily endorse an edit that could EASILY slip through the crack otherwise -- a changing of a year, for instance. It might read as if there's no vandalism, even when there is. Looking at the edit history, though, lets you find these out. ♫ Melodia Chaconne ♫ 21:00, 23 August 2007 (UTC)
- The other problem with this is that we need to be very careful with terminology. This didn't occur to me until I read others comments, but check it out: If this is phrased wrong, it could actually increase the odds of Wikipedia being successfully sued. It's one thing to have a page that reads "Celebrity X worships the devil and hates children" -- it's another thing entirely if, at the top of that same page, it says, "This article is endorsed by Wikipedia!"... or in any way shape or form makes it sound like Wikipedia is asserting its validity.
- Great care must be taken that, if this proposal or some variant is accepted, Wikipedia does not make a stronger statement about accuracy than it already does... rather, Wikipedia is making a weaker statement about Unsighted articles than in the status quo. Otherwise, you could be looking at some real legal hot water. --Jaysweet 21:12, 23 August 2007 (UTC)
- I guess that's why I wanted to get away from a "sighted" concept; a binary quality flag is essentially an endorsement on behalf of the entire community. Endorsements by individual users alleviates that. That said, if there's a better word for "endorsement" that would be preferable, than I'd gladly--ahem--endorse it.(gosh, I'm sick of writing/reading that word already) Cmprince 21:21, 23 August 2007 (UTC)
- No, that has major problems, IMO, because it doesn't nessesarily endorse an edit that could EASILY slip through the crack otherwise -- a changing of a year, for instance. It might read as if there's no vandalism, even when there is. Looking at the edit history, though, lets you find these out. ♫ Melodia Chaconne ♫ 21:00, 23 August 2007 (UTC)
- If there is a "flag" tab, it'll need to flag the exact version that is being viewed, regardless of edits made between when the article was loaded and when the editor flagged it. I recently locked down an article mentioned on Colbert and thought I'd beaten the vandals to the punch, only to find that they'd hit it twice between when I loaded the article and when I hit "submit" on the protection page. EVula // talk // ☯ // 21:20, 23 August 2007 (UTC)
Oppose this counter-proposal; we don't want to count votes on articles. Otherwise the high traffic articles will get 10 times more votes than the less trafficked ones, regardless of reliability. --AnonEMouse (squeak) 21:30, 23 August 2007 (UTC)
- I guess I don't see this as a necessarily bad thing. I would expect there to be more endorsements on higher traffic articles. More eyes watching an article are more likely to identify a higher-quality rev than fewer eyes. Note, though, that absolute endorsement counts shouldn't be compared between articles. A rev of George W. Bush and a rev of Bushwick, Brooklyn may both carry 10 endorsements, but it would be easy to see if there was a GWB rev that had more (perhaps hundreds more), or if there was a Bushwick rev that had more (perhaps not). It's a tool to determine the quality of an article with respect to its own history of edits, not with respect to other articles. Again, it would be asking the reader to do a little investigation to make a determination of trustworthiness for him- or herself (which all researchers should be doing anyway). Cmprince 21:55, 23 August 2007 (UTC)
- The biggest problem with this kind of system is that it is inherently biased against new versions (in a much stronger and more lasting way than flagged revisions). In general, we assume that newer versions are better, assuming nothing counterproductive has been introduced. Voting on specific past versions against each other is a strong indication that there is no consensus for the current version. There usually will be consensus for the current version, but that would get obscured if past versions were the subject of more intense debate, and all changes since then have been uncontroversial but managed by a smaller number of editors. We know that consensus editing works pretty well; the biggest danger for readers is that they are viewing a version that is in the midst of some biased non-consensus editing or vandalism, and that's the main problem this proposal is meant to address. No matter what the past voting patterns were like, I'm going to want read that last version of George W. Bush that a trusted editor reviewed and didn't revert.--ragesoss 22:28, 23 August 2007 (UTC)
- That's well and good, but anyone reading that version should be aware that just one other editor deemed it trustworthy. We should be biased against new versions; that's why people patrol RC. We should never assume a new version is better just because it's newer. Edits should always be vetted for accuracy by the community, and the more editors involved, the better. Also as clarification, this implementation would not enable any sort of debate on endorsing one version versus another. In this way it is not a voting system, in which editors try to influence one another. It would simply be a system for recording endorsements, no more and no less. Cmprince 22:59, 23 August 2007 (UTC)
- The biggest problem with this kind of system is that it is inherently biased against new versions (in a much stronger and more lasting way than flagged revisions). In general, we assume that newer versions are better, assuming nothing counterproductive has been introduced. Voting on specific past versions against each other is a strong indication that there is no consensus for the current version. There usually will be consensus for the current version, but that would get obscured if past versions were the subject of more intense debate, and all changes since then have been uncontroversial but managed by a smaller number of editors. We know that consensus editing works pretty well; the biggest danger for readers is that they are viewing a version that is in the midst of some biased non-consensus editing or vandalism, and that's the main problem this proposal is meant to address. No matter what the past voting patterns were like, I'm going to want read that last version of George W. Bush that a trusted editor reviewed and didn't revert.--ragesoss 22:28, 23 August 2007 (UTC)
What if, instead of the visible version being the revision with the most endorsements, it was the the most recent revision with endorsements above a certain threshhold? The threshhold could be adjusted on a per-article basis, depending on how often the article gets vandalized. For instance, the endorsement threshhold could be set equal to the number of previous edits that have been reverted. Then all articles would work the way they currently do until the first time they get vandalized; after that, subsequent edits would need one endorsement before they become visible. After the next incidence of vandalism, edits would become visible after two endorsements; and so on. --Abou 18:20, 24 August 2007 (UTC)
- Actually, I don't agree with ever showing anything other than the most recent rev by default. It may be that I'm missing the point of this proposal, but I understand it to be a mechanism for indicating to a reader (who may or may not be versed in the ways of the wiki) whether or not any given article meets a standard of quality, including being free of vandalism. If it's meant to prevent vandalism from ever appearing on a page, though, I think it's misguided. The problem with any vandalism-prevention scheme is that so much of it is subtle, as many here have pointed out. Under an endorsement scheme, every new rev would start at 0 endorsements (1, if a qualified editor endorses the article with his edit). IP vandalism would always start at 0 endorsements. This notation, by itself, should serve the purpose of quality control. If readers really want to minimize the amount of vandalism they see, perhaps that could be handled by an account setting ("show most-endorsed rev in last n months", for example). But by default, in the interest of keeping a dynamic, continuously-improving encyclopedia, we should always show the live version. Cmprince 18:58, 24 August 2007 (UTC)
The Wiki Way
People said Wikipedia would fail to produce anything of quality, and look at it now. We can only tell whether or not this is a good proposal through the wiki way, that is, we try it first, and if it's bad, we'll revert to a no-flagged-revisions state. MessedRocker (talk) 21:25, 23 August 2007 (UTC)
- Amen. —Ilmari Karonen (talk) 21:28, 23 August 2007 (UTC)
- This proposal, both halves of it, is fundamentally unwiki. That's what's wrong with it. Septentrionalis PMAnderson 21:40, 23 August 2007 (UTC)
- Fundamentally unwiki? Not sure, but quality assurance is fundamental to an encyclopedia (and Wikipedia is supposed to be one, I think). --B. Wolterding 14:23, 24 August 2007 (UTC)
- This proposal, both halves of it, is fundamentally unwiki. That's what's wrong with it. Septentrionalis PMAnderson 21:40, 23 August 2007 (UTC)
- Agreed. Try, revert, try again. -- Myke Cuthbert (talk) 21:54, 23 August 2007 (UTC)
- I'm pro-bold, but when proposing significant, fundamental changes to something as successful as Wikipedia, the burden of proof that the changes are necessary and practical improvements lies with those proposing it. This proof must be provided before any changes are made; otherwise Wikipedia would be trying out dozens of major changes every month. ~ Danelo 14:36, 24 August 2007 (UTC)
- I agree it is unwiki-ish to hide changes made by a sub-class of editors and privledge others. Basically, I wonder if it will drive new users away. I'm also still unclear about who will decide who gets the ability to change sighted versions and make them. Moonbug 07:22, 26 August 2007 (UTC)
- I'm pro-bold, but when proposing significant, fundamental changes to something as successful as Wikipedia, the burden of proof that the changes are necessary and practical improvements lies with those proposing it. This proof must be provided before any changes are made; otherwise Wikipedia would be trying out dozens of major changes every month. ~ Danelo 14:36, 24 August 2007 (UTC)
Who approves which topics?
As the proposal currently stands, anyone with Surveyor Reviewer [corrected as per below] status would be able to approve of the accuracy of any article. I'm not certain that this is necessarily a good thing - I want mathematicians to check the math articles, biologists to check to the biology articles, etc. While I'm not saying that others are incapable of doing so, I don't think it's ridiculous to say that any true peer review process (which is essentially what this is) is faulty if the reviewers are not peers in the sense of having a deep and fundamental knowledge of the topic. Girolamo Savonarola 22:21, 23 August 2007 (UTC)
- I think this is a definite concern at the Quality/Reviewer level, but not necessarily at the Sighted/Surveyor level, where we're just looking for pure vandalism/nonsense. Unless people start putting particularly subtle libel, I think most of that can be caught without expert knowledge. I agree that Reviewing will eventually need to be topically based: there's a discussion about this in /Archive 2 I believe, where I asked if Reviewer tags could be given on a per category basis (and looked at (and largely dismissed) the potential for abuse of adding and removing category tags). However, Voice-of-All said that this would require a major rewrite of the privilege system as a whole beyond what can be done with a this extension. I think it will eventually need to be done though, butt probably not in this proposal. -- Myke Cuthbert (talk) 23:04, 23 August 2007 (UTC)
- Emended accordingly. (Got them mixed up.) Girolamo Savonarola 23:13, 23 August 2007 (UTC)
- But right now, any editor (even those below the level of 'Surveyor') has exactly this power! Anyone can revert a bunch of edits - even if they know nothing about the subject and, indeed, know nothing whatever about Wikipedia. If a vandal reverts a bunch of my edits (they are still in the database - but nobody will see them without looking into the history) - he's doing the same thing as if he'd sighted some older version of the article and thereby hidden subsequent changes. That's more power than the surveyor has! SteveBaker 23:56, 23 August 2007 (UTC)
- I'm talking about the fact-checkers - whoever is going to review the article for completeness, accuracy, and proper sourcing. Girolamo Savonarola 00:20, 24 August 2007 (UTC)
- The 'quality' flag would be set following a review similar (or perhaps identical to) our current 'featured article' process. Active contributors to the page and people who commonly review featured articles would discuss it, identify things needing improvement, make updates, et cetera. When there is consensus that the page is accurate someone with the access would set the 'quality' flag. This gives input to people who are not 'mathematicians' (per your example), but there should be such input... to assure that the writing style reflects Wikipedia standards, that the information is properly referenced, et cetera. As the users working on the page would generally nominate it in the first place, and are notified of the discussion in any case, there usually wouldn't be any lack of people with knowledge of the subject. If you are looking for something like requiring that only those with a PhD in mathematics are involved in the decision, I don't think that's likely or sustainable. --CBD 11:19, 24 August 2007 (UTC)
- See also #Quality...is dubious. Sighted is OK. above. NB, ragesoss has now removed reviewer status & quality flags from the proposal altogether. PaddyLeahy 11:35, 24 August 2007 (UTC)
Account registration will be attacked next
My biggest concern with this is that account registration will be attacked next. Learn from the email spam war — everything turns into an arms race, a game of cat-and-mouse. If we assign having an account valuable to attackers (vandals, spammers, etc.), we'll get tons of accounts being registered for the purpose. If we add a time delay, we'll get tons of accounts that are idle for the time delay. We should already be looking ahead. —DragonHawk (talk|hist) 22:29, 23 August 2007 (UTC)
- It's not simple account registration (which is already attacked for article creation privileges) that matters here. It requires 300 edits, and if would-be vandals put in 300 good edits before they start abusing the reviewer privilege, then I think we got the better end of the deal, since they'll be caught before they do too much. Would-be vandals have much easier ways to abuse us than through this.--ragesoss 22:36, 23 August 2007 (UTC)
- (edit conflict)So what's your opinion on this proposal then? --YbborTalk 22:37, 23 August 2007 (UTC)
- Well, I think the basic assumption to make here is that vandals are usually quite dumb, and have no persistent interest in Wikipedia. And that seems to be a valid assumption, looking at the current semi-protection system: It reduces the amount of vandalism on those pages considerably; but it would only take an account registration and 4 days waiting time to "circumvent" it. Requirements for surveyor rights will be much higher than that. Certainly, the sighted revisions will not solve all vandalism problems, but they will greatly reduce it. The most subtle threat is, as always, from malicious users who understand how Wikipedia works - but fortunately these are rare. --B. Wolterding 14:04, 24 August 2007 (UTC)
Query to opposers
I think the reason that so many people are in opposition to this is that it is too big of a change all at once. Some people are more resistant to change than others. I'm not saying that that is a bad thing; sometimes it may be best to gradually implement changes. This may be one of those times. My question to those who oppose this is: Would a very gradual implementation be more acceptable than the current version? I'm thinking something like this:
- Initially, we will predominantly use sighted revisions. These will only be used on FAs and possibly GAs (the GA process does not lend itself easily to implementing it like after an FAC). This will help to make articles regarded for quality stay in quality condition and will have very little chance of a major backlog. Surveyor rights will be given automatically with 300 edits or to those who ask an admin on a request page who aren't obvious vandals/trolls/other disruptive editors. Any abuse will be easy to spot because only FA/GAs will be sighted. Abuse may be reported to a new noticeboard or to an existing one such as WP:ANI or obvious abuse to WP:AIV. Reviewer rights will be given only to WP:OFFICE and WP:OTRS people for official work. (estimated time: 2 months)
- If this seems to work, and if the German experiment is a success, we expand it. Sighted may be used in place of semiprotection in some pages to test effectiveness against vandalism. (estimated time: 1 month)
- Begin main implementation. Create a process for assigning "Reviewer" rights so that we have enough reviewers to handle a couple thousand pages (FAs). Requirements should be mainly mainspace experience. This is so that people wwith the right know what to look for in article changes (what is an improvement, what isn't) participation in the FAC/FAR process would also be helpful. If there are no problems, the automatic threshold for granting surveyor rights could also be lowered as we begin to mark other articles as sighted. This should still be used sparingly; stubs should not be marked. Criteria could be created as to what should be marked as sighted. These could be slowly expanded to help with scalability.
Registered users of course would get to choose which version of a tagged article they would like to see. For unregistered users, they should see the sighted/quality version. We do need to remember that this is for readers as much (perhaps more than) as it is for editors and readers will want the accurate version. There will of course be a notice on top of the page and there should be an obvious link to the current version.
Please tell me if this is better/worse, if I'm missing something critical, other suggestions. Mr.Z-mantalk¢ 00:07, 24 August 2007 (UTC)
- I don't see why the reviewers can't be an ad-hoc committee, no special user status, with an admin reviewing their discussions and flagging the version they agree upon.--Father Goose 00:14, 24 August 2007 (UTC)
- I thought about that, but was afraid that that may create too many issues. Every article tagged as quality (if it is FAs there are over 1500) would basically be in a constant FAC/FAR(=possible huge backlog). If it is always the same editors commenting on the same article, there are also WP:OWN or cabal-ish issues. What we could have though, is a policy that any possibly controversial (large additions/removals) edits be submitted to an Wikipedia:Edits for review-type page for general discussion, while minor things (grammar/spelling, small additions/removals) could be flagged by anyone with review rights. Mr.Z-mantalk¢ 00:22, 24 August 2007 (UTC)
- Until we have more formal systems of expert review, "Quality" versions are never going to mean anything more than the consensus of such ad hoc committees (as happens for Featured Articles, for example). So I agree with Father Goose that it's too early to talk about expanding the system beyond handling FA's and possibly GA's for Quality status. Other than that, think a phased rollout would be good, though I think doing the Sighted rollout all at once first, followed by the Quality, would be better. Having a period when Surveyors can only sight certain articles, then changing the rules makes it too confusing.--ragesoss 00:24, 24 August 2007 (UTC)
- I thought about that, but was afraid that that may create too many issues. Every article tagged as quality (if it is FAs there are over 1500) would basically be in a constant FAC/FAR(=possible huge backlog). If it is always the same editors commenting on the same article, there are also WP:OWN or cabal-ish issues. What we could have though, is a policy that any possibly controversial (large additions/removals) edits be submitted to an Wikipedia:Edits for review-type page for general discussion, while minor things (grammar/spelling, small additions/removals) could be flagged by anyone with review rights. Mr.Z-mantalk¢ 00:22, 24 August 2007 (UTC)
Since this is directed at those who oppose, thought I'd throw in my comment on your suggestions/ideas. I don't think it matters whether this is slowly or quickly implemented, it just plain makes me sick. While I know that people who are in favor of this see it as a great way to ensure quality control for the reader, I see it as killing the wonderful, near-absolute transparency of the Wikipedia editing process and creating a much more hierarchical editing process for the core articles. Not to mention that it makes the learning curve harder and longer for any new contributors to join and edit articles they interested in. This is a step in the direction of Citizendium imo. What's the point of allowing IP editing when they can only edit articles of peripheral importance? VanTucky (talk) 00:37, 24 August 2007 (UTC)
- IPs will be able to edit anything (even George W. Bush) and once they start to edit they will choose to look at the current, not flagged, version. The extra click might drive some to register sooner than before. The only problem is that it might also drive vandals to register more often than now. PaddyLeahy 00:51, 24 August 2007 (UTC)
- The third step is a bad idea. We can't claim to be a freely editable encyclopedia and then essentially lock down an article once it reaches a certain level of quality. -Amarkov moo! 00:53, 24 August 2007 (UTC)
- I think an example which demonstrates why even the most solid, high-quality articles don't need to have flagged revisions is the guinea pig article. This has been an FA since May, and it has recieved very little improvements since. One would think it doesn't need much attention, much less from inexperienced users and IPs. But just the other day a new user who's only been around since July 16th uploaded both a useful new image and a good, cited fact that was not previously covered at all. Would they have even attempted to do so if they knew their change would not take effect until it had been vetted through some cabal? VanTucky (talk) 01:58, 24 August 2007 (UTC)
- False dilemma. If they've been contributing since July 16th they'd have access and their change would appear in all views immediately. Even if they were a brand new user/IP their change would be immediately displayed to them, appear immediately for all logged in users, and (since it was beneficial) appear in short order for non-logged in users. I'm not seeing any sort of terrible discouragement from editing here. Especially when you consider that they'll be able to edit pages which they previously couldn't due to protection and semi-protection. Flagged revisions will allow all users to edit more freely than they do now. --CBD 11:10, 24 August 2007 (UTC)
- I'm not sure that this is a false dilemma. There is no way to be sure that it would appear in short order if they were a brand new or ip user or a user with too few edits to become a surveyor or whatever, especially since once this system is implented there are going to be gobs of articles needing constant vetting. This proposal betrays the spirt of wiki and will be a nightmare to implement. I do prefer gradual implementation if people are dead set on implementing it. I also favour a low bar for automatic surveyor status. As a new user I can testify that wikipedia already has a steep learning curve. Making it steeper WILL drive people away.Moonbug 07:31, 26 August 2007 (UTC)
- False dilemma. If they've been contributing since July 16th they'd have access and their change would appear in all views immediately. Even if they were a brand new user/IP their change would be immediately displayed to them, appear immediately for all logged in users, and (since it was beneficial) appear in short order for non-logged in users. I'm not seeing any sort of terrible discouragement from editing here. Especially when you consider that they'll be able to edit pages which they previously couldn't due to protection and semi-protection. Flagged revisions will allow all users to edit more freely than they do now. --CBD 11:10, 24 August 2007 (UTC)
- I think an example which demonstrates why even the most solid, high-quality articles don't need to have flagged revisions is the guinea pig article. This has been an FA since May, and it has recieved very little improvements since. One would think it doesn't need much attention, much less from inexperienced users and IPs. But just the other day a new user who's only been around since July 16th uploaded both a useful new image and a good, cited fact that was not previously covered at all. Would they have even attempted to do so if they knew their change would not take effect until it had been vetted through some cabal? VanTucky (talk) 01:58, 24 August 2007 (UTC)
who is backing this proposal
For such a major proposal, I am curious as to whether anyone "high up" is taking this seriously? As an opponent of this idea -- which amounts to changing "edit this page" to "suggest an edit to this page" -- I would like to know whether to worry that it might actually come to be. Sdedeo (tips) 00:24, 24 August 2007 (UTC)
- I believe, if you read above, this is comming directly from the foundation. They have said (I believe) that they want community input, though it seems that there is some debate as to whether they will listen one way or the other. Adam McCormick 00:30, 24 August 2007 (UTC)
- It's been planned and looked forward to for quite a while now. I'm pretty sure everyone "high up" that you could think of expects some sort of implementation of stable versions to happen sooner or later. But that doesn't mean the implementation will necessarily default to showing anonymous readers the flagged version; it's going to be up to each project to decide how to use it. It may offer readers the chance to see the flagged version, rather than showing the flagged and offering the current version (at least at first, if not permanently).--ragesoss 00:34, 24 August 2007 (UTC)
- Search Archive 2 for comments by User:Eloquence. NB: this was discussed in the candidate Q&A in the board elections, so they have a mandate of sorts. And no, it amounts to editing a "work in progress" page which is not the default view of casual readers, although they can look at it if they want. PaddyLeahy 00:38, 24 August 2007 (UTC)
- That's a very complicated way of saying the same thing. To me, this whole idea -- with its bizarre layers and titles and levels of authority -- is pretty much what you'd expect from wiki's swelling bureaucratic class. Sdedeo (tips) 00:43, 24 August 2007 (UTC)
- Directly from the Foundation? Hmm, that sounds an awful lot the definition of bureaucracy to me. "A system of government in which most of the important decisions are made by state officials." VanTucky (talk) 00:46, 24 August 2007 (UTC)
- Taking it quite seriously indeed -- a contractor was even hired by the WMF for the development of FlaggedRevs (that has, however, for the most part been taken over by Aaron Schulz). It's pretty certain that such a thing will be implemented -- at least for a trial period -- in some form or another; however, it will likely be up to us as to exactly how it will be used. AmiDaniel (talk) 04:19, 24 August 2007 (UTC)
"Quality" aspect removed
Because of the high level of opposition to particularly aspects of this proposal, and the "Quality" versions in particular, I've removed that from the proposal. We should take this one step at a time, and it is perhaps overwhelming to thing of too much change at once.--ragesoss 00:54, 24 August 2007 (UTC)
Another step back: Current version as default
I've modified the proposal further, to have the current version as the default view, at least initially. I feel strongly that eventually having flagged versions as the default view is the way to go, but the opposition to that aspect of it is stronger than was apparent during the drafting of the proposal.--ragesoss 01:25, 24 August 2007 (UTC)
- IMO almost all of the opposition to this proposal has been based on incorrect assumptions of what it is and how it would work. Changing the proposal to conform to inaccurate ideas seems counter-productive. For instance... if the most recent version (rather than most recently flagged) is displayed to IP users by default, per the change you cite above, we won't be able to remove semi-protection from pages. New users and IPs will still be prevented from editing those pages at all and thus we would be keeping restrictions on their editing privileges rather than removing those restrictions as the flagged revisions are intended to do. Given that the opposition has almost universally voiced disapproval of restrictions on editing it is perverse that the result of their efforts would be something which, in fact, causes greater restrictions. It seems self-evident that IP and new users being able to make edits which are immediately visible to all non-logged in users, and which will be visible to ALL users the next time any regular contributor updates the page and approves them, is far far less restrictive than IP and new users being unable to edit the page at all. The change on fully protected pages is even greater... all users will be able to edit the page rather than just admins. Yes, the contributions of IPs and brand new users won't be displayed to non-logged in users immediately, but 'short delay for SOME users to see the change' has to be better than 'you cannot edit this page at all'. --CBD 13:15, 24 August 2007 (UTC)
- This does not make sense to me; it makes the proposal inconsistent. If anonymous users are seeing the latest version, rather than the sighted version, there is almost no point in having sighted versions in the first place. This is supposed to be a quality assurance mechanism (which, in my opinion, Wikipedia desperately needs); but quality assurance is quite useless if you do not deliver the quality-checked version by default. Who will spend the time and mark versions as sighted if this does not have an effect on most users anyway? --B. Wolterding 13:33, 24 August 2007 (UTC)
- Exactly. Sighted versions have only ever been intended for anon users (logged in users are assumed to want to edit, that's why they logged in), and anon users don't have preferences, so they can't change away from the default. The default has to be showing the sighted version to anons, or there is no point having the feature at all. We could have a 2 week test before setting that default to make sure we have the procedures for sighting things all sorted out, but more than that defeats the whole point. --Tango 15:21, 24 August 2007 (UTC)
- I certainly agree, but it might also make sense to begin with having it not default the sighted version, and once editors get used to it and we know how sighting patterns work out (in terms of backlogs, "flag wars", etc.) we can switch to take advantage of the intended purpose. That said, I'd be happy to see it reverted back to sighted-as-default for non-logged-in users (which, now that I check, you've already done).--ragesoss 19:28, 24 August 2007 (UTC)
- I think that showing the latest version (not the sighted version) in an early phase would be the wrong rollout strategy. It would not demonstrate how the system works, and probably editors wouldn't be much interested in it. One should in fact start small; but IMO rather with applying sighting only to a selected number of pages first, say semiprotected ones, and gather feedback from that. I'm quite sure that, this way, users will want to adopt the new system for more pages quite fast, say Wikiprojects who want to protect their top-importance articles from vandalism. --B. Wolterding 11:45, 25 August 2007 (UTC)
My two cents (what I posted on the poll)
I don't like it because it's too liable to cause arguments. I also think that it goes against the "anyone can edit" dictum that we pride ourselves on. Vandalism only accounts for 6% of edits, anyway, and given the level of anonymous edits that are actually valid, I think the log would quickly become umanageable - we have a hard time catching up with vandalism as it is. On the subject of sighted newpages, I don't like more, some requirements, such as "unencyclopedic content" are subjective (for example, trivia is, but that didn't stop at least one article from becoming featured (yes, I know it got demoted)), while the spell-checking and stub requirements discriminate against non-specialists and non-native speakers - anyone can edit, not just fluent English speakers, and stubs are vital as they exemplify the "anyone can edit" dictum more than an FA does, as a stub encourages people to edit ("it's okay, we won't bite") while an FA discourages (because it's complete). Will (talk) 01:07, 24 August 2007 (UTC)
- As far as being 'backed up', one imagines, I'd hope, that there could be a listing of the pages with the most needed edits for sighting, etc. ♫ Melodia Chaconne ♫ 01:33, 24 August 2007 (UTC)
- Given that the current system works pretty spectacularly already, why change it? This seems like a perfect situation to use the adage 'if it ain't broke don't fix it.' Backlogs and betrayal. That is pretty much the essence of this proposal as far as I can see. I've pretty much said all this already, so I'll shush now. Moonbug 07:39, 26 August 2007 (UTC)
the definition of...
From MediaWiki:
Article validation allows for Editor and Reviewer classes of users...[emphasis added]
Since the use of flagged revisions by definition requires the creation of a new class of user with special privileges, and in recognition of the idea that "Wikipedia's success to date is entirely a function of our open community" (from Jimbo) and that we are strongly admonished that "there must be no cabal, there must be no elites, there must be no hierarchy or structure which gets in the way of this openness to newcomers." I believe this proposal should be soundly rejected as a contradiction of the spirit of transparency and welcome that Wikipedia represents. To quote the man once more, "'You can edit this page right now' is a core guiding check on everything that we do. We must respect this principle as sacred." By taking away the "right now" factor of this concept, we discourage, and may effectively destroy, this guiding principle of openness and ease of editing (for everyone) that a wiki encyclopedia embodies. VanTucky (talk) 02:14, 24 August 2007 (UTC)
- I think you hit the nail on the head. .V. [Talk|Email] 02:51, 24 August 2007 (UTC)
- I don't think that this additional feature is directly in conflict with your statement, many of the extra comments made by editors above do and it could be implemented badly, but the tool itself is no more abhorant to openness that any of the already present levels of "Adminship" etc... Adam McCormick 03:19, 24 August 2007 (UTC)
- Adminship is not a requirement for making direct, immediate edits to articles. It is a system of mediating the factors surrounding editing, and does not automatically and permanently preclude the ability of anyone to make substantive changes to an article. No other factor of the current Wikipedia bureaucracy permanently negates the ability of all to edit articles in real time. Actions such as banning individual users and making temporary protections all pass the test of strict scrutiny; meaning that measures instituted for security address a compelling community interest, and are narrowly tailored to achieve that objective and no other. The current proposal does not. VanTucky (talk) 03:31, 24 August 2007 (UTC)
- Sounds like someone's channelling Alexander Hamilton :-). Remind us which clause of the Constitution enshrines the right to direct, immediate edits to all Wikipedia articles in real time? Also, an exhaustive list of all categories of "compelling community interest" will help too, please have that ready asap. I'm betting the list includes "verifiability", "reliability" and "freedom from vandalism". dr.ef.tymac 03:57, 24 August 2007 (UTC)
- And immediate appearance of all edits by everyone is not a requirement for the open community and open editing of Wikipedia. Edits are still recorded, and anyone who wants to see them can. In fact, the 'unsighted edits don't show for everyone' version of this proposal (which is not a necessary part of the implementation) will allow increased open editing on semi-protected pages, which currently can't be edited at all by anons and new users. This would allow us to do away with semi-protection (and with the later addition of other flags, we could even open up the Main Page for editing by everyone).--ragesoss 03:59, 24 August 2007 (UTC)
- Thanks for mocking my sincerity dr. I guess if the basic structure of what a wiki system stands for, and the opinions of Wikipedia's founder, mean nothing to you then there's no convincing you. If you think that the quest for absolute accuracy supersedes the direct, open editing of content, then I personally think Britannica and Citizendium would be a perfect fit for you. The problem is, is not that edits must be enacted immediately, but that a special class of users must vet the edits in order for them to take effect. If it was just a small waiting period (which, btw, doesn't seem to be necessary or make much sense, wiki means quick after all) then there would be no serious problem. But creating two classes of editors, those who must have their edits have a waiting period and those who are "trusted" not to, violates the equality of all content contributors that Wikipedia is meant to have. Another problem is the decision about which articles qualify for the flagging; is it just pages that were often semi-protected, FA's, GAs? It's completely subjective. VanTucky (talk) 04:11, 24 August 2007 (UTC)
- They've already let slip that it's only FAs at first, and if it "works well," they will eventually apply it to all articles. The stated plans have continually changed, since the beginning of this discussion (read it straight from the top, and you'll see what I mean). Badagnani 04:14, 24 August 2007 (UTC)
- I still think that, while they sound nice, the current requirements for sighted pages are extraordinarily too broad and unspecific to be practical and fair in application. And it seems like a disgustingly small number of the hundreds of thousands of contributors to this wiki are shaping this proposal, which would so fundamentally change how the project works. VanTucky (talk) 04:17, 24 August 2007 (UTC)
- Apologies, VanTucky, no offense was intended in my reflections on your earlier statement. You do make legitimate points. (although I respectfully suggest reasonable people might disagree on the seriousness and extent of the changes necessary to make a proposal like this workable) ... my only humble request would be for people to reserve any dire pronouncements of gross inequality and Citizendium-doomsaying until more people have had a chance to weigh in. dr.ef.tymac 04:53, 24 August 2007 (UTC)
- That's a completely reasonable request, but I don't think (from some of the comments made here) that it matters whether 1 or 1,000 users weigh in on this. Several people have made it clear that this was decided by the Foundation and those who got to contribute at Wikimania, and that despite any vocal opposition it will be enacted in some fashion. VanTucky (talk) 06:02, 24 August 2007 (UTC)
That's not the case at all. It's certainly true that where and when it's been discussed before, there has been widespread support. But this isn't going to be imposed on an unwilling Wikipedia. It may be initially implemented without defaulting to the flagged version for non-logged-in readers. Most of the opposition here is to specific aspects of the proposal, not wholesale opposition to any kind of use of flagged versions at all; we will work out a rough consensus for how to begin using the extension, and work from there. Naturally there will be some who vocally oppose anything (just as there as some who vocally oppose the status quo), but this discussion is far from hopeless in terms of reaching a broader agreement.--ragesoss 06:10, 24 August 2007 (UTC)
- This discussion post was very clear:
- "...this has been discussed ad nauseum, and, ultimately, the Board wants it, so it's going to happen. I don't feel much need to get upset about it. AmiDaniel (talk) 10:07, 3 June 2007 (UTC) Badagnani 06:14, 24 August 2007 (UTC)
- Well, I think the last notion is a bit early Rage, considering a patently wide selection of Wikipedia users have not had their opinions solicited. And in terms of whether this has been decided on already, some have asserted otherwise as above (not that the opinion of one or two is fact, but the Foundation hiring someone to oversee the technical side is pretty convincing). VanTucky (talk) 06:19, 24 August 2007 (UTC)
- (edit conflicted) It is going to happen. But it will happen because (in those ad nauseum discussion for it) the arguments for doing it greatly outweigh the arguments against, and sooner or later we'll hammer out a solid consensus proposal on how to implement it. It won't happen against the collective will of Wikipedia, but with it. AmiDaniel has been around long enough to know a discussion that's on the wrong track when he sees one, and he's confident (as am I) the things will get back on track once we start talking about how this can best be used to make Wikipedia better and stop arguing past each other with artificial support/oppose comments. VanTucky, we're still in the "develop the proposal phase", not yet to "solicit everyone's opinion for a yes/no vote" phase. The Foundation hired someone because (this discussion notwithstanding) that's what English Wikipedia and other projects asked for.--ragesoss 06:27, 24 August 2007 (UTC)
- Please strike this comment. If anything could move me to permanent and unconditional opposition, it is this sort of declaration of a "wave of the future". Where did English Wikipedia ask for this, anyway? Septentrionalis PMAnderson 23:00, 24 August 2007 (UTC)
- Putting it mildly, your idea that the arguments for doing it "greatly outweigh" the arguments against, is not exactly sterling in character. This constant hammering of "we're still in development, stop making comments in opposition" needs to cease. It is not your job to tell others when they should or should not make comments pertaining to the subject at hand, and how they should make them. I don't think a discussion about how "this can best be used to make Wikipedia better" is in order, considering it is my opinion that this cannot, in any fashion of implementation, do so. VanTucky (talk) 06:38, 24 August 2007 (UTC)
- Even with the current version displaying as the default (as the proposal currently reads), you think it worthless?--ragesoss 06:53, 24 August 2007 (UTC)
- (edit conflicted) It is going to happen. But it will happen because (in those ad nauseum discussion for it) the arguments for doing it greatly outweigh the arguments against, and sooner or later we'll hammer out a solid consensus proposal on how to implement it. It won't happen against the collective will of Wikipedia, but with it. AmiDaniel has been around long enough to know a discussion that's on the wrong track when he sees one, and he's confident (as am I) the things will get back on track once we start talking about how this can best be used to make Wikipedia better and stop arguing past each other with artificial support/oppose comments. VanTucky, we're still in the "develop the proposal phase", not yet to "solicit everyone's opinion for a yes/no vote" phase. The Foundation hired someone because (this discussion notwithstanding) that's what English Wikipedia and other projects asked for.--ragesoss 06:27, 24 August 2007 (UTC)
- VanTucky, if nearly all active participants are in the 'class of users' then it can't, logically, be called an 'elite'. The idea behind 'sighted' is that everyone who positively contributes to the project will get it in short order. Only people who have no provable record of positive contributions (or a proven record of negative contributions) would be 'excluded'. Basically, that means dynamic IP addresses, new user IDs, vandals, and the like. In that sense it is no different than the current restrictions on semi-protected pages, article creation, or page moves. We already have this 'other' class of users, and we have had it to varying extents since Wikipedia was created. The 'quality' setting will be held by a smaller group of users, but its purpose is to identify versions of pages which community consensus holds to be of high quality... the access is restricted to prevent just anyone from being able to falsely say a particular page has been approved by the community. Thus, the people who have it don't receive any 'special privilege' in debates - they're just agents acting on behalf of community decisions. Finally, nobody loses the ability to 'edit the page right now'... indeed, users who currently don't have that will be GAINING it. Right now an IP/new user can't edit a semi-protected page at all. After this is implemented they will be able to do so immediately. Yes, there will be a delay before that edit is incorporated into one view of the article (it will be immediately visible on the current view), but if it is a good edit it WILL be... and that's obviously an INCREASE in their editing abilities. Ditto fully protected pages. When this change is rolled out there will be far less need to protect pages against vandalism... meaning that all users will be able to edit these pages that they are currently blocked from contributing to. In short, I don't think 'the negatives are outweighed by the positives' so much as I don't agree that the negatives are even accurate. This feature will increase the ability of all helpful users to edit Wikipedia by replacing our current protections that hit everyone with delays that hit only those we don't know mean well. --CBD 10:51, 24 August 2007 (UTC)
- I wanted to be quiet here, but I can't stand the way that some people assume to talk for 'the community'. The problem with special class users is that they don't implement the will of the community, they arbitrate and decide the will of the community. Just like this discussion is assumed to have a foregone conclusion. I just heard about these proposed changes, and I assume I will be lumped into the category of those who " vocally oppose anything" because I oppose this, even though that is not the case. I understand that since I'm a newer user my opinion won't be valued as much, but I never understood just how stratified wikipedia was until now. It really destroys lots of the magic for me. I think there should be a requirement for a certain number of users to vote on a proposal like this, and a greater than 60% rate of acceptance. Of course, as everyone has already pointed out that this whole thing is a foregone conclusion I don't even know why I am bothering to say anything at all. I just don't get the bollucks argument that this will increase access, since changes won't be made visible until they are vetted! I find this argument to be very misleading. Increased access doesn't mean having to wait for weeks to see your change included in the public version. At least semi-protected versions only apply to a few articles. Moonbug 07:52, 26 August 2007 (UTC)
"Has some depth (not a one-line stub)."
Not being a one-line stub is listed as a requirement for being flagged. Later, we're told that "[n]ew pages created by Surveyors can be auto-sighted as well." So let's say an administrator decides I'm suitable to be a surveyor, and the next day I create a one line stub. Would it be dereliction of my duty to not remember to go desight the page which has been automatically flagged? It may seem trivial, but it will come up eventually. Picaroon (t) 04:25, 24 August 2007 (UTC)
- Since the idea that stubs fail the requirements makes good sense to me, I think it is a better idea to scratch the automatic sighting of articles created by surveyors. as a side note: what's with the arbitrary caps on the term surveyor? That doesn't exactly passify the concerns of those accusing the proposal of creating more hierarchy and excess bureaucracy. VanTucky (talk) 04:35, 24 August 2007 (UTC)
- All stubs fail the requirements? What about longish ones, like Idris Legbo Kutigi? Or do you just mean the one-liners? Picaroon (t) 04:54, 24 August 2007 (UTC)
- Sorry for the generality, I just meant one-line dicdef-type ones. VanTucky (talk) 05:00, 24 August 2007 (UTC)
- Good point, I've separate autoreviewing on updates and on create in the software. Voice-of-All 09:40, 24 August 2007 (UTC)
- Personally I think that we should allow one line stubs to be sighted and, like any other edit, do so by default if the user creating it has the privilege. The idea is to identify pages as 'believed to not contain bad information'. Short stubs contain limited information... but still information about the subject. It is better to give a user a page saying, 'Teletubbies is the title of a children's television show and a descriptive name for the characters on it' and hope they expand it than to show them nothing at all. The one line stub tells them what the thing is. That's inherently better than telling them nothing. --CBD 11:00, 24 August 2007 (UTC)
- I agree, stubs and oneliners should be allowed to be sighted. As one of the editors in the WT:WPSPAM and WP:COIN projects, it would be great to just stub down non-neutral articles (those that are notable, but advertising or non-neutral), and then sight that stub. The non-neutral editors can then do what they want, there is a neutral stub that is sighted and which is seen first, and only when another reviewer sights a newer (neutral) version, the stub will stay. --Dirk Beetstra T C 08:39, 31 August 2007 (UTC)
- I also think sighting stubs should be allowed. This will allow us to put wiki's best foot forward, even for the smallest articles. --Falcorian (talk) 17:35, 31 August 2007 (UTC)
I have removed the requirement that it "Not be a one line stub", as it seems people here at least are in favor of it, and no one has spoken against it. --Falcorian (talk) 18:05, 2 September 2007 (UTC)
Exciting stuff
I heard about these changes from Jimbo Wales several months ago. I'm excited that they are possibly soon going into effect. It's an interesting development, and the Quality versions' flag could be very useful if there was a caption to go with it. You could note that this Quality version was referenced by such & such publications. I'd definitely participate in checking off sighted versions for the articles on my watchlist. Roll-out to English Wikipedia ASAP, please.-BillDeanCarter 09:20, 24 August 2007 (UTC)
- There should be benefits for both readers and editors. Wikipedia's readers outnumber its editors by far. If more readers are attracted to Wikipedia, and that happens every day, then more of them will eventually become editors, even if at first they dont edit, or even realize for a while that they can do so, by clicking on whichever button, when they first stumble upon it. More readers, more editors, better project. Newbyguesses - Talk 00:23, 29 August 2007 (UTC)
Analogy with software development
Think of how a group of researchers collaborate to publish their results. Or, better, how a large group of developers write software. You create a first version, make it available. Then continue to improve it till a further releasable version is reached. Then make it public. Then continue the process. --Knverma 09:47, 24 August 2007 (UTC)
Communication and contingency plans
I'd be much happier about this if there was clear communication about this from the developers to the community, and from the Board to the community. Sometimes, shall we say, the communication levels go up and down. As the proposed implementation gets closer, I would like frequent, regular and reliable updates, even if it to say "no more news yet", along with a clear timeline that stays accurate and is updated. Also, I think there should be advance planning for contingency plans - what the plans are if things go wrong, if there is an uproar in the community, and how easy it is to reverse some of these changes. If things do go even slightly wrong, people will be more easily reassured if they can see evidence of clear planning and well-written, clear explanations, plus links to discussions. So, who are the people that will be doing this? Carcharoth 11:47, 24 August 2007 (UTC)
Comments by FT2
Overall, sighted/qality versions have the potential to give some much-needed stability, without disrupting editorship. They also can help to ensure that articles do indeed increase in quality (rather than having no prior standard to be measured and subordinated to, if their quality falls). There are a few concerns and potential problems, I'd value comment on, if not already covered in the many discussions:
- I don't like the idea of surveyors' edits automatically being sighted. A surveyor as much as any other user may make edits they know are not "release ready/consensus" (BRD for example). Autoflagging is likely to cause edits to be flagged that shouldn't be. Automatic sighting is the wrong default - it should be a positive decision for a surveyor to decide a given edit (of theirs or any editor) is a sighted version.
- I have grave concerns about dispute potential of surveyors being able to "sight" their own changes or articles they edit on, thus effectively giving "their" version enhanced standing over other versions. (Similar to page protection if a sysop was allowed to edit a page then protect it.) Reverse problem - suppose some surveyor on the article disagrees with the sighting and thinks its a bad call - there's a problem if they are allowed to mark the current version as unsighted (reverting the most recent sighted version to the previous sighted version), then edit the article's current version too (COI issues). How do we ensure we don't create a new low-grade form of wheel war?
- Special pages such as "whatlinkshere" and "categories" that are created "on the fly" may be problematic. Suppose X sights a version of Tort, and Y then adds a wikilink to Legal terminology and the category Legal terms to the article. Z then comes along, reads the category page Legal terms (or does a "whatlinkshere" on the article just linked). Will they see Tort listed or not listed in these pages?
- If a page is deleted, or moved, what will a reader using sighted versions see?
- If a page has templates such as AFD on them, or some factoid is tagged as [citation needed], readers seeing the stable version will not be aware of this. They may not see the AFD debate or latest version being debated (although this is open to all users), and factoids may be deleted for not having been cited for some time, because readers were now unaware a cite was being sought.
- Will sighting apply to project and policy pages too?
A few edges that need clarifying... FT2 (Talk | email) 12:16, 24 August 2007 (UTC)
- OK, here are my responses at least:
- Sighted changes are meant flag revisions that are not vandalism or BS. It is not a high quality premium stamp of approval. Otherwise, it would face issues of scalability.
- See 1) above ;)
- If either the stable or current version link it it, WLH and the like will show it.
- If deleted, it will be like any other deleted page. If moved, then the new location has the stable versions.
- Readers don't care about some AfD tag cruft. Users will see the current revision be default. Though revisions with cite and AfD tags can be sighted though...heh.
- Not formally, but just like editing, people can integrate it. See this list provided by the software. It is intended to help people sort by article/stub classes to see what work is needed in the areas they like editing in. Projects could use this. Voice-of-All 12:27, 24 August 2007 (UTC)
- I'm not sure I agree with "readers don't care for AFDcruft". Readers are exactly the people who we rely upon to comment - not all do but many become interested or have an interest (or would if they knew). Articles are tagged for reader attention too, to ask for cites to be added, sections expanded, and so on. We also look to readers to respond to these tags and become contributors, not just to editors and account holders.
- On the question of sighting/unsighting, are you really saying that surveyor/sighting disputes wont be a big source of trouble? The power to label a given version as "sighted" is the power to choose which version 95% (or some large proportion) of Wikipedia readers will see and read.
- Yes, we can remove surveyor tags, but in its own way surveyor rights are immensely powerful. They are not merely a means to flag non-vandalism; they are a means to choose which of several versions a visitor sees by default -- which means in most cases the only version they see. We don't let admins edit or fix an article then protect it as the current version (even if their edits would be neutral and an improvement), and yet we trust admins more than surveyors. I think it'd be sensible to have a "request for sighting" page and a rule that surveyors cannot sight article revisions they have worked on (since the last sighting anyhow). The drama, DR avoidance, POV issues it'd avoid would be immense, the removal of temptation would be huge, and if this right is handed out liberally, then surveyors to check others articles will be very easily found. Perhaps we could encourage surveyors to have a set of articles they don't edit, or to have a template on the talk page top that surveyors familiar with this page can list themselves as able to be left a message to re-sight it? So that most sighting is done without request? FT2 (Talk | email) 13:43, 24 August 2007 (UTC)
- I'm not saying that citation tags are cruft though, but I can easily see those getting reviewed with the page anyway. And I don't think it is that powerful. Right now, any user can control the version that %100 percent of readers and editors see. The only thing that would make the right "powerful" was if it as only given to a small few, which is not part of the plan. It wouldn't be powerful if almost everyone had it. Voice-of-All 13:48, 24 August 2007 (UTC)
- Yes, we can remove surveyor tags, but in its own way surveyor rights are immensely powerful. They are not merely a means to flag non-vandalism; they are a means to choose which of several versions a visitor sees by default -- which means in most cases the only version they see. We don't let admins edit or fix an article then protect it as the current version (even if their edits would be neutral and an improvement), and yet we trust admins more than surveyors. I think it'd be sensible to have a "request for sighting" page and a rule that surveyors cannot sight article revisions they have worked on (since the last sighting anyhow). The drama, DR avoidance, POV issues it'd avoid would be immense, the removal of temptation would be huge, and if this right is handed out liberally, then surveyors to check others articles will be very easily found. Perhaps we could encourage surveyors to have a set of articles they don't edit, or to have a template on the talk page top that surveyors familiar with this page can list themselves as able to be left a message to re-sight it? So that most sighting is done without request? FT2 (Talk | email) 13:43, 24 August 2007 (UTC)
- The big issue, though, is one of scope. There NEEDS to be a relative number of surveyors compared to how many articles use the system, otherwise the backlog would just get too deep. Look, I think the main reason for this is to cut down on "ERIC IS A FAG" and page blanking edits...if most readers won't see the blanking, it'll be worthless to the vandalizers, AND remember that any edit can still be reverted even if marked good -- I don't see edit wars, etc, really becoming any better or worse, FWIW. ♫ Melodia Chaconne ♫ 13:54, 24 August 2007 (UTC)
- It'll get warred over. Oh gods, will it ever............
- User 1 edits page. User 1 sights page. User 2 removes addition ("NPOV"). User 2 resights page. User 1's buddy user 3 re-edits page, adds talk page comment "Do not unilaterally break consensus" and resights page. User 2 removes ("reversion of POV, see WP:NPOV") and re-resights page.....
- Multiply by 10,000 articles, obscure and well known...
- Please for the love of all that's good don't let people sight their own edits. Thats a simple and obvious point with few/no drawbacks, if surveyor rights are widely spread. :) FT2 (Talk | email) 14:01, 24 August 2007 (UTC)
- How would that be any worse than edit wars now?Zginder 14:14, 24 August 2007 (UTC)
- I get thought about how I first started. My first edits started a mini edit war, I was trying to make an article neutral. Then I discovered the POV tag, so a tagged a article so that any reader would know that the article was pov. I tagged the article for the reader as much as the editors. Would I have continued editing and been here now, if no one saw the POV tag, because the the non-neutral version was sighted? Zginder 14:29, 24 August 2007 (UTC)
Edit wars exist and probably always will. You try to minimize them, and minimize things that encourage or facilitate them. You also consider removing things (or not adding them) that are likely to add to them but don't bring enough benefit to merit it. FT2 (Talk | email) 16:33, 24 August 2007 (UTC)
Eh?
I see I'm not the only one to reach this proposal and not understand it. Perhaps the Proposal could be reworked by those who do understand it, so that people don't need to come here to ask what it means. --Dweller 13:08, 24 August 2007 (UTC)
- Agreed, the proposal is incomprehensible at present. To the extent I comprehend it, I think it will be a net negative for the project. GRBerry 15:12, 31 August 2007 (UTC)
- I think the only way to really understand this would be to play around on the test wiki. A more full scale public test which will be made available before any decisions are made on this. Even so, I think the only way to really understand the net effect would be to experiment on the semi-protected pages. There is so much speculation about whether people will be discouraged or not, but the only real way to find out is to try this out and read what they write in the slashdot comments. --Merzul 17:31, 31 August 2007 (UTC)
Suggested words to replace "sighted/unsighted"
I am not the only one to think "sighted" sounds just plain odd. Here are alternatives I thought of, largely during my waking hours this morning. (In no particular order, some of which I think aren't much better than "sighted/unsighted").
- vetted/unvetted
- externalized/unexternalized
- inspected/uninspected
- approved/unapproved
- authorized/unauthorized
- published/unpublished
- propagated/unpropagated
To clearly convey the idea of "observed by authority", perhaps...
- observed/unobserved
- discovered/undiscovered
- uncovered/covered
- seen/unseen
- spotted/unspotted
- revealed/unrevealed (unhidden/hidden)
- unconcealed/concealed
- publicated/unpublicated
(The last one satisfies the suggestion that the word be deliberately contrived and unfamiliar).
"Sighted" is largely bad because of possible confusion with "cited" (not to mention "sited"). However, it's also bad because its use as a stand-alone adjective tends to be avoided because it might be understood as "not blind". Note that "spotted" is similarly bad because it can mean "possessing physical spots" (e.g. a leopard).
It is probably best to use a base word, with an "un-" negation. (But using a "non-un" word from each column could be considered; that's what I meant to indicate with the suggestion of "hidden"). It's probably also best to use a base word with a natural verb (and probably a word which is actually a verb itself). There must be lots of other possibilities which I haven't listed. --SportWagon 14:07, 24 August 2007 (UTC)
- If the suggestion is that "the word be deliberately contrived and unfamiliar", then surely the way of choosing a pair of terms that try to minimize any pre-existing associations with existing English words is to aim to use a deliberately made-up word, and so consult a friendly Experimental Psychologist/Cognitive Psycholoist who knows various bits of research that might inform such a choice? I am no longer active in those areas myself, but I was for o0ver 25 years, and I do still know enough to realise that if one really wants to eliminate unwanted pre-existing associations, then this might well be a good way forward to consider. (I did think using a word made p from an acronym of "obba" and "unobba", derived from "Observed By Authority", but this seems to me to suggest "observed" a bit too much.) DDStretch (talk) 14:30, 24 August 2007 (UTC)
- I think that published/unpublished is the strongest analog to what is actually being done. I think this may be premature though, we need substantive policy before we start quibbling over terminology. Adam McCormick 16:33, 24 August 2007 (UTC)
- I was and am not suggesting that "the word be deliberately contrived and unfamiliar", but someone else seemed to do so. [2] [3] Therefore I tried to coin such a possibly suitable word. I, personally, would be inclined to go with a "natural" (therefore "familiar"?) choice.--SportWagon 17:37, 24 August 2007 (UTC)
- What about "proofed" or "proofread"? As I understand the proposal, it's basically equivalent to what a printer or typesetter would do when looking over a proof copy of a printed work -- checking for obvious errors and misprints, without necessarily doing the more thorough investigation that would be expected of an editor or fact-checker. --benadhem 02:07, 25 August 2007 (UTC)
- I think "inspected" is a good choice. It conveys both the idea that it's been looked over to some degree and the idea that it may not be perfect and that there may be mistakes. Pyrospirit (talk · contribs) 15:21, 25 August 2007 (UTC)
- I have an issue with "inspected"- to me it implies, those who have been appointed (usually available because they cannot actual succeed in doing the job) criticising those who actually can. Those who blythely destroy all that is good because it fails one artificial tick box.Ofsted. It may be a good choice of word- but if it, then the change is not benign and Wikipedia should avoid this system with a bargepole. ClemRutter 21:30, 28 August 2007 (UTC)
If changing the status of the revision comes with "surveyor rights", why not use "surveyed" (and "unsurveyed")?--SportWagon 21:01, 28 August 2007 (UTC)
Critical features
IMO there are several specific features which any implementation of flagged revisions either must or must not have in order to be useful and viable;
Must have
- The ability to flag revisions granted automatically by the software after a short period of activity. Without this the system will not scale and will have backlogs on both granting the access and flagging revisions. My strongly preferred criteria continues to be something like 'must have made edits to articles on 30 different days'. Any 'total days and/or total edits' criteria will be vulnerable to blitz attacks where a vandal can easily get multiple accounts 'authorized' simultaneously... and efforts to discourage that by raising the counts will impact legitimate users more than vandals.
- Logged in users see most recent version. Non logged in users see most recent flagged version, or most recent version if none are flagged. All users see the most recent version after saving an edit. Initially nothing is flagged, so all users see the most recent version just as they do now. This system will allow protection and semi-protection to be removed from all the pages which are protected to deter vandalism. Instead all users can now edit the page and see their updates immediately - but edits by very new users and IPs are not shown to casual (not logged in) readers until looked at by a regular contributor. Having edits delayed for default viewing by some users is less restrictive than not being able to edit at all.
- Special page, bot, or the like which lists pages for which there are unflagged revisions more recent than the last flagged revision. This could be reviewed as 'recent changes' is now and, if all active registered users have access as intended, should mean that all pages are kept in a continually flagged state after a short period for review.
Must NOT have
- The ability for users with access to flag revisions to give it to other users. Vandals would play nice with one account until they got the access and then start handing it out to anyone and everyone... including dozens of sockpuppets. Removing access from everyone who the vandal gave access to (and everyone THOSE people gave access to... and everyone THOSE people gave access to... and...) would hit tons of innocent people. Not doing so would leave vandal sockpuppets free to continue handing out the access. Makes the system worthless for controlling vandalism.
- Requirements for 'first level' flagging of revisions which cannot be easily met by the majority of users. For instance, we cannot have a criteria that 'all information on the page be verified against reliable sources'... because a user who comes along to add a missing comma isn't going to have several hours to check that everything on the page is verified and thus will not flag the page. Nor will the next person who wants to add a verified fact, but not check everything else. Nor the next. Et cetera. Even if someone does eventually take the time to review and approve the entire page - once people start making unflagged revisions the problem starts all over again as each doesn't have time to go and fact check what was done by the one before. The criteria have to be things like, 'there is no obvious vandalism, there is no obviously defamatory uncited info about a living person, et cetera'... things which people can determine with a very quick review. Otherwise there will be an insurmountable backlog on unflagged revisions. More restrictive criteria should be applied to a 'second level' flagging to identify pages which are very high quality, like our current 'featured article' system.
For flagged revisions to work it has to be something that is easily kept in a constantly updated state. Users should seldom see anything other than the most recent version of the page because most changes should be flagged, and the unflagged ones should be reviewed and approved or reversed quickly. If positive contributions sit unflagged and unreviewed for days at a time then the system is not working. The system should work such that Wikipedia seems almost exactly as it does today EXCEPT that pages are protected/semi-protected much less often and vandalism is seldom seen by casual non-logged in readers. --CBD 14:45, 24 August 2007 (UTC)
- I completely agree with the above, although I might add an additional Must have:
- 4. The ability for users to flag their own revisions.
- I think this is a critical feature needed in order to keep the number of unflagged revisions low. Note that if most regular users are able to flag revisions, and can flag their own revisions, then for these users there's effectively no difference from today except rather than reverting visible vandalism most regular users would be working on adding content (and occasionally reverting generally invisible vandalism). -- Rick Block (talk) 15:02, 24 August 2007 (UTC)
- Would this be the default setting for registered users? and if not, would there be a way to set this, as there is a way to set "do not save w/o edit summary"? Septentrionalis PMAnderson 17:06, 24 August 2007 (UTC)
- I would make this another check box below the edit summary field (like "This is a minor edit" and "Watch this page") and, by default, have it unchecked. There could be a preference item to auto-check this field, but I'm not sure this is either necessary or a wise idea. Adding a "prompt me if I haven't set the flagged box" preference item seems like a frill, but might be nice (and would certainly be safer than an auto-flag preference). -- Rick Block (talk) 17:52, 24 August 2007 (UTC)
- I agree with the "must" and "must not" points above, including nr.4 suggested by Rick Block. If established editors cannot sight their own edits, problems might arise with certain editorial procedures that involve multiple pages (mergers, for example). Sighted pages involved in a merger should be sighted again immediately afterwards; otherwise inconsistencies can arise in the "public" version: Anon users might see the redirect placed on the source page (which may never have been sighted), but miss the content on the target page (which has not yet been sighted after the merger). All this becomes terribly complicated if it needs two people to perform the merger completely (one who merges, one who sights). --B. Wolterding 18:25, 24 August 2007 (UTC)
- I agree with all of these "must" and "must not" points above. If the system is absolutely going to be implemented, these seem like reasonable guidelines. Moonbug 08:05, 26 August 2007 (UTC)
Accuracy/Depth/Readability
On AmiDaniel's test wiki there are drop down boxes for Accuracy, Depth and Readability. Will Wikipedia have a similar system, or will it just be a yes/no on whether or not it's been sighted? I hope the latter, a more complicated system is unnecessarily confusing. --Tango 17:47, 24 August 2007 (UTC)
- I believe it will likely be the latter, though it's not really certain at this point. I'll try later on to make the specific configuration on the testwiki reflect the desires of the proposal so we can see better what we're dealing with. The extension is, however, highly configurable -- even in its currently immature state -- so just about anything we want we can have. Don't be misconstrued into thinking that the way it is on my testwiki is the way that it absolutely has to be. AmiDaniel (talk) 18:52, 24 August 2007 (UTC)
Interaction with templates, css files, etc.
Any thoughts regarding how this feature interacts with templates and other transclusions, and css and other similar sorts of files? Perhaps difficult, but I'd expect a change to a template transcluded by a sighted page would not be visible on the page until either the change to the template is itself sighted (if the template is marked as a page requiring sighting) or the page is resighted (if the template is not so marked). This seems to imply the need for tranclusion versioning. It might be simpler to globally change the transclusion mechanism so that any transclusion of a sight-required page (the transcluded page, not the page doing the transcluding) transcludes only the most recently sighted version, although doing this would make testing a change to a sight-required template kind of cumbersome (the change would not really take effect until it was sighted - perhaps this simply forces use of sandbox templates for trying out changes which is arguably a good thing). In fact, with this implementation and the ability to enable the entire feature by namespace, we could try it out in template space first to see how it scales in practice - potentially unprotecting some or all of the high use templates. I assume this would be implemented in part by not adding the change to a sight-required template to the job queue until the change is sighted. -- Rick Block (talk) 18:46, 24 August 2007 (UTC)
- As it seems to be written now, the sighted version of a page will display all templates as they were at the time of the article's "sighting." I'm not sure if this would be my preference, however, as I can foresee this causing a lot of headaches. AmiDaniel (talk) 18:54, 24 August 2007 (UTC)
- Right. I suppose I can configure to fetch the current ones if desired. Voice-of-All 18:51, 25 August 2007 (UTC)
- Hmmm. So marking templates as requiring sighting effectively does nothing? -- Rick Block (talk) 19:14, 24 August 2007 (UTC)
- I agree with transcluding the most recent sighted template, though not with idea that this would permit high use templates to be unprotected. These templates don't need to be edited all that often and it would be too tempting for vandals to gain sighting permissions (we really need a better term than "sighting") and wreak havoc. Chaz Beckett 19:39, 24 August 2007 (UTC)
- I think this depends somewhat on how revision flagging is viewed/implemented. I see the first, 'sighted', level as being a continuously maintained thing... every unsighted edit subsequent to at least one sighted revision on a page is quickly reviewed and approved or reversed. In such case there really isn't much difference in how templates are treated except that if templates are displayed from their current state then a vandalized template would show up immediately on an otherwise 'sighted' page. Showing the most recent 'sighted' version of templates could address this, but only if we required that all templates be sighted before being transcluded. Or we could just keep the outside elements 'frozen' for the short time between unflagged updates and review thereof. Where this becomes a bigger issue is if flags don't get updated for a long period of time... such as in the second 'quality' level. There a page might get marked as being completely reviewed/high quality and then go weeks without that status being updated again. In such case I think templates, images, CSS, and the like have to be 'frozen' in the state they were when the flag was placed... otherwise you might be trying to display the contents of a template which has since been replaced and deleted. It would hardly seem like 'quality' to have red links to deleted templates/images and/or new CSS which doesn't mesh with the formatting originally intended.
- Thus in both cases I think it would be better to keep these 'external' items 'frozen' in the appearance they had when the page was flagged. What drawbacks would there be to this? I was thinking of the 'date sensitive' templates on the Main Page, but those are actually keyed to call entirely different templates as the day changes... so unless the date/time magic words and parserfunctions are also 'frozen' on a sighted page those should continue to update and point to different sub-page each day. --CBD 12:29, 29 August 2007 (UTC)
Instead of different versions for different users, show ALL users the sighted version plus a diff list.
What if all users just see the sighted version, with a list of "pending changes" below it? And users with "sighting" ability would also get an "approve changes" checkbox, or something along those lines. It might not be quite as strong a deterrent against vandalism (since vandalism would still be immediately visible), but at least it would segregate potential vandalism apart from the rest of the article without hiding potential improvements from anyone. --benadhem 19:12, 24 August 2007 (UTC)
- One problem is that the diff list is confusing for many readers and editors. I know a couple people who edit occasionally who have no clue what those green boxes they see sometimes at the top of the screen are.
- BTW -- I think if the most recent version is also the most recent sighted version, there should definitely not be any "There are no unreviewed versions" text or anything like that at the top of the page. I suspect (and hope) that in general, the latest version will be sighted for most (non-FA?) pages, so we won't need anything like that. I also hope that we can have two special pages, one for greatest number of unsighted revisions, and one for pages with the oldest unsighted revisions. I know I'd patrol the second one. -- Myke Cuthbert (talk) 21:54, 24 August 2007 (UTC)
- I guess the Pending Changes list could just as easily show the final format of the changed paragraphs instead of the side-by-side highlighted source text. The affected paragraphs would just need to be scanned for vandalism and typos; it wouldn't be critical to see exactly what was changed. --benadhem 22:32, 24 August 2007 (UTC)
Possible compromise between sighted vs. "live"
One option would be to have the editable version be the default one viewed, with a moderately-visible banner at the top offering the reader to view the most recent "sighted" version, accompanied by a check box ("Always view sighted version").--Father Goose 20:37, 24 August 2007 (UTC)
- Now that I've cooled down from yesterday...I thought of another possible compromise: more domain names. Just as there exists simple.wikipedia.org, if sighted versions are the default, we could have "current.wikipedia.org". Conversely, if the current version is the default, we could have "sighted.wikipedia.org", or "verified.wikipedia.org", or something. This seems like it might be a better option to me, so that those who want to avoid seeing vandalism could use that URL; but that's the wrong debate for this section. -- Ken g6 21:28, 24 August 2007 (UTC)
Rethinking the System
If it was implemented, It would mean the complete redo of all systems, and creating a backlog so high that manually fighting vandalism would take less time. This would also mean that : 10000 hours of work goes into waste. And that before we get used to the new system, that would leave us virtually defenseless for a matter of months. Imagine wikipedia without IRC, without Martinbot or RC Patrollers. It would be a big mess and wikipedia would fail within a year. All experienced editors would have to start learning the program again. Flagged revisions is giving Wikipedia to the vandals and throwing it into Bronze age again, creating what would make newspapers laugh at us. An unterminable backlog of over6,875,496 edits would be created and getting bigger every minute. -FlubecaTalk 21:35, 24 August 2007 (UTC)
- Huh? I don't follow. Reading through articles for readability and obvious errors isn't a waste of time, it's something that very much needs to be done. And the "backlog" articles that have not been flagged won't be an increased liability; they'll behave just as they do now.--ragesoss 21:41, 24 August 2007 (UTC)
- But we can do that already, and also Wikifying is not a waste of time, there is still a huge backlog of it. It's not what we think is good, it's just what we can do, in a period of time. "If it ain't broken, don't fix it.-FlubecaTalk 21:45, 24 August 2007 (UTC)
- I'm not sure what you mean, Flubeca. Why would we be left without IRC, Martinbot, RC Patrol, etc? Why would we be defenseless, and how would this be throwing WP back to the Bronze Age? Jpers36 22:40, 24 August 2007 (UTC)
- I'm confused. If we take all the folks who do RC patroling now, and instead they are looking at the sighting of articles it will actually increase the amount of people reading and fixing articles. Right now a fair bit of RC stuff is just look at the diff, revert. in the new system it would be look at the 5 diffs, revert the vandalism, check that the others are good, sight to anon view. How would that go back in time? --Rocksanddirt 22:57, 24 August 2007 (UTC)
- I have know idea. Not to mention that the incentive to vandalize will increasingly evaporate. Voice-of-All 00:12, 25 August 2007 (UTC)
- And this will not be automatically applied to every page. It has to be manually applied, page by page. We won't automatically start with 1.9 million articles. Since this won't apply to many (including most stubs), I would estimate the eventual number (after a few months or years of surveying) would be far less than 1 million articles depending on how much we decide to use this. And we would not have to "start learning the program again." editors do not have to flag revisions and even for those that do, it is learning a new tool not a radical software change. editing pages will still be the same, there may just be an extra step for sighted articles. Mr.Z-mantalk¢ 01:19, 25 August 2007 (UTC)
- I have know idea. Not to mention that the incentive to vandalize will increasingly evaporate. Voice-of-All 00:12, 25 August 2007 (UTC)
- I think as long as scripts are developed for fast "sighting" (or "inspecting," which I think is a better term) of articles, similar to the current scripts for fast reverting, tagging, warning, etc., it won't cause any more work for us than RC patrol is now. Of course, it'll take some time for script developers to implement this stuff, but in the end it might be worth it. Pyrospirit (talk · contribs) 15:30, 25 August 2007 (UTC)
Abandon sighted versions
Comments
- Sighted and quality versions
I think that the distinction between "sighted" and "quality" versions is an artificial one. The proposal states that in order for a page to be tagged as "sighted", it must be clear of vandalism, of libel, of unsourced statements about living persons, of unencyclopedic content. In addition, it must have been around for a few days, must have been checked for accuracy and prose, and must not be tagged for cleanup.
Given the above requirements (especially those regarding absence of vandalism and unencyclopedic content), an editor must, prior to tagging an article as "sighted", read the entire article, check every single source, and ensure that the article meets the content criteria of WP:NOT, WP:NPOV, WP:NOR, WP:BLP, and WP:V. This is not something that can be done during the course of recent changes patrol. This requires a serious commitment of time and effort for every article. Vandalism (especially in the form of false information and hoaxes) and unencyclopedic content (e.g. original research and trivia) is not always easy to spot or correct. It will be impossible to tag an article as sighted without checking all of the sources and removing all unsourced content.
- People will cut corners
Let's assume that a surveyor tags an article as "sighted" but does not actually ensure that all content is supported by a reliable source and does not ensure that all content conforms to Wikipedia's content and relevant style policies. (This will happen. People cut corners or satisfice all the time, and it's not always in bad faith.) If subsequent editors only check for changes since that revision, any old problems automatically be missed. However, we will be giving the content of the article an artificial legitimacy.
Questions
Given the above, I have a few questions:
- What is the point of "sighted" versions if (a) sighting requires a large investment of time, (b) people will cut corners, and (c) there is little or no oversight.
- Who checks the accuracy of tags applied by surveyors?
- What should be done to surveyors who tag (not necessarily deliberately) articles as "sighted" when those articles have one or more problems?
- What should be done to administrators who give "surveyor" rights to editors who tag articles as "sighted" when those articles have one or more problems?
Suggestions
If this idea is implemented, I am of the opinion that we should abandon the idea of "sighted versions" and go directly to "quality" or "confirmed" versions. That is, we should check the accuracy of articles and ensure that they meet relevant content policies and guidelines.
Since we cannot expect that every editor will be aware of all relevant policies and guidelines, nor that they will be rigorous in their fact-checking, I also propose that we impose relatively strict conditions as to who can tag an article as "quality". Since a reviewer should be someone that we trust to (a) know Wikipedia's policies/guidelines, and (b) implement them, I think it would make sense to make this an additional "tool" of adminship.
I also do not believe we should limit "quality"/confirmed tagging to FAs and GAs only. Confirming the accuracy and encyclopedicity of short articles is substantially easier to do and many short articles could be identified, fixed, and tagged in such a manner.
Conclusion
In short, I oppose the current form of this proposal, mostly because it is a half-measure. It creates a tremendous amount of work and bureaucracy without directly pursuing the goal of full verification. — Black Falcon (Talk) 23:40, 24 August 2007 (UTC)
- I don't disagree with the evaluation based on the way you phrased the intent of "sighting." However, I didn't get the sense that the "sighting" was going to be a GA review, but rather just that vandalism, Gross BLP violations, Gross violations of OR and Verifiability were the target. If what you suggest is the case, then the "sighting" program is totally unworkable and should be abandoned (IMO). --Rocksanddirt 23:46, 24 August 2007 (UTC)
- Your analogy is a good one, I think. The current "Requirements for Sighted pages" section effectively requires a detailed point-by-point review of the article. Much vandalism cannot be spotted through a quick one- or two-minute check (spotting and correcting original research and POV problems takes even longer). If the sole point of sighting is to spot and remove something like Wikipedia SUCKS ERFISP SD*SBC KLJnj-8unrfas8-d(, then I really don't think it's worth the extra work and bureaucracy. — Black Falcon (Talk) 00:18, 25 August 2007 (UTC)
- Unless it was changed recently, I don't see where the proposal says we have to check that every statement is sourced and that all sources are reliable (with the exception of BLPs, but we technically should be doing that already). All it says is that "Introduction and major statements checked for accuracy/cited." - my interpretation of this is that they should be correct in the context of the article and verifiable. I would also hope that editors would not just choose random articles to go around and "sight", that they would pick articles they at least have some background knowledge on or were involved in writing. Mr.Z-mantalk¢ 01:27, 25 August 2007 (UTC)
- I was referring to the "Requirements for Sighted pages" section. Ensuring that an article is clear of vandalism, unsourced statements about BLPs, and unencyclopedic content requires thoroughly evaluating the article for conformance to key content policies on verifiability, original research, NPOV, BLP, trivia, WP:NOT, and so on. I'm not sure if that is a recent addition. — Black Falcon (Talk) 01:37, 25 August 2007 (UTC)
- I agree with Black Falcon. The way the requirements are formulated is stronger than what I gathered to be the intention. I thought that somebody familiar with the subject should be able to sight an article by reading through it, but "Introduction and major statements checked for accuracy" means a more careful review, taking at least ten or fifteen minutes even if you're familiar with the topic. I fear that would definitely lead to backlogs and edits not disappearing for months. I thought that it would be enough to check that the article is plausible. -- Jitse Niesen (talk) 02:56, 25 August 2007 (UTC)
- If that is how it is being interpreted, then the requirements should be rewritten. I don't think that the intention was to make this into GA review (which can often take weeks for a result because few people want to do it) but rather an assurance that the article is decent and presentable. The way it should work is: Someone sees an article that they think may be "sightable." It is fairly well developed, and they have some background knowledge of the subject. They check to make sure it has some sources and that some of the sources are reliable. They read it to check for obvious spelling, grammar, style, and factual problems (or templates indicating some). They check to make sure it is not in clear violation of policy such as [{WP:NOT]] and WP:NPOV (clear violations are generally easy to spot when reading). They check the history for recent vandalism. Finally, they decide "Is this a decent, presentable article?". Mr.Z-mantalk¢ 15:01, 25 August 2007 (UTC)
- Anything relating to "fact-checking" has been mostly removed. Now it is:
- "Is clear of libel and unsourced statements about living persons." - As all articles should be.
- "Is clear of unencyclopaedic content." - Should not be too difficult to determine.
- "Contains some references to reliable sources." - It should have at least some references to be flagged.
- Remember also that no one is forcing people to review articles, if they are not sure about something, they can choose not to review it or can ask someone else. If they start and think, "This is too much work", they can stop and do a simpler article instead to gain experience. Mr.Z-mantalk¢ 18:53, 25 August 2007 (UTC)
- I'm glad you added the "contains references" line. I almost added something similar, but held off since I hadn't seen much discussion about that. But that's not too onerous a requirement, and it would definitely be a bad idea to give any sort of mark of quality to unsourced articles.--ragesoss 19:10, 25 August 2007 (UTC)
- Anything relating to "fact-checking" has been mostly removed. Now it is:
- If that is how it is being interpreted, then the requirements should be rewritten. I don't think that the intention was to make this into GA review (which can often take weeks for a result because few people want to do it) but rather an assurance that the article is decent and presentable. The way it should work is: Someone sees an article that they think may be "sightable." It is fairly well developed, and they have some background knowledge of the subject. They check to make sure it has some sources and that some of the sources are reliable. They read it to check for obvious spelling, grammar, style, and factual problems (or templates indicating some). They check to make sure it is not in clear violation of policy such as [{WP:NOT]] and WP:NPOV (clear violations are generally easy to spot when reading). They check the history for recent vandalism. Finally, they decide "Is this a decent, presentable article?". Mr.Z-mantalk¢ 15:01, 25 August 2007 (UTC)
- I agree with Black Falcon. The way the requirements are formulated is stronger than what I gathered to be the intention. I thought that somebody familiar with the subject should be able to sight an article by reading through it, but "Introduction and major statements checked for accuracy" means a more careful review, taking at least ten or fifteen minutes even if you're familiar with the topic. I fear that would definitely lead to backlogs and edits not disappearing for months. I thought that it would be enough to check that the article is plausible. -- Jitse Niesen (talk) 02:56, 25 August 2007 (UTC)
- I was referring to the "Requirements for Sighted pages" section. Ensuring that an article is clear of vandalism, unsourced statements about BLPs, and unencyclopedic content requires thoroughly evaluating the article for conformance to key content policies on verifiability, original research, NPOV, BLP, trivia, WP:NOT, and so on. I'm not sure if that is a recent addition. — Black Falcon (Talk) 01:37, 25 August 2007 (UTC)
- Unless it was changed recently, I don't see where the proposal says we have to check that every statement is sourced and that all sources are reliable (with the exception of BLPs, but we technically should be doing that already). All it says is that "Introduction and major statements checked for accuracy/cited." - my interpretation of this is that they should be correct in the context of the article and verifiable. I would also hope that editors would not just choose random articles to go around and "sight", that they would pick articles they at least have some background knowledge on or were involved in writing. Mr.Z-mantalk¢ 01:27, 25 August 2007 (UTC)
- Your analogy is a good one, I think. The current "Requirements for Sighted pages" section effectively requires a detailed point-by-point review of the article. Much vandalism cannot be spotted through a quick one- or two-minute check (spotting and correcting original research and POV problems takes even longer). If the sole point of sighting is to spot and remove something like Wikipedia SUCKS ERFISP SD*SBC KLJnj-8unrfas8-d(, then I really don't think it's worth the extra work and bureaucracy. — Black Falcon (Talk) 00:18, 25 August 2007 (UTC)
(outdent) Let's look at the current requirements:
- Is clear of vandalism. As I've noted above, vandalism is not always obvious. Often it's in the form of changing a single number, date, or letter, or of adding false, but plausible, information. Correcting this type of vandalism requires checking/finding sources.
- Contains no spam in the external links. This requires checking every external link. This is not an issue with most articles but some, like Scientology, contain dozens of links and require more time.
- Is clear of libel and unsourced statements about living persons. Checking for libel is fairly easy. It's equally simple to delete any unsourced information about living persons. However, what about missourced information? A sentence may have a pair of
<ref>
tags after it, but it's impossible to know whether the sentence is verified without actually checking the link. - Is clear of unencyclopaedic content. This is generally not a quick and simple task. Yes, it's easy to spot long directory-like lists and blatant bias, but subtle original research and/or bias is not so easy to catch or correct.
- Contains some references to reliable sources. The number of references only matters if they support the content of the article. Checking that and checking whether a reference is reliable requires clicking each link and then considering it in the context of the statement it is supposedly supporting.
- Has been around for several days. Easy enough to check.
- Is not a one-sentence stub. OK.
- Has been spell-checked. Fairly straight-forward.
- Is readable (uncluttered) and is not tagged for cleanup. OK.
If sighting is not to require a serious investment of time (something like a GA review), then a "sighted versions" is little more than a version that looks good but is neither checked for accuracy nor carefully reviewed for conformance to content policies. Improving appearance without checking accuracy does not seem to match the goal of "quality over quantity". It inspires confidence in the content of the article without actually verifying it. In short, it misleads both readers and editors.
If en.wikipedia proceeds with the idea of flagged revisions, I think it is more productive to focus on accuracy and invest the time needed for a detailed review, rather than focus on appearance and create the impression of a decent article without actually confirming that an article is decent. — Black Falcon (Talk) 21:01, 25 August 2007 (UTC)
- What we should describe sighted versions as to readers is a very minimal check that the article has not just been maliciously disrupted. In order to achieve that, we have to have a little more extensive procedure; inevitably, it will fail sometimes but it should at least do a decent job of screening for vandalism and new spam. I don't think these requirements are especially tough, since once an article has been checked once, a simple diff check is all that is required to update the sighting. Long articles may be a chore to sight the first time, but this is inevitable with any worthwhile flagging system. Future extensions (which I'm eager to see) can deal with more substantial quality-checking procedures, but we have to start somewhere. If the basic sighting requirements are substantially less than the current proposal, then the procedure will catch hardly any current problems in articles (which is a big advantage of the current proposal: systematically clearing the most obvious crap out of articles, without too much duplication of effort).--ragesoss 21:17, 25 August 2007 (UTC)
- Agreed. We should have a formal fact-checking review process, with stable fact-checked versions available via some technical means. Anything more than that, while desirable, is overambitious at this point. And anything less than that will send a false message of quality -- the famous Siegenthaler article would probably have been "sighted", which would have made us look like worse idiots.--Father Goose 21:19, 25 August 2007 (UTC)
If someone is looking for it bias are easy to spot even by some with the same bias. Zginder 01:43, 26 August 2007 (UTC)
- Currently in 'recent changes patrol' people look at the text added/removed and if they don't see anything clearly against our policies they move on. I envision 'sighted' as requiring a similar level of review... if you read the unsighted changes and don't see any obvious vandalism/unsourced defamation/copyright violations/et cetera you can go ahead and mark your update, along with those changes, sighted. The key is that the problem(s) should be readily apparent to anyone who reads the text. Yes, this will undoubtedly miss many things... but it will catch most of the nonsense that generates the hundreds of complaints we receive every day. Which is why I think the 'two tier' system makes sense. One tier for 'somebody looked at this and did not see anything obviously wrong' and another for 'this article is as good as (or better than) what you would find in Britannica'. We need to both keep the 40 million casual users who view our pages daily from frequently seeing obvious vandalism and thoroughly review articles so that we can say with confidence that they are accurate and useful. Currently we can't do either of those things. Even 'featured articles' are not reviewed to verify that all of the references check out and cover everything claimed in the article. Even recent changes patrol can't catch and revert every vandal edit before someone sees it and complains to OTRS (or sees it and goes away angry without notifying us).
- I don't agree with the argument that we will 'look like worse idiots' for the junk missed by a quick review. People already blame us for whatever they see on a particular page. Even if it was there for all of three seconds before being reverted the irate e-mails inevitably say, 'How could you put that on your website!'. In the vast majority of cases you go to the page and find that some recent changes patroller already fixed it in the time it took the person to write the complaint e-mail. With sighted versions, as I describe above, almost all of that stuff would never even be seen by 99.9999% of our readers... it'd be unsighted and never displayed to them at all. That would dramatically decrease the number of complaints we get and the number of people angry with us over things that some dingus wrote and they saw before it was cleaned up. Many people already assume that we approve everything on the site. Those who are a bit more clueful know that it could be written by anyone, but generally still hold us accountable for not finding and removing things immediately. Removing the vast majority of the garbage from their sight during the brief time it takes to clean up will be a VERY good thing. Likewise, being able to say, 'We have really checked and approve this' will be a tremendous benefit, but should only be done under the strictest of standards. --CBD 14:42, 27 August 2007 (UTC)
- Your description of what sighting involves (a quick 15-second overview) does not match what is described in the "Requirements for Sighted pages" section of the proposal. To actually meet the requirements specified, one would need to spend at least a few minutes for every KB of text, checking the reliability of sources, checking for NPOV and NOR violations, checking external links for spam, checking for spelling errors, and so on. I support investing time to do the latter (a detailed review); but see no value in creating and running an entire bureaucracy just to say the equivalent of "this article on quantum physics does not contain 15 repetitions of the word 'penis'" (which is the only thing a quick review can prove). — Black Falcon (Talk) 18:29, 31 August 2007 (UTC)
I agree that we should abandon the concept of sighted versions. We'll get disputes, flaggng wars, additional bureaucracy, which Wikipeidia is not, lots of additional edits (to flag), a whole new type of vandalism to deal with, and other negative consequences we haven't even thought of. I, for one, am not going to flag any versions. I've seen too many cases where I read entire articles, kept them on my watchlist, and later discovered obvious vandalism that I'd missed several times in reviewing the article. As a reader, I wouldn't trust a sighted version any more than the current version, so as a reader I think this offers no benefit. Our recent change patrollers are pretty good about cleaning up bathroom stall quality vandalism, and readers will be at most irritated by that; subtle vandalism is very likely to get marked as sighted and that is what should be irritating our readers. The energy that will be spent in dealing with sighting is better spent on recent change patrol, solving real disputes, or generally improving articles. I see this as a proposal with significant and generally understated costs and no real benefit to our readers. GRBerry 17:40, 31 August 2007 (UTC)
- I would just like to say that I find the arguments in this section very convincing. However, one good argument for sighting that still remains (for me at least) is that it could serve as a deterrent, and as a result we would really have less work to do. The counterargument was that it could lead to a more sophisticated form of "intelligent" vandalism, but who knows...
- I still believe that it makes sense to experiment with sighting on semi-protected pages, and I would support the use of sighting strictly as a replacement to semi-protection. But I now fully agree that any the real push for quality would require some form of syncing with the more rigorous QA processes. --Merzul 18:10, 31 August 2007 (UTC)
Some more notes from the Foundation
I'm sorry, I'm not able to follow the discussion at the rapid rate at which it is happening right now. I've already commented a bit on the deployment strategy here. We've discussed this further now on the Foundation-level and have an agreement that the sequence will be as follows:
- security review of the extension
- beta test on a dummy site, with a few different configurations to play with
- approval from WMF for production use (probably around mid-October)
- interested wikis can request the extension to be activated through BugZilla with a specified configuration (they will have to point to a consensus or vote that approves the decision, details to be determined).
To understand a bit more where I'm coming from in all this, please read m:User:Eloquence/WikiQA, which I wrote almost a year ago in response to the original German proposal. So let's be very clear: I think there are many extremely valid concerns on this page. Nothing will be imposed on the English Wikipedia community (or any other) by the Foundation Board of Trustees!
The main thing I advocated throughout the development process is that the feature must be highly flexible so wiki communities will have a lot of freedom in how to configure it. This, for the most part, is the case for the current extension. So please let's not jump to conclusions about what will happen based on this wiki page, it's a specific proposal for the adoption of FlaggedRevs that does not reflect our current internal strategy for making this feature available to interested communities. Indeed, even the levels that we set and their meanings can be freely configured by any wiki community.
I encourage those interested in this feature to already work on separate propositions for how it should be deployed. To see what configuration options there are, take a look at this file, specifically, all the variables beginning with $wg. It's source code but it should be fairly readable with comments explaining what everything means.--Eloquence * 23:58, 24 August 2007 (UTC)
- this might be easier to read :) Voice-of-All 00:07, 25 August 2007 (UTC)
"Is readable (uncluttered) and is not tagged for cleanup."
How will this one work with an important article that is contentious. I have been working on an article that has been tagged for clean up since December 2006- and cleanup is the least of the editors worries. I could sight it clean of obvious vandalism and libel but it carries a cleanup tag (and several others) which may have been placed there as an act of mischief. When I consider it accurate and free of POVs and otherwise clean I will ask a third party to judge. If articles such as this are ommitted then Wiki is going to be accused of censorship. Would changing the wording to not recently tagged for cleanup. help? ClemRutter 19:10, 25 August 2007 (UTC)
- If an article is contentious and clean-up is a relative minor issue compared to others, then it probably should not be sighted. However, this will just mean that the current version always shows; such unsighted articles will behave just as they do now, without FlaggedRevs.--ragesoss 19:18, 25 August 2007 (UTC)
New pages created by a surveyor
I notice this has been amended to say that new pages created by surveyors should be sighted by another editor. I am inclined to agree with Rick Block above, that surveyors should be able to sight their on articles and would be interested in what the reasoning behind adding this was? Davewild 19:49, 25 August 2007 (UTC)
- I think it's not unreasonable, and would be helpful, if someone besides the creator has read through an article, to catch obvious errors. I know I often overlook typos and grammatical errors in my new articles. This will not make any difference in terms of visibility, as the article will appear to everyone whether or not it has been immediately sighted. As I see it, there's no reason not to have someone besides the creator do the sighting.--ragesoss 19:53, 25 August 2007 (UTC)
- There are problems as raised in an above discussion. One is that there would be often be a backlog of articles awaiting sighting especially with articles which are not often edited. This reduces the benefits of introducing sighted versions by leaving more articles in which vandalism will be visible. Also as raised above by B. Wolterding not sure how it would work with mergers. Davewild 20:07, 25 August 2007 (UTC)
- There would be no vandalism if no one has edited the article besides the creator. If others edit it, then the creator should be free to sight it, but random vandalism to new articles is very rare, except when they appear on the main page... by which time multiple editors have already read the article as it went through the Did You Know process. I suggest that this be a social rule rather than a software rule, so that exception can be made where appropriate.--ragesoss 20:17, 25 August 2007 (UTC)
- Ah I misunderstood what you added then as from the way I understood it was a surveyor should not sight an article they created ever, not as from what I understand now, that they should only sight it once someone else has edited it.
- The main reason I am concerned about this is that there are large numbers of articles which have only been edited by their creator - I have quite a number on my watchlist (some from over a year ago). Perhaps a flexible guideline whereby the creator should be advised not to sight an article they have created until either someone else has edited it or a period of time has gone by. Davewild 20:35, 25 August 2007 (UTC)
- If an article has only ever been edited by its creator, then what's the difference between it being sighted and not being sighted? I think it's very important that more than one set of eyes has been over sighted articles. I honestly don't think a new-article sighting backlog will be a big problem, even if it exists. The problem of vandalism scales with the number of readers and editors for an article, so low-profile articles seen only by their creators are in the least danger from others (and the most danger in terms of subtle, deliberate misinformation or hoaxes by article creators). It's going to take a while before most articles have sighted versions, and there isn't much urgency in avoiding backlogs except for articles that have earlier sighted versions and subsequent good edits.--ragesoss 20:48, 25 August 2007 (UTC)
- Based on the summary below this would require a software change to force. At first glance it wouldn't seem to make any difference. Either way, all users would see the most recent revision to the page... either because it was sighted or because there were no sighted versions of the page available. However, if we had some way of listing 'never flagged' pages then this might have some benefit in allowing users to go through and check every page. That said, a list of 'pages with only one editor' would accomplish much the same... and either of those could be easily foiled by someone who creates articles with one account and flags them with another. Thus, overall I don't think it would make too much difference, but it'd be reasonable as a social custom to encourage people to follow. I'd eventually love to see something like a 'list of pages with sighted revisions from fewer than X different users'... with the value of 'X' at 1 until people had gone through and updated/sighted nearly every page (there will always be a few newly created), then 'X' being changed to 2 until all of those had been updated/re-sighted, then 'X' at 3, et cetera. If it eventually topped out around 'X=5' and every newly created page were edited and sighted by five accounts before coming off the list it'd go a long way towards assuring quality. --CBD 13:43, 27 August 2007 (UTC)
- If an article has only ever been edited by its creator, then what's the difference between it being sighted and not being sighted? I think it's very important that more than one set of eyes has been over sighted articles. I honestly don't think a new-article sighting backlog will be a big problem, even if it exists. The problem of vandalism scales with the number of readers and editors for an article, so low-profile articles seen only by their creators are in the least danger from others (and the most danger in terms of subtle, deliberate misinformation or hoaxes by article creators). It's going to take a while before most articles have sighted versions, and there isn't much urgency in avoiding backlogs except for articles that have earlier sighted versions and subsequent good edits.--ragesoss 20:48, 25 August 2007 (UTC)
- There would be no vandalism if no one has edited the article besides the creator. If others edit it, then the creator should be free to sight it, but random vandalism to new articles is very rare, except when they appear on the main page... by which time multiple editors have already read the article as it went through the Did You Know process. I suggest that this be a social rule rather than a software rule, so that exception can be made where appropriate.--ragesoss 20:17, 25 August 2007 (UTC)
- There are problems as raised in an above discussion. One is that there would be often be a backlog of articles awaiting sighting especially with articles which are not often edited. This reduces the benefits of introducing sighted versions by leaving more articles in which vandalism will be visible. Also as raised above by B. Wolterding not sure how it would work with mergers. Davewild 20:07, 25 August 2007 (UTC)
How many surveyors?
Currently the proposal says:
- Reviewing pages requires the surveyor right. This right will be given liberally. Any trusted editor may be granted rights by an administrator, regardless of their edit count.
Would this form an "elite"? Some numbers on "classes of users" may help. On the English Wikipedia we have
- About 40 million visitors every day.[4] Lest it be forgot, wikipedia's primary purpose is to provide useful content for these people. For us, the "wiki way" should be a means to this end, not an end in itself. The main aim of this proposal is to improve wikipedia for such casual readers.
- 5 million user accounts, mostly inactive.
- About 18,300 editors qualifed to vote in the last Board elections, i.e. registered 3 months and with more than 400 edits (in any namespace).[5]. Also, a list was accidentally constructed based on a minimum of 201 edits, which had 50% more voters. [6]
- IP users who choose to edit seem to be outnumbered by registered users: The recent changes log shows that 10-20% of non-minor, non-bot edits are from anon IPs.
- 1308 admins, 24 bureaucrats, 15 arbcom members and 7 Board members. Plus a few ex-arbcom members, foundation employees, and developers with privileges like oversight and checkuser. Currently these are our "elite", if that is an appropriate term.
At one point this proposal said that surveyor rights would be granted automatically to users with 2 months activity and 300 main space edits, which is a comparable barrier to the board election hurdle. This would give about 20 times as many surveyors as admins. At the same time, there are many more users who have been registered for a long time but rarely edit. Such people can always ask an admin, who are instructed to give out rights liberally as above, but the act of asking is quite a barrier in itself. If a lot of them asked, and admins checked their contributions before agreeing, there would be a significant backlog. An alternative proposal made on this page was "30 days on which main-space edits were made", which would be much lower barrier...perhaps 100,000 editors would qualify, and yet this would be harder for would-be vandals to clock up deliberately.
The proposal now omits statements about automatic granting of surveyor rights. I think this is a big mistake as the numbers above imply that it would be a very large job to personally grant rights to anything like as many people. PaddyLeahy 13:01, 26 August 2007 (UTC)
- Update Since this post has been cited as showing that surveyors would indeed be an "elite", I'd like to emphasise that the vast majority of our 5 M users are inactive. Wikicharts listed 158,000 wikipedians with more than 10 edits in October 2006 (the latest update for en:wiki).[7] I don't know how many of the 5 M accounts have been registered since then, but surely a minority. October 2006 was about the point when en:wiki changed from exponential growth with a doubling time of about a year, to linear growth. This very likely reflects the rate of recruitment of new editors. Hence I would guess 3 or 4 M users in October 2006, most of whom were registered for at least a year. The fact that only 5% of them had managed even 10 edits shows that almost all registered accounts are used only very briefly. Therefore, 20-30,000 surveyors would be 15-20% of users with any significant contribution, and if we could get the number to 100,000 it would be a majority. PaddyLeahy 10:53, 30 August 2007 (UTC)
- See also Brianna Laugher's slides from her wikimania talk suggesting why so few new users contribute. PaddyLeahy 11:33, 30 August 2007 (UTC)
Translating the Programmer-Speak
OK, so I'm trying to make some kind of sense of the Configuration Options and so I'm going to post what I've learned from the configuration options themselves. And since, per a post above, the Foundation is going to let us decide how all this works, I think it's important that we all have a solid understanding of what our options are. Alright, here goes, we can choose:
- Which Namespaces to apply this tool on
- What type of "Flagged Revisions" we use
- What priority flagged types are given
- The flag with highest priority is the "Quality" revision
- Whether "quality/sighted" revisions will be viewed directly
- Detirmines if "quality/sighted" revisions are shown first
- Also whether only anonymous and logged out users will view the "quality/sighted" version first
- Whether editors/reviewers can leave notes at the bottom of pages
- Not sure how this would work, like edit summaries or like added content
- How long flagged revisions last in the cache (after which they will be refreshed)
- All this does is decide is how long the stable version is held in the sever cache. When the cache expires, it is refreshed. This is like purging a page, and will not change its content, unless the page had external templates/images that were changed (which don't refresh all pages that use them when changed). Voice-of-All 22:21, 27 August 2007 (UTC)
- Thanks for the clarrification, then is there even a possibility of "Auto-Sighting" in any sense? Adam McCormick 23:32, 27 August 2007 (UTC)
- All this does is decide is how long the stable version is held in the sever cache. When the cache expires, it is refreshed. This is like purging a page, and will not change its content, unless the page had external templates/images that were changed (which don't refresh all pages that use them when changed). Voice-of-All 22:21, 27 August 2007 (UTC)
- Autopromoting of users to "Editor" status according to a minimum number of days, a minimum number of edits, and a confirmed email address
- Autopromoting, as built in, seemingly cannot be done by a combination of edits and days such as "30 days with at least one edit each"
- Whether an Editor who watches pages he/she edits will also watch pages he/she reviews
- Whether Editors will be asked to review unflagged edits after they make an edit
- Whether edits made by Editors to flagged versions are also flagged
- Templates and images necessarily enable this option
- Whether to provide "stable" and "current" tabs
- Whether to use some kind of simpler UI
- Whether reviewers can make new editors/reviewers
- Whether editors can make new editors/reviewers
Now the easy part:
- Bureaucrats can make reviewers and editors
- Sysops/Admins can make editors
I think that covers it. This is everything that is listed as being possible. To make the extension do anything else would require the extension itself to be changed. Adam McCormick 04:46, 27 August 2007 (UTC)
- Thanks for this summary/translation. Does anyone know if something like 'X different days with article edits' could be added to the system? We don't have that for any Wikipedia feature that I know of currently, but it doesn't seem like the logic should be too complicated. I can't see Admins/Bureaucrats keeping up with requests for the access, and certainly not with users who should have it but don't ask. Allowing everyone with the access to give it out is an open door policy for vandals. Autopromoting after X days and/or Y edits may be viable, but you run into problems where 30 days/30 edits is an easy barrier for vandals to get armies of sockpuppets past (create a couple each day, wait until they reach 30 days, make 30 useless talk/userspace edits, go vandalize) while 60 days/300 edits keeps 90% of our useful contributors from ever having the access. A '30 different days with article edits' standard would require serious commitment by a vandal to get one account authorized (let alone dozens) but not much more onerous than 30 days & 30 edits for helpful contributors. --CBD 12:35, 27 August 2007 (UTC)
- It would be nice to also require edit summaries for those (to make it harder to just BS it). I'm playing around with how well
SELECT COUNT(DISTINCT SUBSTRING(rev_timestamp,1,8)) FROM `mw_revision` WHERE rev_user = 1 AND rev_comment != ”;
- scales. .0032 seconds for 4000 rows. Still, user_editcount is faster (already loaded anyway, and only .0012 sec, independent of the # of edits). Requiring 100 edits in addition to 30 edits in 30 days could make it faster, since user_editcount could be checked first to avoid wasting the time to scan further. Voice-of-All 23:02, 28 August 2007 (UTC)
- I wouldn't have any problems with that as long as it isn't limited to mainspace edits. Adam McCormick
- That would require a JOIN, so it's not happening anyway ;). Either way, it's still unlikely that querying the revision table would be efficient enough on a large site. So I'll be looking for other heuristics. Voice-of-All 00:55, 29 August 2007 (UTC)
- Hmmm. So more technical hurdles than I'd thought.
- On the speed issue, perhaps something like 'X edits before <today - Y days>' could be done more quickly? Or the equivalent, 'total edits - <edits from last Y days> greater than X'. The primary intent of the 'edits on X different days' concept was to build in a delay for review between the time the edits were made and the access granted. Thus, '30 edits which are more than 3 days old' would accomplish much the same thing... a reasonable sample of contributions and plenty of time for people at recent changes patrol to have checked these and thrown up warning flags if there are problems.
- As to the other bit, I think promotion very much should be limited to counting only mainspace edits... because a person decorating their user page or playing around in the sandbox tells us virtually nothing about whether they are going to be a vandal, POV pusher, or whatever when they get access to flag revisions in the article namespace. Nor does it do anything to improve the encyclopedia and thus 'offset' the effort which would be involved in cleaning up any vandalism they might engage in... it's a 'net loss' to the encyclopedia. Chatting on talk pages may or may not give us some indication of the person but wouldn't directly improve the encyclopedia. Edits to the Wikipedia, Template, MediaWiki, Portal, Help and suchlike namespaces might be beneficial/give us some idea about the person, but the article namespace is the central mission/focus of Wikipedia, by far the most heavily read, where most vandalism and POV problems occur, and thus the best place to evaluate a user's contributions. One of the main purposes of flagged revisions is to remove vandalism and nonsense from 'public view' during the time it takes to locate and correct... the vast majority of that 'public view' concern (virtually all of it) being our articles. Basically, it just doesn't make sense for someone to gain the ability to flag revisions in the article namespace without ever having made a single edit there. If they need to provide more in the way of positive contributions to articles to get the revision flagging than the damage they could do in the time between getting it and being blocked, then even the 'clever' vandals provide a 'net gain' for us. If they can just futz around with their user page and then make a mess we lose ground. --CBD 11:41, 29 August 2007 (UTC)
- That would require a JOIN, so it's not happening anyway ;). Either way, it's still unlikely that querying the revision table would be efficient enough on a large site. So I'll be looking for other heuristics. Voice-of-All 00:55, 29 August 2007 (UTC)
- I wouldn't have any problems with that as long as it isn't limited to mainspace edits. Adam McCormick
- As a hurdle before the main check, 100 edits anywhere seems fine. Few people with normal edit patterns will manage to make 30 mainspace edits on different days with less than 100 edits overall (usually edits come in bursts). And only a tiny fraction of our 5 million registered users have a total of 100 edits (95% of them have less than 10),[8] so this will be an effective screen. PaddyLeahy 10:40, 30 August 2007 (UTC)
Concern and modified proposal
Thinking further I have put 3 major concerns together and would like to check views on the rather surprising conclusions:
Concerns. |
---|
|
Impact of these. |
---|
|
I think on balance, #2 is actually a rather neat solution. If everyone who creates an account and uses it for a number of days and number of edits, has all their edits "sighted" by default (unless the right is removed or suspended), then you have the following situation:
- Most articles will be automatically and routinely sighted on the latest vandalism-free version.
- Most non-logged-in users will therefore see the latest version (and also see all templates) unless there is a good reason such as "latest version was by a newcomer or an editor stripped of sighting rights".
- Edit war potential is not enhanced cut since there is no fight to get "a preferred version" sighted. In an edit war, the latest versions by both sides are auto-sighted, but that would probably happen anyway; there is no separate sighting war or need to deal with reported sighting wars in most cases.
- Last, a nice side effect: Surveyor rights could be made contingent upon conduct - disruptive conduct = automatic loss of surveyor rights for 7 days, similar to 3RR. That'd be a huge incentive to behave, because at that point everyone else in the dispute has their versions auto-sighted, but someone who edits disruptively loses that right for a few days. (Note: they could still edit, and their edits could still be sighted by another editor, so it doesnt have the effect of a block.) What's the point in edit warring if it means you automatically at admin judgement lose your edits being sighted for a while. This would probably have a huge effect on edit wars and reduce the amount of POV war versions presented to non-editorial readers too.
Proposal
So I'd support a flagging system with these criteria:
- Any editor with an account and "30 days and also some number of edits" automatically gets surveyor rights, and also, their own edits are all automatically sighted.
- An editor who has edited vandalistically or disruptively may have their surveyor right suspended for a period of a week, by any administrator. (Or given communal consensus/repetition, for longer).
Sweet and simple. Suddenly most vandalistic edits are hidden from public view, most sighted versions are current (unless vandalized), there's no wars over sighting since everyone's edits are all auto-sighted (with very few exceptions), and there's a huge new disincentive to pov war in the traditional sense since it leads instantly to a state where one can still edit but one's edits won't be auto-sighted for the next few days and will specifically not be shown first to readers.
So as well as removing vandalism from public viewing, sighting would then also tend to remove disruptive editor's posts from public as well. Act up, and you can still edit, but you become the one person whose edits are not auto-sighted for a while.....
Thoughts? FT2 (Talk | email) 11:12, 27 August 2007 (UTC)
- I could be wrong, but I think that like so many have the wrong idea. First, and I think you must be missing this, ANYONE logged in or not will be able to see the most recent version, it just sometimes might require an extra click. And if something is so 'important' that the user simply must see the most recent findings/happenings, then it'll almost certainly be watched and sighted with quickness anyway. ♫ Melodia Chaconne ♫ 11:44, 27 August 2007 (UTC)
- You're mistaken. Its not a case of the wrong idea. It's long term experience of how POV wars and their editors work in practice, and having dealt with such situations as a long term Wikipedia activity. "Be able to see" does not mean "default to"; 90% of editors will not, and the ability to push a specific version to "the world at large" is going to be seen as valuable in edit wars, and will enhance them. That seems very evident. The flip side of that is, if disruptive editing or vandalism makes a non-new editor the one person whose edits are (for a while) not sighted automatically, then this disincentivizes edit warring. —Preceding unsigned comment added by FT2 (talk • contribs) 12:12, August 27, 2007 (UTC)
- I completely agree on this. Also, the idea that sighted status should be given to accounts which have made edits on more than 30 different days floated above sounds like a good idea to prevent the creation of sleeper sighted accounts. ELIMINATORJR 12:45, 27 August 2007 (UTC)
- Unfortunately, the software can't do the "edits on 30 different days" thing (yet).--ragesoss 16:39, 27 August 2007 (UTC)
- I don't see how this could "enhance" edit wars... as it is exactly the same as the current situation. Users with access to 'flag' revisions will be able to "push a specific version to 'the world at large'"... just like all users can right now. Thus, I agree that edit wars will continue to exist, but I don't see how they will be any more intense or significant than they are currently. We will still have page protection (not to mention blocks) to deal with edit wars. We just won't need to use it for vandalism as much - hopefully not at all. Otherwise I agree with what you say and do think we could do alot of good by replacing many blocks and warnings for 'edit warring' and 'POV pushing' with some sort of equivalent temporary loss of access to flag revisions... it would encourage users to be more neutral and cooperative so that their edits will be more likely included in a flagged update - and so that they don't get hit with longer waiting periods on being able to flag things themselves. --CBD 12:59, 27 August 2007 (UTC)
- I also don't see how edit wars would be any worse with this. It would still be a violation of Wikipedia's standards of conduct (and grounds for having surveyor rights removed), and in the vast majority if cases, editing disputes are resolved by compromise and mutual agreement (at which point, the new version can be flagged). For many edit wars, it's regular editors who are arguing so every new edit will probably be sighted anyway. Generally neither warring party is happy with the current or past state of the article, so I expect more energy will continue to be put in to making changes than arguing over which past version is best.--ragesoss 16:37, 27 August 2007 (UTC)
- If everyone who is not a vandal pretty much, can sight versions, then it won't, because each version in a dispute will be sighted by its creator. But if some are, some aren't, then the potential for sighting rights to be used in edit wars increases. A better use is to make it a universal right, that's removed or suspended for bad play. That way edit warring is strongly disincentivized; a warrior can stil edit, but their version won't be so easy to get sighted if they do. Thoughts? FT2 (Talk | email) 19:33, 27 August 2007 (UTC)
- I also don't see how edit wars would be any worse with this. It would still be a violation of Wikipedia's standards of conduct (and grounds for having surveyor rights removed), and in the vast majority if cases, editing disputes are resolved by compromise and mutual agreement (at which point, the new version can be flagged). For many edit wars, it's regular editors who are arguing so every new edit will probably be sighted anyway. Generally neither warring party is happy with the current or past state of the article, so I expect more energy will continue to be put in to making changes than arguing over which past version is best.--ragesoss 16:37, 27 August 2007 (UTC)
- I completely agree on this. Also, the idea that sighted status should be given to accounts which have made edits on more than 30 different days floated above sounds like a good idea to prevent the creation of sleeper sighted accounts. ELIMINATORJR 12:45, 27 August 2007 (UTC)
- You're mistaken. Its not a case of the wrong idea. It's long term experience of how POV wars and their editors work in practice, and having dealt with such situations as a long term Wikipedia activity. "Be able to see" does not mean "default to"; 90% of editors will not, and the ability to push a specific version to "the world at large" is going to be seen as valuable in edit wars, and will enhance them. That seems very evident. The flip side of that is, if disruptive editing or vandalism makes a non-new editor the one person whose edits are (for a while) not sighted automatically, then this disincentivizes edit warring. —Preceding unsigned comment added by FT2 (talk • contribs) 12:12, August 27, 2007 (UTC)
- Here's a further suggestion: auto-sighting is enabled for a surveyor if the previous version is also sighted (i.e., going from a "trustworthy" version to another "trustworthy" version). Auto-sighting is disabled if the previous version is unsighted, since the previous version may not be "trustworthy"; the surveyor can still manually sight the version, of course, via a checkbox. Just brainstorming here; let me know if there would be unintended consequences. Jpers36 17:30, 27 August 2007 (UTC)
- That would be better, because if someone finds a article needing a lot of work it may take many edits and a few days to get it to sighted quality. Zginder 20:56, 27 August 2007 (UTC)
- That is currently how it works atm. Voice-of-All 21:02, 27 August 2007 (UTC)
- That would be better, because if someone finds a article needing a lot of work it may take many edits and a few days to get it to sighted quality. Zginder 20:56, 27 August 2007 (UTC)
- I disagree with "We may be unable to avoid edit wars; but this will make "getting your preferred version" as the visible version immensely easier and immensely more desirable." I would assume that page protection will be used in a sight war just as it is used in an edit war. Also, as stripping the rights is less destructive than blocking (in the sense that they can still edit), right-stripping should be applied much more liberally in sight-wars than blocking is in edit wars. 3RR will still apply to sighting (at least it should). I also believe that we should be giving the flagged version to non-logged in users. People have said that this will inconvenience anons who edit, do RCP, etc. and it will. However, we should be doing this for the readers, not the editors. Mr.Z-mantalk¢ 21:30, 27 August 2007 (UTC)
- Also note that, if you can't edit a page (via block/protection or such), then you can't review it either. Voice-of-All 22:26, 27 August 2007 (UTC)
I don't agree with the "'edits on 30 different days' thing". For a long time, the message boards at GameFAQs had a system similar to this, where making at least one post a day netted you a point of "Karma". Having enough Karma eventually removes posting restrictions and raises your status as a board user. However, during this period of time, some message boards became filled with "Karma topics" consisting of messages containing just the words "Karma post", which of course, contribute nothing to discussion and are essentially just a way to WP:GAME the system. I'm worried that a similar thing might happen here, (ie, people making edits for the sake of having an edit, not for the sake of contribution) because even a sufficently sneaky vandal would be able to gain surveyor status. shoy 16:08, 28 August 2007 (UTC)
- Obviously users who made article edits which were the equivalent of 'Karma post' from your example above would have those reverted and be warned and eventually blocked or, preferably to me, have the number of editing days before they could flag revisions increased. Having a system based on 'X days and/or Y edits' could be 'gamed' in the same way, except in that case all of the nonsense edits could be made in a matter of minutes and the user off vandalizing freely. By making it 'article edits on X different days' we build in a period for review of those edits and the ability to take action if they aren't beneficial. Even if a user is just correcting punctuation or fixing wiki-links one time a day for 30 days that is a benefit to the encyclopedia which can be retained long after any vandalism they then commit is wiped away. If the edits they are making during those '30 different days' aren't helpful in any way then we have time to see this and act on it before they ever get the permission. The situation at the GameFAQs message boards you describe would presumably have been different if they had a policy of warning and then banning people who wrote just "Karma post" or the like. --CBD 11:07, 29 August 2007 (UTC)
- I believe part of the issue (and correct me if I'm wrong) is that there needs to be a significantly large number of sighters, so that they will automatically be granted the ability after the edit. Not to sprout some WP:BEANS, but I can see it being really easy to goto random pages in varying fields and making small edits like deleting spaces, or changing something and then fixing it back (I've seen IPs do this a number of times). ♫ Melodia Chaconne ♫ 11:35, 29 August 2007 (UTC)
- That's exactly the sort of thing I'm talking about. With automatic granting of surveyor status, whether it be by number of days or posts, or days and posts, determined-enough individuals can and will slip through. I can see this becoming a problem in contentious topic areas with users who want to push a particular POV. shoy 18:21, 29 August 2007 (UTC)
- So what? It wouldn't stop all vandalism? That's true. People could game the system, and some undoubtedly will. People game the system today; they click 'edit this page' and add their POV or vandalism. Poof, the change appears to all readers of Wikipedia. If there's a 30-edit or -day or -editing-day barrier, the situation will be much improved. -R. S. Shaw 19:18, 29 August 2007 (UTC)
- Yep, people will game the system, no matter what we do. I don't think we need to be worrying about how determined people can game the system, other than just knowing what to look for. If it helps reduce the number of casual vandals, without placing an undue burden on good editors, then it seems reasonable to me. Sxeptomaniac 21:09, 29 August 2007 (UTC)
- So what? It wouldn't stop all vandalism? That's true. People could game the system, and some undoubtedly will. People game the system today; they click 'edit this page' and add their POV or vandalism. Poof, the change appears to all readers of Wikipedia. If there's a 30-edit or -day or -editing-day barrier, the situation will be much improved. -R. S. Shaw 19:18, 29 August 2007 (UTC)
- That's exactly the sort of thing I'm talking about. With automatic granting of surveyor status, whether it be by number of days or posts, or days and posts, determined-enough individuals can and will slip through. I can see this becoming a problem in contentious topic areas with users who want to push a particular POV. shoy 18:21, 29 August 2007 (UTC)
- I believe part of the issue (and correct me if I'm wrong) is that there needs to be a significantly large number of sighters, so that they will automatically be granted the ability after the edit. Not to sprout some WP:BEANS, but I can see it being really easy to goto random pages in varying fields and making small edits like deleting spaces, or changing something and then fixing it back (I've seen IPs do this a number of times). ♫ Melodia Chaconne ♫ 11:35, 29 August 2007 (UTC)
- Obviously users who made article edits which were the equivalent of 'Karma post' from your example above would have those reverted and be warned and eventually blocked or, preferably to me, have the number of editing days before they could flag revisions increased. Having a system based on 'X days and/or Y edits' could be 'gamed' in the same way, except in that case all of the nonsense edits could be made in a matter of minutes and the user off vandalizing freely. By making it 'article edits on X different days' we build in a period for review of those edits and the ability to take action if they aren't beneficial. Even if a user is just correcting punctuation or fixing wiki-links one time a day for 30 days that is a benefit to the encyclopedia which can be retained long after any vandalism they then commit is wiped away. If the edits they are making during those '30 different days' aren't helpful in any way then we have time to see this and act on it before they ever get the permission. The situation at the GameFAQs message boards you describe would presumably have been different if they had a policy of warning and then banning people who wrote just "Karma post" or the like. --CBD 11:07, 29 August 2007 (UTC)
- To prevent backlog, the right to mark a version as sighted, (i.e., free of vandalism and BS) should devolve automatically to trusted editors. And a trusted editor could be one who has editted (productively) on at least thirty different days, or to thirty different articles. Newbyguesses - Talk 02:32, 30 August 2007 (UTC)
Auto-sighting
Is it possible to have all edits to sighted pages that are not reviewed in a certain amount of time to be auto-sighted, as this would avoid long backlogs, we could say all edits will be reviewed or visible within _____. Alternatively, these pages could be auto unsighted. Zginder 20:37, 27 August 2007 (UTC)
- As per my post above, we can set a timeout after which sightings will expire, thus making it so that all revisions X-days after a sight will just be visible. I didn't see anything about making edits become sighted after a specific period though. Adam McCormick 21:49, 27 August 2007 (UTC)
- That would be better, since it avoids possibly vandalised revisions getting automatically marked as vandalism. However, if the privilege is given out freely enough, there shouldn't be a long backlog. A special page listing pages that have at least one sighted revision, and a more recent unsighted revision that's been around for more than a week (or whatever) would be good, though (assuming such a thing doesn't already exist). --Tango 22:37, 27 August 2007 (UTC)
- A week is way to long. A day at the most. Zginder 23:04, 27 August 2007 (UTC)
- There are a lot of specialized articles, for which there are only a couple of Wikipedians able to determine whether an edit is vandalism or not. For instance, this edit is nonsense, but not that easy to spot. For such articles, a day would be too short in my opinion. But the idea that pages with pending edits become automatically unsighted after some time is interesting. -- Jitse Niesen (talk) 02:19, 30 August 2007 (UTC)
- There enough time in one day for one of those knowledgeable Wikipedians to log on and check one's watchlist Zginder 12:26, 30 August 2007 (UTC)
- Nope. There are lots of pages that are on nobody's watchlist. Special:Unwatchedpages (admin access only) currently lists 1000 pages with names beginning with a punctuation character or two character combinations from "11" to "19" - and then doesn't list the remainder of the unwatched page list. It is not a safe assumption that any particular page is on any Wikipedian's watchlist, much less on the watchlist of an active Wikipedian. GRBerry 14:40, 31 August 2007 (UTC)
- There enough time in one day for one of those knowledgeable Wikipedians to log on and check one's watchlist Zginder 12:26, 30 August 2007 (UTC)
- There are a lot of specialized articles, for which there are only a couple of Wikipedians able to determine whether an edit is vandalism or not. For instance, this edit is nonsense, but not that easy to spot. For such articles, a day would be too short in my opinion. But the idea that pages with pending edits become automatically unsighted after some time is interesting. -- Jitse Niesen (talk) 02:19, 30 August 2007 (UTC)
- A week is way to long. A day at the most. Zginder 23:04, 27 August 2007 (UTC)
- That would be better, since it avoids possibly vandalised revisions getting automatically marked as vandalism. However, if the privilege is given out freely enough, there shouldn't be a long backlog. A special page listing pages that have at least one sighted revision, and a more recent unsighted revision that's been around for more than a week (or whatever) would be good, though (assuming such a thing doesn't already exist). --Tango 22:37, 27 August 2007 (UTC)
Comment from project page
I am the one of the two [9] people Jimbo is accusing of FUD; my post is here [10]; Chris' is here [11]. Jimbo is claiming that this feature will only be applied to semi-protected/protected articles; this is nowhere explicitly stated in the proposal, which proposes this as a model for the entire wiki. Sdedeo (tips) 21:27, 29 August 2007 (UTC)
(Note from Melodia: Sorry if I didn't title it right or anything. Feel free to fix it up. ♫ Melodia Chaconne ♫ 22:05, 29 August 2007 (UTC))
- Jimbo claimed nothing of the sort. He pointed out that in the case of protected articles, flagged revisions would allow users to edit who were previously unable to edit. The corollary to this is that on other pages this policy has no effect on editing rights, not that in other cases flagged revisions wouldn't be put to use. Jpers36 22:26, 29 August 2007 (UTC)
I put the most reasonable interpretation on Jimbo's words ("don't worry, this will only apply to semi-pro/pro.") You're welcome to your Walesology. Sdedeo (tips) 22:41, 29 August 2007 (UTC)
- There has certainly been an inordinate amount of 'FUD' on this issue. The reality is that as envisioned it should cause Wikipedia to appear and work just as it does today except;
- Nearly all pages currently protected or semi-protected against vandalism will be unblocked so that they are editable by all users.
- The unlogged users who do 99.99% of Wikipedia page reading will rarely see the kind of obvious vandalism that currently generates hundreds of complaints every day.
- There will be a brief delay (generally a few minutes) before changes made by relatively unknown/new users are shown to unlogged readers.
- That's what we're really talking about. Details on how to implement flagged revisions to meet those goals still need to be worked out, tested, and refined, but that's where we are going. There are additional (even more impressive IMO) benefits to the second 'quality' level of flagging, but it is this base level 'sighted' system which seems to be generating most of the unwarranted hyperventilating. --CBD 01:14, 30 August 2007 (UTC)
- Please don't be unnecessarily rude or dismissive of those who believe (or *know*) that this proposal is unnecessary at best and damaging at worst. It has already been stated that users will have to *apply* for "surveyor" (i.e. "Constable," as on Citizendium) privileges. Those editors who refuse to apply for these privileges will be shut out of our own encyclopedia. That is wrong! Badagnani 01:18, 30 August 2007 (UTC)
- You're certainly not being shut out of the project, you will still be able to edit, indeed, able to edit substantially more articles than at present, all that might happen is that your edits won't go live instantly. That means no more embarrassing typos or syntax errors going live the moment you hit save. You'll all probably end up being able to sight your edits anyway, it'll just give you an extra layer of protection against the silly little mistakes we all make. Nick 01:28, 30 August 2007 (UTC)
- (e/c) The rights can be applied automatically, I don't think we have decided that yet (We haven't really decided anything yet). I for one think they should be given automatically. I sure don't want to waste much time looking through people's contribs to make sure they aren't a vandal while CAT:CSD grows to even bigger backlogs. And we won't be shutting people out either. As has been said before multiple times, this will be manually applied to each page it is used on, it will not be used on every page, the majority of pages may not use this. And please remember, we are mainly doing this for the readers so they will see less vandalism. And even if editors don't yet have the rights, its not like the pages will be protected to those without the rights; they will still edit the same version as everyone else and if their edits are helpful they should be sighted quickly. Pleas stop saying this is intended to emulate Citizendium. Jimbo has been talking about stable versions for months, if not years, as a way to improve the quality of what we present to the readers. Note that unlike CZ, we are using a more "utilitarian" (for lack of a better word) name, unlike "Constable" which implies authority. Mr.Z-man 01:37, 30 August 2007 (UTC)
- Badagnani, most of what you "know" about this proposal seems to be false. You state as 'fact' that, "users will have to *apply* for 'surveyor'" - while in reality very few people have ever advocated that and it has no chance of actually happening. It would directly prevent everything Jimbo and the Foundation hope to get out of flagged revisions. You go on about the 'secrecy' of this change... apparently unaware that it has been publicly and prominently discussed for months.
- You are concerned? Fine. Ask questions. Find out a bit more about this long planned feature that you are apparently just now hearing (incorrect) information about. Stop implying that those of us who actually know the history and intent are all lying about how it will work. Get a little knowledge... and then discuss the realities of the situation rather than these fictions which are being bandied about by scare-mongers, liars, and those uninformed who have gotten the wrong idea from them. Being concerned is fine. Change can do that. But don't go barreling into something you know almost nothing about and insist that it must be evil because all the things you don't really know prove it to be so. --CBD 11:09, 30 August 2007 (UTC)
- Is this what is called "a troll"? Because it sure looks like one. I've been reading and contributing to this discussion from the very beginning; perhaps it is you who needs to follow the discussion a bit more closely, because your statements regarding the proposed application for these most unnecessary "privileges" are quite opposite what has been stated as fact by those "in the know" for several months here. I will not hesitate to speak my mind on this issue, unnecessarily rude edit summaries notwithstanding. Badagnani 11:20, 30 August 2007 (UTC)
- Regarding misinformation, I know it when I see it, and oh, do I see it. In regard to having to apply for "surveyor" status, it appears quite clearly, in black and white, in the archives of this page (just do a CTRL-F search for "grant"). Regarding whether this feature is projected for just a few prominent articles (primarily FAs) or for every WP article, at first it was stated, emphatically and with certitude, that it would be applied to just a few prominent articles. Then it was stated that it would eventually be applied to all articles. Then it was stated that it would only be applied to a few articles. Then it was stated that it would be applied to all articles. Now we're back to that it will be applied to just a few articles. This is obfuscation, which the feeble protest above simply cannot wish away. That's the beauty of the Wiki system; all the edits and discussion history are still there for all to view. Human memories are short--and understandably so due to the high volume of discussion here--but please do not impute ignorance of the issues at hand to other editors simply because they find this proposal entirely unnecessary and without merit. Badagnani 11:37, 30 August 2007 (UTC)
- Welcome to Wikipedia. People say all kinds of things. That doesn't make them all true. You describe each of these different ideas by different people as if they were some campaign of misinformation to hide the true nefarious plans. It was just people talking about different ideas of how to do things. Some people think it should be applied to all pages from the start. Some think limiting it to 'just FA' or 'just currently protected/semi-protected pages' or some other subset would allow the fearful to see that this is in no way detrimental before spreading it out to all pages. All this stuff about how 'first we were told A and then we were told B and now A again... so it is obvious you are all lying' is exactly what I was talking about. It's a recasting of reality through a distorted lens. You start from the premise that you are being deceived and disenfranchised and wind up turning the different ideas of various people into a carefully coordinated plot. That this would be applied to all pages is hardly new or hidden information... it was stated in the Signpost (and outside news articles) from nearly a year ago that I linked above. You obviously aren't as 'in the know' as you seem to think you are. Everything which you are ranting about being bad... I agree is, in fact, bad. It just isn't what is actually happening. Some people have suggested pieces of it, ironically as part of efforts to calm the fears of people who have the wrong idea, but there is a very clear and long-standing plan to roll this feature out such that it should have minimal impact beyond allowing us to unprotect pages and delay public display of the minority of edits made by relatively unknown users for the brief time (on average a couple of minutes currently) it takes to weed vandalism out of them. That's it. Should be all that changes. If the implementation doesn't achieve that it will be tweaked until we get there. --CBD 12:26, 30 August 2007 (UTC)
- I'm afraid the "it's not going to be applied to every page" ruse isn't going to work anymore; maybe you missed that one of your colleagues already let it slip a week or two ago that once it's applied to the FAs and similar articles it will eventually be extended to all articles. This was stated with certainty at this page by someone "in the know." This is the same editor who keeps removing notice of this discussion from prominent locations, apparently to prevent the wider WP community from knowing about it. Badagnani 01:41, 30 August 2007 (UTC)
- Hi Badagnani. We're not trying to hide anything here, the feature will be employed by all pages, either from the start, or through a progressive roll out. Pages that go un-reviewed will automagically be flagged and displayed after a length of time to be determined. There's suggestions that this be 15 mins, 30 mins, an hour, or more problematically, a day or a week. All articles will initially be flagged when the system is rolled out, I believe, and from then on, any changes will require to be flagged. If you, as an anonymous editor, edit a largely ignored article, then if someone doesn't flag it as being suitable within a couple of minutes, as is likely to happen, it'll, by default, go live after a predetermined length of time. We are expecting editors to flag articles as being OK when doing RC Patrol as a matter of routine, and we all know how quick the RC Patrollers are.
- It might be, however, that we have to disable automatic flagging for articles which are high traffic, suffer lots of vandalism etc, so every edit is checked and flagged or not, this would be the new equivalent of semi protection for articles like George W Bush
- As an aside, If you have any knowledge of editors removing note of this discussion, please do let me know, I am concerned that some editors do feel this is all being done rather secretively. I'll be happy to investigate. Nick 01:59, 30 August 2007 (UTC)
- Well, I don't recall the idea to start of with everything flagged being popular, or good. It would result in many pages being flagged that are not watched, which goes against much of the usefulness of flagging (updates would never show, vandalism could be frozen). I'd rather pages start of as they are now, and people review pages that they (or members of a wikiproject, or if it's high profile, other editors) can be expected to check in on and keep up dated. Also, initially, we would not have yet built up a large surveyor base for all those pages, since all pages would be flagged without anyone getting rights yet. Voice-of-All 02:23, 30 August 2007 (UTC)
- What happened was, there was a note on the watchlist for about four hours, before someone took it off (that's how I found out, for instance). ♫ Melodia Chaconne ♫ 02:44, 30 August 2007 (UTC)
- Nick could start his investigation by reading the archives of this page, where AmiDaniel gave his reasons for removing the notice (on two occasions).
- Nick seems to imply that a page must be flagged to be displayed, In fact it will be displayed as now unless there is a flagged version. Per Voice-of-All, auto-sighting is controversial, and the proposal has never been to sight all articles initially. I have added this issue to the discussion topics below. Nick assumes that RC patrol will continue as now, which is worth debating. RC patrol is motivated by the need to minimise vandalism visible to general readers. Arguably this effort would be better spent at least in part on article improvement, once sighting is widespread. Consequently, sighting within 30 minutes is also unrealistic: even now vandalism on obscure pages often survives much longer than that. PaddyLeahy 08:13, 30 August 2007 (UTC)
Outstanding issues
In an attempt to focus discussion (and facilitate another archiving of this talk page), I'm going to try to lay out the main points of contention regarding the proposal. Please add significant threads that I missed, and correct my summaries if they give specific arguments short shrift.--ragesoss 02:59, 30 August 2007 (UTC)
Autoconfirm surveyor rights?
Earlier version of the proposal included the provision that surveyor rights would be granted automatically after a certain edit threshold (e.g., 300) and time threshold (e.g., 2 months), along with a confirmed email address. An alternative or supplemental way of populating the surveyor pool is to give every surveyor the ability to grant the right (as opposed to just admins and/or autoconfirmation). However, manually granting surveyor rights increases the amount of social contact that happens between editors (which is often regretfully rare with new editors) and reduces the threat of "sleeper" accounts. Many discussants have liked the idea of autoconfirmation of accounts with at least 1 edit on each of 30 different days, but the software does not currently permit such criteria.
Discussion
Again, I do not think that either admins handing out access to those who request it or users with the rights being able to give them to others is at all viable. The first leaves out users who do not request the access and will almost certainly generate an enormous backlog of requests... discouraging users from contributing. The second would eliminate the backlogs, but allow vandals to 'play nice' with one account to gain the ability to 'approve' an unlimited number of other accounts to use for vandalism... invalidating the whole purpose of flagged revisions. Automatic granting of the rights is the best course IMO. Consider; semi-protection has proven successful in reducing vandalism. That system requires only that the user account be at least 5 days old... yet it has proven effective in decreasing 'drive by' vandalism. Applying the same '5 day' criteria to flagged revisions would be exactly as effective at reducing vandalism seen by the general reading public (on ALL pages) while allowing all users to continue editing all pages - just with a brief delay before edits from very new accounts are seen by some users. I'd actually prefer somewhat stricter controls to weed out some of the vandalism which now gets past semi-protection, and some people have advocated MUCH stricter controls (e.g. 300 edits and/or 2 months), but I think almost everyone agrees that in order to work, flagged revisions have to be available to nearly all active contributors to the encyclopedia. --CBD 11:43, 30 August 2007 (UTC)
- Yes, IMO, automatic granting is a must. I sure don't want to waste my time looking through the contribs of 100s or 1000s of users to make sure they aren't a vandal, troll, or other disruption (because that's pretty much what the criteria boil down to). I think 150-300 edits (less than 150 and they may not know enough of the rules, more than 300 would take too long to get for many users) and 10-30 days (again, less than 10 days is not enough to learn enough rules and more than 30 is too long) should be enough of a requirement. If people don't meet the criteria but they think they should get the rights, then they can apply to a page for manual right-granting that should not backlog. Mr.Z-man 13:47, 30 August 2007 (UTC)
- I'd prefer starting off with a relatively high threshold (e.g. 1 month, 300 edits) on the initial implementation. After the system has been in place for at least a month, it would be possible to make an informed decision on whether the standard needs to be relaxed. Chaz Beckett 16:11, 30 August 2007 (UTC)
- I'd go the other way... start off with a low thresh-hold and then raise it if the amount of vandalism getting through is still too high. I think that even a minimal '5 day' requirement, equivalent in terms of vandalism control to semi-protecting every article (though allowing continued editing), would result in a massive and immediate decrease in the amount of vandalism we display to unlogged readers. Most vandalism comes from unlogged IPs and newly created accounts doing something on the spur of the moment. That simple '5 day' requirement would be enough to stop all of those problems... just as semi-protection (which uses the same '5 day' standard) does currently for individual pages. Only the 'dedicated vandals' who create multiple 'sleeper accounts' to vandalize from would still get through. I'd like to see standards which would make things difficult for them too (e.g. 'article edits on 30 different days' or '30 article edits at least 5 days old') without significantly restricting access for new contributors. If a vandal wants to make 30 edits which, when reviewed, will be approved as valid then I think that's a net gain for the project... the effort in reverting their later vandalism and blocking them shouldn't be more than the benefits of those 30 positive edits. If the 30 edits aren't positive then they get some sort of block or delay in gaining the access. --CBD 16:41, 30 August 2007 (UTC)
- I'd prefer starting off with a relatively high threshold (e.g. 1 month, 300 edits) on the initial implementation. After the system has been in place for at least a month, it would be possible to make an informed decision on whether the standard needs to be relaxed. Chaz Beckett 16:11, 30 August 2007 (UTC)
As others have said, automatic assignment is a must. Admins handing it out doesn't scale and we'd end up with a massive backlog. Surveyors handing it out would effectively give all logged in users the right, which is better than what we have now, but a little more restriction would be beneficial in my opinion. I think automatic criteria of 1 month, 50 mainspace edits and no blocks would work well (if the software doesn't support a namespace requirement, and the feature can't be easily added, then 150 total edits would work well enough - mainspace edits are better, though, since it avoids social networkers and trolls getting the right). Admins need to be able to remove the right, and should be allowed to do so with any reasonable cause if the permission was granted automatically. If the permission was granted by an admin, it should only be removed for a genuine policy violation that has happened since it was granted (otherwise it is wheel warring). A way for admins to mark an account as not being suitable for automatic assignment would be good, but we can use short blocks for that if we have to. --Tango 16:54, 30 August 2007 (UTC)
- I'm going to re-add the autoconfirmation to the proposal (1 month, 150 edits), and leave a message inviting Moreschi (who removed it) to elaborate on the downside of autoconfirmation. I would set the bar a little higher (2 months, 300 edits, as before), but 1 month and 150 edits seems reasonable to me.--ragesoss 22:20, 30 August 2007 (UTC)
I support the idea of users being autoconfirmed but dislike the recent addition of a requirement for users to have both a user page and talk page. This seems to be unecessary and I am not even sure if the software could make this a requirement? Some very long established users prefer to not have a user page or have it redirect to their talk page. This could be as they do not want to spend time on their user page or to avoid having a page for vandals to attack. Davewild 17:22, 31 August 2007 (UTC)
- I also dislike the user/usertalk page requirement. By the time a user reaches 150 edits/30 days, I would hope that some other user has tried to communicate with them. As far as a userpage goes, many people think they are unnecessary and don't want them. Especially as WP:USERPAGE is only a guideline, it seems weird to quasi-require users to have userpages. (sorry if that last bit sounds odd, it made sense as I typed it) Mr.Z-man 18:23, 31 August 2007 (UTC)
- A redirect counts as having a page. Voice-of-All 01:51, 2 September 2007 (UTC)
- Nevertheless, it does seem a strange requirement and I don't see a reason for it. Why did you add this requirement to the code, and why is there no configuration variable covering it? -- Jitse Niesen (talk) 05:50, 2 September 2007 (UTC)
- A redirect counts as having a page. Voice-of-All 01:51, 2 September 2007 (UTC)
- Yeah, each one of these requirements seems like a technical answer to a social question: whom do we trust? Having a user page is commendable, but where this proposal is concerned, it looks like nothing more than a hoop to jump through. These weird requirements are not going to hold up to community-wide scrutiny.--Father Goose 06:47, 2 September 2007 (UTC)
- I've been bold and removed the user page requirement, cannot see any arguments for why it should be required and cannot see why a user such as User:FrozenPurpleCube should be prevented from having this right for not having a user page. Davewild 07:59, 2 September 2007 (UTC)
- Yeah, each one of these requirements seems like a technical answer to a social question: whom do we trust? Having a user page is commendable, but where this proposal is concerned, it looks like nothing more than a hoop to jump through. These weird requirements are not going to hold up to community-wide scrutiny.--Father Goose 06:47, 2 September 2007 (UTC)
Should flagged versions be the default reader view?
Some discussants suggest that we keep the current versions of articles as the default view for readers, and offer flagged versions rather than simply foist them upon the unsuspecting readers. However, this largely defeats the purpose of the proposal; it would not allow un-protection of semi-protected articles and offers no benefit to most readers (whose common concern is that they don't know who "used the facilities" last).
Discussion
There are two different 'goals' which people want to get out of flagged revisions. The first is to remove obvious vandalism and disruption from public view in the brief time before it is cleaned up (allowing removal of protection/semi-protection as a 'side benefit'). The second is to be able to identify pages which have been thoroughly reviewed and found to be of the absolute highest quality. The first can only happen if unlogged readers, who are responsible for the vast majority of page views on any given day, see the flagged version by default. The second goal can be met either way... though since it requires extensive review such flags might be updated infrequently and thus argue for display of a more recent version. To me the best approach would be to show the most recently flagged (either 'sighted' or 'quality') version to unlogged users by default or the most recent version if there are no flagged copies of the page. Logged in users of course always see the most recent version. --CBD 11:43, 30 August 2007 (UTC)
- In short, the use of flagged revisions only makes sense if they are displayed per default. The whole point in quality checking articles (on a lower or higher level) is to display the checked versions to readers, unless the reader explicitely chooses otherwise. Without that, I don't see why quality checking would be worth the effort. --B. Wolterding 15:42, 30 August 2007 (UTC)
If we're sticking with just a "sighted" flag, then the flagged version has to be shown by default to be worth having it at all. If we introduce a "quality" flag, then there might be a point in showing the latest sighted version to anons and letting them choose to see the latest quality version instead. --Tango 16:58, 30 August 2007 (UTC)
Is "Sighted" the best term?
The term "sighted" is ambiguous (since it can refer to the opposite of blind) also might get mixed up (especially in non-textual off-wiki situations) with "cited". Instead of surveyors and sighted versions we could have surveyor/surveyed, or inspector/inspected, or proofer/proofer, or something else.
Discussion
We should remove the names for the users who have this access entirely. Just have terms for describing the access... consider that for semi-protection we don't have even that. There is no 'specific term' for the users 'who can edit semi-protected pages'. There isn't even a special term for the access which allows that. Accounts which have existed for 5 days can do it. With flagged revisions there needs to be some term to identify whether a particular edit is 'flagged' or not... I'd actually suggest 'flagged' except that in most conceptions there are two different levels that we should differentiate between. In which case 'flagged' or 'reviewed' for the first level and 'approved' or 'featured' for the second sound better to me than 'sighted' and 'quality'. --CBD 11:43, 30 August 2007 (UTC)
- The name for people that can edit semi-protect pages is "autoconfirmed". A name is required on a technical level (for the permission, and the user group of people with that permission). Whether we use that name, another name or no name at all in non-technical stuff is up to us, and in my opinion, there is no need to use any name. My suggestion for articles that have been flagged is "clean". --Tango 17:02, 30 August 2007 (UTC)
- I originally like the surveyor term, but I like the idea of downplaying any terminology for the user class. I also like "clean", which conveys the main issue that we want to get across with the first implementation of flagged revisions.--ragesoss 17:09, 30 August 2007 (UTC)
- I like the term "inspected" rather than "sighted," but I think the "surveyor" term is fine. Pyrospirit (talk · contribs) 01:30, 2 September 2007 (UTC)
Is the proposal fundamentally un-wiki?
This proposal will create another "class" of users, between (autoconfirmed) non-admin editors and admin editors. In principle, this has the potential to exacerbate problems of elitism and cliquishness among editors. It also might make new users feel less welcome, since their edits no longer instantly become incorporated into the default version of an article; new editors without the surveyor right may feel like second-class editors (and may not want to/know how to request surveyor rights). On the other hand, it might make users more involved from the outset, since they will have a sense of validation if another editor sights their work. They will also have an extra reason to return to Wikipedia (to check on the validation status of their edits).
Discussion
Again I refer to semi-protection... it also 'created another class of users', those who have accounts less than 5 days old. The world did not end. So long as most active contributors receive the access in short order there will be nothing to express 'elitism' over. Most of the people you are talking to on talk pages will have exactly the same ability. The higher the bar to getting the access the more valid these concerns become - and thus why I think most people are opposed to any sort of bar which most new users wouldn't pass over before even realizing it existed. --CBD 12:07, 30 August 2007 (UTC)
- I agree. I think our test case should be if a person who averages only a few edits per day (such as myself) would gain the reviewer status within a month or two. I've already seen some cases of elitism on Wikipedia, but I think that the baseline for this is currently set so low it can do little to contribute to the problem. I am mildly concerned that the bar might creep higher and higher with time (much the way RfAs have) but I don't believe in opposing based on what might happen. Sxeptomaniac 21:43, 30 August 2007 (UTC)
Extent of sighting
Should sighting eventually apply (1) just to pages currently semi-protected or protected (2) some other small subsets as well (e.g. where we now have a "work in progress" subpage for a major re-design) (3) All articles which are not one-sentence stubs and very new articles.
Discussion
(4) To every page in the article namespace. (5) To every page in most or all namespaces. I don't see any reason to exclude new articles or stubs from the flagged revisions system. If stubs were excluded then a vandal could always get nonsense widely displayed by changing a popular article into a stub. If new articles aren't included then whenever the 'new' period has expired someone will have to go through and review the entire thing to create the first flagged copy... not a big deal for most articles, but when there are major events in the news we often have articles go from non-existant to a page long and dozens of footnotes in the course of a day. Also, there are tremendous benefits to this beyond the article space... policies and guidelines in the Wikipedia namespace could be flagged at the 'quality' level such that people could make changes and revisions just as freely as they do now, but they wouldn't be re-flagged as actually part of the policy unless approved by consensus. User pages frequently hit by vandalism could be warded against that without preventing valid updates. Ditto talk pages in all namespaces. Et cetera. --CBD 12:07, 30 August 2007 (UTC)
- In my opinion, we should start introducing sighted versions on the currently semi-protected articles, in order to see how the system scales, but with the perspective of applying it everywhere in the article namespace. For a later stage, I could even think of requiring sighting for new articles before they are ever displayed to the public, in order to catch spam etc. before it actually goes live. --B. Wolterding 15:48, 30 August 2007 (UTC)
- The idea of not showing pages with no 'sighted' revisions makes sense for dealing with spam, new page vandalism, and the like... but I don't think it should be implemented until all existing pages in Wikipedia have been 'sighted'. Otherwise, alot of valid articles which have been around for years would 'suddenly disappear'. If/when revision flagging takes off I expect that there will be a project to go through and review and flag every single article... after which we should be able to track and review all unflagged edits - including new unflagged articles. --CBD 16:25, 30 August 2007 (UTC)
- I agree, first the vast majority of existing articles should be sighted, before requiring sighting for new pages. --B. Wolterding 16:45, 30 August 2007 (UTC)
Sighting should be possible for all pages. If there isn't a sighted version for a page, then just display the most recent. Whether we sight stubs or not is another point entirely, the feature should still be turned on for them. --Tango 17:08, 30 August 2007 (UTC)
- I think sighting should be allowed even for stubs (except possibly one-sentence stubs, as an incentive to either expand it or merge or remove it, since nothing should be a perpetually one-sentence stub). Depending on how successful Wikipedia as managing backlogs, at some point we would hopefully have everything except very new articles sighted. At that point, we could have some sort of default warning template for non-sighted (new) articles, and we could turn anonymous article creation back on. (Of course, that isn't strictly necessary; we could turn anon page creation on anyway. But that's a whole different proposal).--ragesoss 17:19, 30 August 2007 (UTC)
- That's a good point, however it's a completely different point to what we're discussing here. It's just a matter of deciding what the requirements should be for sighting, it's not a matter of technically turning it on or off for particular pages. --Tango 17:24, 30 August 2007 (UTC)
- If every editor who, has an account for more than 30 days, has more than 150 edits, has confirmed an e-mail account, has a user and user talk page gets the surveyor right then we can not try to roll it out slowly. Zginder 12:22, 31 August 2007 (UTC)
Automatic sighting or unsighting
To prevent pages getting overlooked, it has been suggested that after some period (days or weeks have been mentioned) either the latest version should be automatically sighted, or the the sighting should "expire". Both of these would have the effect that the latest version of the page would become visible to casual readers. Contrary to some rumors, the software does not currently allow this, partly for technical reasons and partly because the developers feel it would be confusing to users.
Discussion
The standard argument against this is the Seigenthaler incident. A difference between the two approaches mentioned is that currently there would be numerous sighted version in the history, allowing reverts to a previous sighted version if a problem is found in the current one that cannot be quickly fixed. If sighting expires, there will be no previous sighted versions. PaddyLeahy 10:19, 30 August 2007 (UTC)
- I would prefer that new special pages similar to 'recent changes' be implemented. It should (IMO must) be possible to list all pages with unflagged revisions subsequent to flagged revisions... that is, all pages where users have made contributions that are not yet being displayed to all readers. If volunteers go through those with the same alacrity that they do recent changes patrol today then these new updates should usually be reviewed and approved or reverted in a matter of minutes. If there is a list and it is kept up then no change will ever languish 'overlooked' for long. Automatic changing of the flagged status invalidates the goal... we want to keep changes from relatively unknown users out of the public eye until there has been time for someone to look at them and see that they aren't vandalism. If the status changes automatically then the vandalism gets out before anyone has seen it and we haven't achieved what the system was meant to do. --CBD 12:07, 30 August 2007 (UTC)
- I would prefer that flags expire (if that's possible with the extension) after unreviewed edits sit for a certain amount of time. Some have suggested a day - that is fine for many articles that have many people watching them; but other articles, especially those in more specialized subjects, may only have one or two people watching them and they might not edit every day. Auto-sighting would be bad, that would defeat the purpose. Mr.Z-man 13:57, 30 August 2007 (UTC)
- Or zero people watching them. There are thousands of unwatched pages. In fact, there are more than a thousand unwatched pages whose first character in the title is a punctuation character or the digit "1". Yes, sighting should expire if unsighted edits have sat for a sufficient period of time, but 1 day is far too fast. Perhaps if the most recent change is unsighted and more than 30 days old then it should be shown. GRBerry 15:04, 31 August 2007 (UTC)
- I would prefer that flags expire (if that's possible with the extension) after unreviewed edits sit for a certain amount of time. Some have suggested a day - that is fine for many articles that have many people watching them; but other articles, especially those in more specialized subjects, may only have one or two people watching them and they might not edit every day. Auto-sighting would be bad, that would defeat the purpose. Mr.Z-man 13:57, 30 August 2007 (UTC)
I think that expiring sighted versions would undermine the sighting system, and would open a loophole to vandals. Some people have pointed out that unsighted articles could pile up in a backlog; but avoiding the backlog by auto-sighting the articles amounts to ignoring a problem rather than solving it. Just compare with the current situation: Vandalism patrol does not generate any backlogs - if there are not enough patrollers, vandalism will just slip through uncorrected, and there's no way to tell whether and when this has happened. A backlog would at least show the problem, and attract people's attention. --B. Wolterding 15:55, 30 August 2007 (UTC)
- I agree that it would not be desirable to have expiring sighted version. I also agree with CBD's suggestion of a special page that lists all pages with unflagged revisions subsequent to flagged revisions. It would be very useful to be able to sort such a page by the # of unflagged revisions and/or the amount of time since these revisions were made. Finally, being able to track various metrics on the flagged revision system would be essential to ensuring it benefits the encyclopedia. Chaz Beckett 16:05, 30 August 2007 (UTC)
- Yes, I think putting the metrics and support pages in place from the start will be an important part of assuring success. For instance, instead of flagging 'automatically expiring' would it be worthwhile to set in advance a 'criterion for success' of something like, 'the average time for unflagged edits to flagged pages to be reviewed and approved or reverted must be less than an hour'? It addresses the same concern (valuable edits 'getting lost' and not included in the displayed version) by treating it as an indication that the new process is failing and needs to be adjusted (or removed if unfixable) rather than as a built in software adjustment which could let vandalism slip past. Similar things I might suggest would be; 'the number of protected and semi-protected articles should decrease by at least 50%' and 'the number of daily vandalism complaints received by OTRS should decrease by at least 50%'. If those three things could be accomplished then I'd call flagged revisions wildly successful. I suspect it will do considerably better than those metrics, but changes of even those magnitudes would make it very much worthwhile. --CBD 16:16, 30 August 2007 (UTC)
- I'll agree that without success metrics determined before the experiment, we'll never know if the experiment worked. We already know that the system will have costs on existing editors, require additional processes, additional database and edit load, et cetera. During the experiment phase, we are likely to be only testing pages that are already well watched and contentious. So the standard we test in the experiment phase should be significantly more stringent than the standard we want to live with once all pages are being affected. I think a 30 day standard is about right for once all pages are affected, and I think that a 24 hour standard would be the equivalent for semi-protected pages. (I don't think 1 hour is feasible except for a minor subset of pages where the benefit of this system will be negligible anyway.) GRBerry 15:04, 31 August 2007 (UTC)
Automatically sighting pages is a definite "no". It would make it meaningless. Automatically unsighting is worth considering, but I wouldn't support it - as long as we have enough surveyors, it shouldn't be an issue (a Special: page showing old unsighted revisions would be good, though). --Tango 17:10, 30 August 2007 (UTC)
- Agreed. Hopefully that is the last time someone will mention that ;) Voice-of-All 20:44, 30 August 2007 (UTC)
'Sighted Wikipedia'
After a while, we will have a set of sighted articles, but if we browse those, we will be blue-linked to many articles which may not have a sighted version. Would it be possible to have an option 'sighted only browsing of the wiki', which results in that (registered) editors only see sighted versions at that moment, and that pages that do not have a sighted version turn into redlinks. After all, only the sighted versions are (supposed to be) good. On the articles one could then easily go through the 'redlinks' and see whether the article does exist but does not have a sighted version (and then try to find/create a sightable revision of those 'redlinked' pages). In that way not only pages are turned into good, reliable pages, but complete subjects are becoming that way. The reliability of a page is dependent on the links that are there ..
What about considering that users who are not logged in only see sighted versions, and pages which do not have a sighted version (yet) are not shown (but can be shown via a tab, or when editing that page). I know that that depends on how many reviewers we have, and this should certainly not be enabled in the beginning.
Such a feature would also allow to only see the part of wikipedia which is deemed OK ... as in, we see a reliable encyclopedia. --Dirk Beetstra T C 08:51, 31 August 2007 (UTC)
- The first part of the idea makes some sense and is very interesting as a conscious option (if it could be implemented is another story), but I don't think the idea that it could be applied to non-logged in users should EVER be considered, unless somehow 1.8 million of our 2 million articles actually eventually get sighted... ♫ Melodia Chaconne ♫ 11:41, 31 August 2007 (UTC)
- We should have almost all our articles with at least one recent sighted version. That's the point of having almost everyone able to sight. --Tango 12:05, 31 August 2007 (UTC)
- I don't believe that it is going to be such a big deal. There are about 1500 administrators, if they all would give the sight rights to 5 users (on avarage) you have almost 10,000 users that can sight .. which would take about half a year if they would sight only 1 page a day. But I expect these 10,000 users to be more active in sighting (as they are probably quite dedicated users already).. So even if the threshold is 'all administrators and all registered editors with more than 1000 edits' we get a substantial group that is capable of sighting, and yes, it might be that we are quite busy with that for some time, but I expect that it can be done in a couple of months (most of the pages that don't get sighted in that time, are not seen by many people anyway, and people can always put up a request). After that you can put the system up of 'seeing a sighted wikipedia'. --Dirk Beetstra T C 12:49, 31 August 2007 (UTC)
- Do it quick. I will participate.-BillDeanCarter 12:59, 31 August 2007 (UTC)
Forking this proposal
It's nice to see how progress is being made on various details, but I notice that some bad feelings are caused by people confusing this proposal with the Foundation's work on Flagged Revisions. I suggest splitting this proposal into:
- /Sighted revisions: Basically the current proposal.
- /Quality revisions: The part of this proposal that was removed.
- /Opening protected pages: Jimbo's approach.
This page would then serve to describe the software features in human readable form, and only contain information that is really backed by the Foundation. It would then link to the various community generated proposals on how to use this feature on the English Wikipedia.
How does this sound? --Merzul 09:17, 31 August 2007 (UTC)
- I don't really see the relevance of the Foundation to this discussion. It should be determined by community consensus, the same as any other new policy (that isn't related to a legal issue, at least). The fact that the foundation is supporting this proposal (apparently - I haven't actually seen any such support, I've just seen people saying that it exists) is great, but it's not their decision to make. --Tango 12:08, 31 August 2007 (UTC)
- Separating the "sighted revisions" proposal from "quality revisions" might be a good idea, since some seem to confuse the two concepts. I don't quite understand how the third approach differs from the first one, however. --B. Wolterding 14:02, 31 August 2007 (UTC)
- Separating this mess (which is the best description I can give it) somehow is a good idea. We will eventually need a Help namespace page on the technical details of the system, which we might as well draft now, even if we don't put it into the Help namespace yet.
- We should have one page for each proposal. As I see it, the only information "that is really backed by the Foundation" is that it is up to each Wikipedia community to decide whether or not to implement this system at all, and if they do implement it how and when to do so. That isn't worth a separate page; it can fit in one sentence at the top of all pages.
- Forking the sighted section from the quality section is reasonable. There is no solid reason to intermingle those proposals and discussions. I personally am leaning toward not implementing sighted versions at all, but have no opinion yet on quality versions. GRBerry 15:11, 31 August 2007 (UTC)
- Yes, the Foundation has made this capability available. The problem is when the current proposal is interpreted as also part of their work. A base page could include the general outline of the system, news about the development, and any official statements from the foundation and developers, see this valid concern. I think some people seem to have a hard time distinguishing what fanatic supporters of flagging (like myself) are pushing, and what the Foundation has actually stated.
- In any case, some editors have argued very carefully against sighted revision, but are more favorable towards flagging revisions when articles go through more QA processes, like FAC and such, and so that's indeed a separate proposal to flesh out.
- Finally, Jimbo's approach is basically the same as this page, but he seems to want to stress the openness instead of the quality improvement aspects. Personally, I don't think that being "even more open" is a good thing. I would prefer Wikipedia to move from the current "The Encyclopedia that any vandal can edit" to "The Encyclopedia that any well intended person can edit". The basic idea is the same, but Jimbo's proposal would start out stating that the main problem is how vandalism forces us to effectively close down entire pages to public contributions. He would probably then suggest that sighted revision is a more fine-grained mechanism that let's us keep pages open for well-intended editors, while still deterring vandalism. Ah, ehm, it seems that it is indeed actually the same thing, just different ways of selling it :) --Merzul 17:19, 31 August 2007 (UTC)
- I don't think this is a good idea. We've pretty much winnowed the original proposal down to just focusing on Sighted versions, which seems to have the least resistance and will probably be the first step in the usage of Flagged Revisions. The split will confuse newcomers to the discussion. I think Quality versions proposals and anything else can be linked from Wikipedia:Flagged revisions, but the sighted revisions material should be moved back.--ragesoss 19:43, 31 August 2007 (UTC)
- I disagree. There are already editors, myself included, who think sighted versions is a harmful idea, but are open to quality versions. Realistically, sighted versions as a long way to go before it is even worth experimenting with on a few test pages. Before we get there we'll probably see competing proposals, and making each proposal a sub-page of the overall explanation is an excellent idea. As a newcomer today, I'd much rather have come to the new main page and gotten the opportunity to understand the context before I had to deal with any specific proposal for any of the types of versions. GRBerry 21:09, 31 August 2007 (UTC)
- I fully support Merzul's move. Some people support Sighted versions but not Quality versions, some the other way around. The only similarity between the two should be the technical implementation; all the procedural and user-rights details should be workshopped completely separately.--Father Goose 22:09, 31 August 2007 (UTC)
Not "user friendly" policy
The main page is extremely off-putting for a new visitor. It opens with a message from Jimbo going into the detail of something that's not yet been explained. Then, we have a techy section about something that's not yet been explained. Finally, after the ToC, you finally hit some explanation (phew!) but are immediately hit by a header that refers to "Sighted versions" and opening words that refer to "Sighted revisions" (or is it the other way round? Not important - it's off-putting) Frankly, anyone who persists to actually working out what this important proposal is about deserves a medal. This is Wikipedia, where we love to "edit mercilessly" to improve articles - so why's this proposal such an exception. Simple suggestion: create a lead section that follows the principles of WP:LEAD and all will be well. --Dweller 18:17, 31 August 2007 (UTC)
- I did the second best thing, although that's not very conventional, but I put an infobox saying other people should be bold in editing :) About the signed comment, uhm normally one would remove it as totally inappropriate, perhaps paste a warning on the User's talk page. :) In this case, perhaps it should be moved to the top of this talk page? --Merzul 19:12, 31 August 2007 (UTC)
- I agree that Jimbo's comment should be moved to the top of the talk page, above the TOC. Its immediate purpose has been served, and blog link traffic typically falls off quickly.--ragesoss 19:16, 31 August 2007 (UTC)
It's now much improved. Whoever inserted the "nutshell" made a valuable contribution. I've amended it for greater clarity. Thanks all. --Dweller 09:57, 2 September 2007 (UTC)
Anonymous editing
One of the advantages of "sighting" is that it reduces the incentive to vandalise, since vandalism will not show up immediately. But, isn't the opposite true as well? Doesn't it also reduce the incentive to make positive contributions because they too will not show up immediately? In this world of ever-faster Internet connection speeds, no one is going to tolerate even a 5-minute delay for their change to appear. I hadn't thought of it before, but this essentially shuts out all anonymous users. They can technically still edit, but there is absolutely no incentive for them to do so. — Black Falcon (Talk) 18:22, 31 August 2007 (UTC)
- It gives incentive to sign up, and I also think one shouldn't underestimate the psychological situation this creates. New anonymous editors, especially the most promising ones who might become regular contributors, will be eager to see if their edits get approved. They'll check back obsessively, and in the meantime they'll find other things to edit. I remember when I first made major edits, I kept checking back to see if anyone changed it or commented, and I think a lot of new editors do the same. Additionally, for some people who never edited because they saw no attempt at quality control, they'll be more willing edit. Although some will no doubt respond as you suggest, I think overall it's more likely to increase the level of contributions than decrease it.--ragesoss 18:40, 31 August 2007 (UTC)
- You checked back obsessively because your edits appeared. I don't think most new users will immediately grasp the notion of 'sighting' and understand why their edits didn't show up. Also, "sighting" is not really "quality control" unless it involves a significant investment of time and effort (which various comments on this talk page suggest is not the intent of the proposal, despite the text of the "Requirements of Sighted versions" section). — Black Falcon (Talk) 19:43, 31 August 2007 (UTC)
- After the initial (first-time) sighting is done, are we really talking about that significant of an effort to flag revisions? Unless I'm misuderstanding something, we'd only need to look at the changed text, not the entire article. If an article is 30K and only two lines are changed, the reviewer only has to examine those two lines. Chaz Beckett 19:53, 31 August 2007 (UTC)
- Yes, but someone has to make that initial, detailed, GA-review type evaluation in order for subsequent sightings to be valid. Otherwise we'll just be burying original research, non-obvious vandalism, false information, and other problems behind dozens of layers of sightings. Here we must rely on the meticulousness of surveyors. If we limit surveyor rights to a small group (i.e. admins), we'll have a hard time sighting 1.9+ million articles (and growing). If we make everyone with an account (30 days, 150+ mainspace edits) a surveyor, the reliability of the average sighting goes down since (1) people will cut corners and (2) plenty of POV-pushers and the like meet those criteria. See also my comments in the section #Abandon sighted versions. — Black Falcon (Talk) 20:06, 31 August 2007 (UTC)
- I'd like to see us having a big push on getting anon editors to register when they edit, explaining the future benefits, why it'll be good and so on. It might also be time to have a purge on accounts that have never edited and don't have an e-mail address registered. It's possible lots of users are being put off from registering because they can't remember their username or can't find a short, easy to remember username. Nick 18:48, 31 August 2007 (UTC)
- A proposal to purge accounts was suggested about 2 years ago. It seemed to have gained consensus, but it was denied by the developers. My personal preference is not to force the issue of registration: if someone doesn't want to register, that's fine. Gentle nudging is appropriate, but delaying the appearance of edits has nothing 'gentle' about it. — Black Falcon (Talk) 19:40, 31 August 2007 (UTC)
- There's a belief in about half the camp here that people create accounts so they can customize the appearance of Wikipedia which is one reason the devs aren't keen to have a purge. It's possible we'll see a fairly substantial upsurge in the number of accounts being used to actively edit if that's the case. I'd still like to see a message that goes with the anon notice about not being logged in that gently nudges the user to create an account. Nick 01:01, 1 September 2007 (UTC)
- I don't really agree that no one will tolerate even a 5 minute delay, but there's always the options to create an account. It's really not that hard to create one and actually provides more privacy than editing anonymously. Chaz Beckett 18:54, 31 August 2007 (UTC)
- I think a lot of new users don't really believe that their changes will appear instantly until they try anyway (judging by the number of test edits we end up reverting), so what different does it make? --Tango 19:26, 31 August 2007 (UTC)
- Good point. New users may not even have the expectation that their edits will immediately be visible to the world. If/when flagged revisions go live, anons should probably be notified on the edit screen that their contributions will be reviewed before becoming visible. Chaz Beckett 19:31, 31 August 2007 (UTC)
- I don't think that they "don't really believe that their changes will appear instantly", but rather that they just want to see what happens. That is, they want to effect a change. — Black Falcon (Talk) 19:40, 31 August 2007 (UTC)
- Perhaps we should change the text at MediaWiki:Userlogin (the text displayed to non-logged in users in the top right corner) to "Sign in / Create username" instead of "Sign in / Create account"(the current version). Account sounds a lot more formal (like a bank account) when creating a username here is a really easy process (way easier than most other sites) and all you are need to enter in the form is a captcha word, a username, and a password; even an email is optional. It would not be a huge difference, but it could help stress the easiness and simplicity of creating an account. Another small, but noticeable, thing to encourage anons to create an account could be adding a link to Wikipedia:Why create an account? on MediaWiki:Anonnotice. Mr.Z-man 20:48, 31 August 2007 (UTC)
- Two-thirds of registered usernames are never used to edit. Seems plausible that there are more unused usernames than anonymous IP editors, so perhaps getting people to register is not the main problem... PaddyLeahy 22:19, 31 August 2007 (UTC)
- There's a ways round this problem, I believe. Those users without surveyor permission would always see the page as they last edited it, but with a small disclaimer that says the page may not appear this way to other visitors. Nick 00:19, 2 September 2007 (UTC)
- Perhaps we should change the text at MediaWiki:Userlogin (the text displayed to non-logged in users in the top right corner) to "Sign in / Create username" instead of "Sign in / Create account"(the current version). Account sounds a lot more formal (like a bank account) when creating a username here is a really easy process (way easier than most other sites) and all you are need to enter in the form is a captcha word, a username, and a password; even an email is optional. It would not be a huge difference, but it could help stress the easiness and simplicity of creating an account. Another small, but noticeable, thing to encourage anons to create an account could be adding a link to Wikipedia:Why create an account? on MediaWiki:Anonnotice. Mr.Z-man 20:48, 31 August 2007 (UTC)
- I don't think that they "don't really believe that their changes will appear instantly", but rather that they just want to see what happens. That is, they want to effect a change. — Black Falcon (Talk) 19:40, 31 August 2007 (UTC)
Some people are saying that a 30 day backlog to sight edits no one will wait that long they will leave the project. Zginder 21:01, 31 August 2007 (UTC)
- I highly doubt it will be 30 days. On more technical articles it may take a few days for someone with enough technical knowledge to come around to it but for most articles they should have new edits sighted or revrted within 48 hours (ideally within a couple hours). Mr.Z-man 22:11, 31 August 2007 (UTC)
- Good edits should be sighted at the same speed as vandalism is reverted now, which is generally within an hour or two. Recent changes patrollers already look at pretty much every edit, now they'll just have to click "sight" instead of just ignoring good ones. --Tango 12:59, 2 September 2007 (UTC)
- There is all this "should be" stuff, but no promises. If I were an anonymous editor and it said "Your edit will be review" I would think that some big company with ten editors were going to get around to it sometime, maybe never. Zginder 13:02, 4 September 2007 (UTC)
- The message should dispel that notion by its phrasing, along the lines of "When another editor agrees the change is an improvement, it will be made visible". -R. S. Shaw 20:19, 4 September 2007 (UTC)
- (e/c reply to Zginder) Wikipedia is all volunteer, we can't require people to do things in a certain timeframe (or even require them to do anything at all) and this software has never been used on a large wiki before so there is no real way of knowing precisely how it will work here. It will probably be used on the German Wikipedia first, but their implementation may be quite different from ours, so our results may differ. We can make estimations based on similar processes (which is what we've been doing), but even those are not precise. This is not an automatic system here. This is thousands of people at individual computers around the world, each editing at different times, different articles, and different tasks. Everyone does whatever they want to do, whenever they want to do it; its near impossible to predict. Hypothetically you could survey every user who would be eligible for this (the number would be in the thousands) as to whether they would use the system and how much they would use it. Combine that data with the number of articles this could be used on (possibly all) and the number of edits per day (or hour for a better estimate) on those articles by users who are and are not eligible to "sight" articles. Even then, whatever statistics you may gather, it is still only an approximation and relies on individual responses. Of course you could save time by only using a percentage of users and articles but as your percentage gets smaller, so does your accuracy. Or, instead of wasting hundreds of hours collecting and processing data to get an estimate, we could hope this works as well as current processes. If it does, great, if it doesn't, we tinker with the policy and if its a horrible failure we deactivate/uninstall the extension. Mr.Z-man 20:31, 4 September 2007 (UTC)
- The message should dispel that notion by its phrasing, along the lines of "When another editor agrees the change is an improvement, it will be made visible". -R. S. Shaw 20:19, 4 September 2007 (UTC)
- There is all this "should be" stuff, but no promises. If I were an anonymous editor and it said "Your edit will be review" I would think that some big company with ten editors were going to get around to it sometime, maybe never. Zginder 13:02, 4 September 2007 (UTC)
- Good edits should be sighted at the same speed as vandalism is reverted now, which is generally within an hour or two. Recent changes patrollers already look at pretty much every edit, now they'll just have to click "sight" instead of just ignoring good ones. --Tango 12:59, 2 September 2007 (UTC)
Reversions to sighted versions automatically sighted?
Is there any reason that a reversion to a sighted version should not be automaticall sighted. If the diff between the two versions shows no differences, I can't see why we wouldn't have the new version auto-sighted. That would at least eliminate the burden of sight checking straight reversions, so that a vandalous edit doesn't mean the article has to be sight checked all over again. (Yes, I consider sight checking as currently proposed to require significantly more work than just reading the article, and a lot more work than reading the diff.) GRBerry 02:51, 1 September 2007 (UTC)
- One advantage of sighted versions is that there is really no need for simple reverting, hence we could have much cleaner and shorter history lists which more clearly reflect the development of the article. Furthermore, reverting feeds trollish vandals, encouraging them to vandalise again to make more work for us. If a new unsighted edit is bad, don't sight. If you spot a problem in a sighted version, remove the flag so the previous sighted version becomes current. Only make a new revision if you have an improvement to offer over the current sighted version.
- Now I think of it, there is a problem with this advice: to avoid duplicated effort we need a way of alerting other editors that a version has been deliberately left unsighted. Since flaggedRevs offers many levels of flagging, one option would be a "don't use" flag which would have no effect on page display but, like other flags, shows up in history, watchlists etc. Like the sighted flag, it could be removed by any "surveyor". PaddyLeahy 08:22, 1 September 2007 (UTC)
- We will still need to revert vandalism for the sake of logged in editors. Automatically sighting identical versions shouldn't be a problem, but I don't see much point - non-logged in readers will see exactly the same text either way. --Tango 13:01, 2 September 2007 (UTC)
- I thought the point of sighting reversions was well explained by PaddyLeahy: sighting serves as a signal to fellow editors that it has been at least cursorily checked. -- Jitse Niesen (talk) 01:08, 3 September 2007 (UTC)
- Once the first version is sighted, to sight a new version one just has to look at diff. To make the first sight only a lot of work is needed. Zginder 23:22, 4 September 2007 (UTC)
- I thought the point of sighting reversions was well explained by PaddyLeahy: sighting serves as a signal to fellow editors that it has been at least cursorily checked. -- Jitse Niesen (talk) 01:08, 3 September 2007 (UTC)
- We will still need to revert vandalism for the sake of logged in editors. Automatically sighting identical versions shouldn't be a problem, but I don't see much point - non-logged in readers will see exactly the same text either way. --Tango 13:01, 2 September 2007 (UTC)
New opinion
I changed my opinion on this. The page is very convincing. I hope this gets accepted. -FlubecaTalk 23:58, 1 September 2007 (UTC)
- I also have slowly been drawn to support this. We definitely need some form of quality control system. Most people think "people can put whatever they want on Wikipedia" and therefore, they really don't trust it. The proposal has flaws in it, but I'm too ambivalent in my position to seriously input on anything. What I will say is auto-sighting should not be done. Wikipedia as a whole cannot endorse any article automatically. Also, the "surveyor" requirements are too strict. Two weeks and ten edits would be plenty. Most vandals (in my experience) are the hit and run type. The others we will just have to deal with. It's better to have to stop a few more vandals that alienate a lot of constructive contributors. I know we all want to stop vandals, but we cannot close our doors. Our goals should be to improve our reliability, reduce vandalism, and do this without anyone needing to yell "the citizendium is near". Psych less 23:18, 5 September 2007 (UTC)