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.

I'd like to see a way for an issue to be marked as resolved. This would mean it's believed resolved by users. In the interface it would appear as "This issue was marked as 'probably resolved' on <DATE>" and items marked as resolved with no further "upvotes" would show as dimmed. Subsequent upvotes would remove the "resolved" flag.

Rationale - issues may be resolved quickly or take forever. It doesn't happen as a function of time. The ability to dim issues believed fixed will keep feedback relevant. Otherwise items rated highly will stay at the top a long time when they are no longer real priorities. An issue that's still live can be undimmed just by re-upvoting it, indicating it's still a concern. FT2 ^{(Talk | email)} 20:07, 9 January 2012 (UTC)

Yup; we're already experimenting with that feature. Okeyes (WMF) (talk) 22:34, 9 January 2012 (UTC)

I'd like to see a basic categorization dropdown for issues (spelling/readability, out of date, factual error, balance, mere opinion [OR], dubious/unverified fact, multiple issues/other). That way the feedback page can be filtered for specific kinds of issue such as spelling or serious errors, and a quick skim will give anyone a sense what areas the article's getting feedback about. The wireframe mockup suggests skimming feedback would be easier that way. FT2 ^{(Talk | email)} 20:07, 9 January 2012 (UTC)

We're definitely going to have some sort of filtering and sorting, and we're experimenting with tags; the two could go hand in hand! I'll bring your suggestion up :). Okeyes (WMF) (talk) 22:35, 9 January 2012 (UTC)

What I'd ideally like is to capture a textual comment, plus a green thumb up/red thumb down, plus an article area, which would help a lot when skimming to see which areas have issues. Also very useful for tracking. The difficulty is the more we ask for the more likely it just puts people off and the loss outweighs the benefits. Still, mentioning it here in case there's a way to do it. Perhaps after capturing textual comment + area of concern, and the user clicks "ok", maybe then ask "is this a compliment, concern or question?" -- if they've already entered their feedback perhaps it would be easier accepted. It would be useful. FT2 ^{(Talk | email)} 12:33, 10 January 2012 (UTC)

Possibly useful; as you say, it's always a tradeoff between detail and ease of use. One of the designs does actually include "praise" "issue" "criticism" "question" tabs, so we'll see whether that gets used. Okeyes (WMF) (talk) 12:34, 10 January 2012 (UTC)

A quick thought on wording. A lot of the time even if "told" something's ok, people don't truly believe it, or hesitate, or have qualms. It's one reason we emphasize "YOU can edit" so strongly.

In designing the form, can we test messages which are very strongly worded to encourage people to comment?

• "Can you think of anything that would improve this article? Tell our editors now!"
• "Is there anything you expected to see, or would make this more helpful? Let an editor know now!"
• "Anything you think isn't top quality here? Let us know now!"
• "Any outdated or missing content you can see? Give a link to a reputable newspaper, book, or journal to our editors now!"

FT2 ^{(Talk | email)} 16:32, 12 January 2012 (UTC)

We're definitely going to test new wordings, but this'll be after we've picked a final design for the form :). Okeyes (WMF) (talk) 18:04, 16 January 2012 (UTC)

The Article Feedback Tool, Version 5 is the new version of the existing Article Feedback Tool and is being developed by the Wikimedia Foundation in partnership with the editing community. The major change from the previous version is the replacement of the star rating system by a box which allows free-text comments and suggestions. The hope is that this tool will allow readers to provide genuine, helpful feedback that editors can use to improve the page. We're asking readers: How can this page be improved?

Their feedback will be displayed on a special “Feedback Page” associated with the article where both readers and editors may view the feedback. The goal of that page isn't so much to surface expressions of opinion, but to give editors information and tools that will help them improve articles in partnership with readers who have made useful suggestions.

While the exact features on the Feedback Page are still under discussion, there are certain issues we know we will need to address. Specifically:

1. Who can hide feedback?
2. How can illegal feedback be deleted?

The Wikimedia Foundation is requesting comments on a proposal for addressing the above two issues. Those interested in discussing other issues related to the Article Feedback Tool should visit the project page. This RfC will run for two weeks, from 6th January until midnight UTC on the 20th. 20:43, 06 January 2012 (UTC)

The wireframe on the right provides an illustration of what the Feedback Page might look like and which tools might be shown for different user groups. Please note that this wireframe is only illustrative – the features on the page should not be taken literally, but as representations of ideas. As such, the features on the wireframe are sure to change and the final version will look different than this rough graphic. Those interested in helping use develop these ideas should visit the project page.

Here is a list of the proposed permissions:

We would like feedback as to whether this is acceptable, or what other suggestions or ideas you might have. In particular, we would like feedback on the following areas:

Thanks to all the editors who commented, and to Tom Morris for summarising what is, in a lot of ways, a novel kind of RfC :). I should have a response for you all by the end of the day! Okeyes (WMF) (talk) 11:50, 24 January 2012 (UTC)

I've misunderstood something up till now - it was my understanding that feedback would be posted to the talk page of an article. This is the logical thing to do, fits into our current policies/practices, reduces the extra work (i.e. having to keep an eye on the new page) and allows an obvious way to work on ideas posted as feedback (and respond). Why are we re-inventing the wheel with a new feedback page? It doesn't seem very sensible design... :( --Errant ^(chat!) 09:42, 11 January 2012 (UTC)

(BTW the reason I consider the talk page a much better venue is because this then also allows us to show people giving feedback the collaborative stage used to work on articles - and hopefully encourage them to contribute more. Hiding the normal way of editing/discussing behind a feedback page is going to lead to confusion) --Errant ^(chat!) 10:27, 11 January 2012 (UTC)

Good reasons all :). Well, we're stuck in a bind on that over here. On the one hand, it's logical to send them talkpage-wise. On the other, well...have you seen our talkpage architecture? :P. The reasons for liking an interim dedicated "feedback page" before moving to talk:

1. Our talkpage architecture sucks. Give a standard talkpage 200 comments a day from the Justin Bieber article and it will simply lag; it has no way of dealing with concepts like "only show us the top 20", say.
2. The distribution of feedback is a long tail; if we move straight to talkpage we're asking editors on the really high-profile pages to deal with a constant and unremitting stream of comments without any of the useful tools that are being provided to balance the scales.
3. It means forcing commenters to use markup in their responses.
4. The current tool just isn't designed to translate words into syntax ;p.

Now, obviously there'll be a "post to talkpage" function that can be used, but the issues with talkpages and feedback (particularly no. 2) makes us quite wary of just dumping stuff straight to the talkpage. Whether the special feedback page could be subst'd directly into the talkpage is another thing entirely; I can find out, if you'd find that useful? Okeyes (WMF) (talk) 11:23, 11 January 2012 (UTC)

There must be some way to integrate this tool into the talk page, whether by setting aside a section of the talk page for feedback or some other magicoding. I agree that posting feedback directly on the talk page is unhelpful, especially for extremely popular articles like Barack Obama or Lady Gaga, which may receive hundreds of (often useless) feedback messages each month. However, if a feedback box or section could be stuck in the talk page, under a flaggedrevs-type approval based system (where submitted feedback is considered pending and only helpful suggestions are approved for general view—and especially if similar suggestions could be merged under one heading) it would allow for discussion of the feedback as well as easier tracking for popular pages. Otherwise, I don't see how a list of five hundred ideas that are either quite similar, or possibly controversial, will help as people are going to have to copy-paste the ideas back onto the talk page for discussion (and then delete the duplicate feedback?) /ƒETCHCOMMS/ 23:58, 12 January 2012 (UTC)

Yeh I see the difficulties there.. I'd suggest that maybe finding a way to weed out the useless/uneeded comments would be a good exercise though, and see what comes of that. For example at the moment the UX tries to get you to enter a comment - ideally you should be able to just say Yes/No, that gets saved, and then a comment is optional. --Errant ^(chat!) 14:14, 13 January 2012 (UTC)

A comment should be optional, actually; I'll check with the foundation devs. Fetchcomm's suggestion is very close to what we're investigating; a dedicated subst'd stream of "comments which have been posted about this article and have been marked by a user with 'post to talkpage'". Okeyes (WMF) (talk) 15:23, 13 January 2012 (UTC)

It is, I think, optional. But the process now is: click yes/no, get prompted for a comment, then hit send. I wonder if a better path would be: click yes/no, this is recorded with feedback that it worked, get prompted (in an overlay?) for an additional comment - with instructions on what would be useful to hear. We A/B tested a similar sort of thing at work (aimed at getting feedback on our products from the ~100,000 users) and this change of workflow did a lot of the quality of the feedback. --Errant ^(chat!) 15:27, 13 January 2012 (UTC)

Gotcha! Altering the methodology normally comes up with some pretty interesting tweaks; I'll forward this idea to the devs. Worst case scenario, we can stick it in our "what to do if even the best design comes out with a whole lot of junk" warchest. Okeyes (WMF) (talk) 15:28, 13 January 2012 (UTC)

Apparently, to paraphrase ESB, they got there just before you did ;p. This is totally where we're going with the methodology. Okeyes (WMF) (talk) 18:05, 16 January 2012 (UTC)

My vote is that the comments not go to the Talk page. I believe that most of the feedback will be from IPs and a certain portion will be spam. Can there be a "Feedback" page, maybe inside the Talk page? In this manner, we can get quick and dirty feedback on that page, and leave the Talk page for more in-depth and positive improvement suggestions. Otherwise the Talk page will be cluttered with "I like this page" or "your mom", instead of proper things like "Can we incorporate the information from THIS source? I propose something like ...". YouTube has a feature where if you post comments too quickly, it asks for a CAPTCHA. That might be a good idea to prevent bots from blasting us with junk. But all in all, I'm really happy that this feedback tool is going live. I saw it on SOPA and was happily surprised, but then was a bit disappointed when I didn't see it on Folding@home. Get it going! :) Jessemv (talk) 03:59, 17 January 2012 (UTC)

Will do! :). Yeah, we're going to have spam filters up and running before full deployment, but it's probably a good idea not to post direct to the talkpage. I do like the idea of having the feedback page substituted into a portion of the talkpage. Okeyes (WMF) (talk) 04:49, 17 January 2012 (UTC)

Perhaps a compromise is to both have a feedback page, and also to embed a frame or a subview in the talk page giving a subset of the feedback, like the top feedback or most recent feedback? On a related note, how will article authors indicate when a particular feedback comment has already been addressed? Dcoetzee 08:17, 17 January 2012 (UTC)

"Feedback" is a type of "Talk", only easier and less formal. I'm thinking something like an inner tab at the top of the page, or at least a big obvious nice-looking link. I think some kind of embedded frame is nice that shows either the most recent feedback or the most constructive unaddressed feedback (or both?) but of course there should be an easy way to view more of it. I believe we should be able to vote up or down on whether the feedback is constructive or needed, and then mark it as addressed when the change is made. I don't know whether the vote shows up as "12 for, 3 against" or if there should be some kind of bar that illustrates feedback support, but I think some kind of feedback on the feedback is important. Maybe there should be an option to move a particular feedback to the talk page to launch a discussion on it? Just some ideas. As for the spam filter, I would strongly recommend looking at the technology behind Cluebot-NG, because that tool can revert Wikipedia vandalism without any sort of blacklist. Of course spam of Wikipedia articles seems different than spam of the feedback tool, so there might have to be some modifications of its software or something. Cluebot also gets active feedback as to how correct it is based on reports from Wikipedians when they revert false positives. This kind of thing could be done by looking at how many Wikipedians think the feedback is constructive/appropriate or not. Jessemv (talk) 23:20, 20 January 2012 (UTC)

this might be a useful thing to look through :). (there are images of it further in development, but I cannae find them right now, cap'n). Okeyes (WMF) (talk) 01:09, 21 January 2012 (UTC)

On page 3, I like that little link to the Feedback section. I would suggest that you change the font, or make the background blue. This way it's more obvious and noticeable, without going overboard. But how did they get to the screen on Page 4? I can't seem to follow that step. Did they edit the Talk page? Because providing quick feedback is completely different from editing the Talk page and I believe we should maintain that current ability. This feedback tool should be supplementing the Talk page, not replacing it in any way. But I think the screen on pages 12-20 are pretty much perfect. Looks good! Jessemv (talk) 00:25, 22 January 2012 (UTC)

I see some commonality between the free-style feedback tool as described here and the objective of this strategy-wiki proposal (which I put forward). While the tool as described here only addresses making it easier for a reader to signal an issue concerning article quality, the strategy-wiki proposal also discusses means for making it easier for editors to find articles with issues in the scope of their expertise or interest.

I think it will make the feedback tool much more useful if we can likewise implement some way to search for feedback by topic and nature of the issue.

Note also that in the strategy-wiki proposal the feedback delivered gets integrated with the article's talk page. That too may be useful here.  --Lambiam 19:49, 9 January 2012 (UTC)

That's a really awesome idea! There are a couple of proposed features we're playing around with (subject-specific tags, for example, tied to comments) that might work well with that; I'll bring it up with the developers. Unfortunately everything is a bit hectic at the moment, so I can't promise when they'll get back to you on that :). Okeyes (WMF) (talk) 20:13, 9 January 2012 (UTC)

Looking here: whereas most of the comments do not require any action, some do (errors, comments etc). These need to be handled by either editing the article or moving the useful feedback on the talk page of the article. To avoid double work, it would be handy to let editors mark the feedback pieces where the action has been taken. In Russian Wikipedia, the feedback ends up in the Wiki page as separate sections, and there just whoever takes action, closes the section with ((close)). The closed sections get archived by bot. Here, since the output is not a wikipage, this is not applicable; besides, 90% of the feedback requires no action. May be it would be a good idea to convert the output to the wiki page, and then handle the feedback as I described?--Ymblanter (talk) 22:04, 9 January 2012 (UTC)

I wouldn't say 90 percent; you have to take several things into account:

1. This is a stream of comments from 3 different versions, 2 of which will be deprecated; there's thus a worse signal to noise ratio than there will be when it's actually deployed;
2. This is a stream of comments without an abuse filter or spam blacklist plugin (see above);
3. The feedback form is deployed on 0.6 percent of articles - it's likely that the ratio will alter, at least slightly, when full deployment comes around.

That being said, we have taken these things into account :). We're messing around with both a "move this to the talkpage" function and a "mark this as a resolved issue" button. Okeyes (WMF) (talk) 22:33, 9 January 2012 (UTC)

Just saw this in the feedback stream: "Where was my feedback saved?" I have to admit that was my first question too after using it. Perhaps you could provide the user with a link to the feedback stream after they have submitted their feedback. Kaldari (talk) 09:43, 12 January 2012 (UTC)

Yeah... saw that too, but I presume it will be evident in the final product where a user can go to find their feedback listed. For the moment it's quite hidden because there's no systems in place to remove obscenities etc.

But on that note... The ratio of obscenities to useful feedback is off the scale! I'd thought there would be some people making drive-by random comments but nothing on this level... There's a couple of compliments from people saying things like "I was looking for her ancestry and found it quickly!" but the overwhelming majority of posts are things like (and I quote), "Not enough farts.", "This made me wanna /wrists." and "poopoo." I assume that the compliments come from the versions asking "Was this article helpful", and the vandalism can come from any of the versions, but I'm genuinely surprised how few actually contain constructive feedback :-( Wittylama 13:54, 13 January 2012 (UTC)

The stream of junk is the main reason we're not linking to the stream at the moment; we're testing lots of different versions, some of which (unsuprisingly) produce a load of junk (or "poopoo", as one astute reader has noticed ;p). The final version is not going to look anything like this comment stream; the most promising design is 60 percent "useful" to 40 percent "not useful" (and a large portion of the "not useful" comments aren't vandalism or abuse, but are more "I have no idea what the hell this person meant" - I have some ideas as to where that issue is coming from). Then you take into account the fact that this thing will have an abuse filter, a spam blacklist, and no blocked IP or account can use it, and things start to look more promising :). Okeyes (WMF) (talk) 15:26, 13 January 2012 (UTC)

That sounds promising :-) Which is the version that's getting the 60/40? I'd like to hear your thoughts on where that issue of nonsensical comments is coming from too. When you're looking at identifying the 'successful' versions, can I request that your question to yourselves be "Are these comments 'actionable' for improving the article" and not merely "are these comments comprehensible". "I found what I wanted very quickly" might be a satisfied customer (someone who might later start editing too), which is great, but that is still completely useless in terms of helping the existing community to improve the article. Wittylama 23:04, 13 January 2012 (UTC)

It's "version 1"; my money sez it's because of the directness of the question. We have been making the non-offensive/actionable distinction; the editors rating the comments were asked to divide into "useful" or "not useful" (that's where the 60/40 split comes from) as opposed to simply "offensive/actively unhelpful" "not". The numbers may shift slightly, because that feedback was categorised during the holiday period, which is always a bit weird, but we've just finished a second round of hand-coding in a "normal" period that should provide additional data :). Okeyes (WMF) (talk) 20:35, 14 January 2012 (UTC)

Increase width of the feedback box to 100% of page width. Currently it appears in left half of the page looks asymmetric and out of place. Change "submit ratings" button alignment to left, and increase width of < div > to same as article or navigation boxes that appear just above, for sake of aesthetics. Heading indexing (talk) 04:52, 10 January 2012 (UTC)

The problem there is that it would appear to be incredibly large - something that also messes with aesthetics. Okeyes (WMF) (talk) 10:15, 10 January 2012 (UTC)

Is there any way to disable the AFT on a per-article basis, please. There are a number of small/medium articles for which I am "curator" and I don't think they are particularly useful ways to help me gather improvement suggestions. I had an email from someone asking whether I had read their comment, which obviously I had not. My preference is to use the talk page :) (there are other articles I would find this more useful on, this is just for the occasional article) --Errant ^(chat!) 09:45, 11 January 2012 (UTC)

There isn't, no :(. One thing to take into account is that there will be a "centralised" feedback page containing all feedback from all articles, so hopefully with small-to-middling articles, the workload can be picked up there :). If it doesn't work, we'll revisit. Okeyes (WMF) (talk) 11:31, 11 January 2012 (UTC)

Whatever happened to Category:Article Feedback Blacklist? MER-C 06:39, 12 January 2012 (UTC)

That should still be active and working with version 5; is something wrong with it? Okeyes (WMF) (talk) 11:50, 12 January 2012 (UTC)

No. I was wondering why you didn't mention it -- slapping it on a page seems to do what Errant wants. MER-C 13:49, 12 January 2012 (UTC)

Well, the blacklist is meant for things like disambig pages, primarily. Okeyes (WMF) (talk) 14:22, 12 January 2012 (UTC)

Couldn't DAB pages also benefit from feedback, for example if a user sees an inappropriate entry or wants to propose a missing one? Ocaasi ^{t | c} 06:34, 13 January 2012 (UTC)

• Quite possibly; my understanding for the rationale is "disambig articles aren't real articles", but I'll check to see what the precise reasoning was. Okeyes (WMF) (talk) 15:27, 13 January 2012 (UTC)

  Apparently it was requested by enwiki users with the last version of the tool; "many community members found it inappropriate to enable it on DAB pages as they don't host any actual content (or content that users can meaningfully rate". Since we're moving towards "proper" ratings, we're up for re-enabling AFT5 on DAB pages if people are okay with that? Okeyes (WMF) (talk) 18:01, 16 January 2012 (UTC)

  • I think the new AFT would be useful on DAB pages. Star ratings make little sense, but comments about them are welcome and potentially useful. Dcoetzee 09:48, 17 January 2012 (UTC)

Is there any useful information whatsoever that has been generated by previous iterations of the Article Feedback Tool? I'm not aware of any. Does the WMF have any evidence whatsoever that having a comments section is going to change readers into editors? Or is this not, as it appears to me, just one more BLP/incivility nightmare getting ready to be unleashed just because the bureaucracy has decided it'd be swell to have a comments section? Have you all spent any time perusing the comments left at YouTube lately? How about in the comments sections of newspaper websites? Gained any insight from that??? Another crackpot scheme, if you ask me... Carrite (talk) 01:10, 12 January 2012 (UTC) Last edit: Carrite (talk) 01:12, 12 January 2012 (UTC)

• To go through in order:

  • Yes; the limitations on that information are what have got us aimed at developing a version with a free text box.

    Yes; the research on conversions has been fairly widely publicised. Would you like me to provide you with a link?

    No, I don't spend much time perusing youtube comments. Nor would it be helpful for my job, since youtube has a far greater number of comments with a smaller pool of moderators and thus isn't comparable. Okeyes (WMF) (talk) 11:49, 12 January 2012 (UTC)

The entirety of our public pages is based on the idea that people may want to edit, and a proportion of readers can be encouraged to add and improve, not just read. We have seen with donations that if approached thoughtfully more people will engage themselves than if there was no attempt made; the same is likely here as well. Bear in mind that because readers vastly outnumber editors (500m readers, 80-100k editors) a conversion of just 0.01% of readers to good-faith editing following careful testing of new approaches would be a huge increase, so it doesn't need that much to make a really big difference.

Keeping and further engaging those users is a separate important step. There's no point attracting people to engage or raising curiosity about it, or encouraging them to try new things in the wiki-world, if it's too difficult, or met with unhelpful confusion, or attacks. So help, guidance and a good quality of community is the other crucial thing. Measures such as visual editor, terms of use, etc are aimed at that, but the community needs to do its share too.

As for "look at youtube" I'm not sure that's helpful. People on youtube (or indeed, people posting comments to BBC News) don't usually add citations for their statements either, but they often do here. Different sites, different norms. Again, we only need a tiny proportion to make a big difference. Encouraging readers to understand they can improve pages if they spot something, is the most important first step. Like many things in life, asking "how?" is easy once you have the realization something's possible. Realizing the possibility exists in the first place (in this case that you, the reader, could yourself improve what you read), is often the primary problem. Anything raising or encouraging that awareness is going to be crucial to work on. FT2 ^{(Talk | email)} 16:21, 12 January 2012 (UTC)

Despite differences, I think some of the YouTube comment features might be helpful here, like the ability of editors to quickly mark the comments they think are most useful, to point them out to other editors. Dcoetzee 08:22, 17 January 2012 (UTC)

I'm not a big fan of the floating "Improve this page" tab, especially since there isn't any way to get rid of it. Is this part of the final design or something you guys are A/B testing? Kaldari (talk) 09:36, 12 January 2012 (UTC)

That's just for a/b testing; an X button on it is in the works (to be honest, I didn't like it either, but...). Okeyes (WMF) (talk) 11:45, 12 January 2012 (UTC)

I also dislike floating tabs in general, and this one especially. The article feedback tool is, in my opinion, a useless thing but if it is at the bottom of the page it can be safely ignored. This floating tab is a distraction in the peripheral vision and should go. — Preceding unsigned comment added by 69.128.55.230 (talk) 18:03, 15 January 2012 (UTC)

• I think it's fine, but two things: (1) it should be on the left where it will live in the Vector skin sidebar without covering up anything important, and (2) perhaps the wording should be "Comment on this page" or "Send feedback" (since it's clearly copied from Google+ :) Selery (talk) 15:24, 16 January 2012 (UTC)

  Heh. Actually, google isn't the main influence; we've been looking at sites like GetSatisfaction.com. Sticking it on the left is a really great idea - I'll pass it on :). Okeyes (WMF) (talk) 18:01, 16 January 2012 (UTC)

We are pis*ed off this and that internal testing. Give us the results!--Kozuch (talk) 13:34, 16 January 2012 (UTC)

http://toolserver.org/~dartar/aft5/stream/ Have fun! Selery (talk) 15:25, 16 January 2012 (UTC)

Kozuch; we don't have results to give you. They should be tabulated later today, and I'll be able to release them Tuesday evening. We do have a stream above (thanks, Selery, for passing it on!) but it doesn't accurately display the tool's useage (it's an amalgamation of data from 3 different versions, 2 of which we won't use, and is without a spam blacklist or filter). Okeyes (WMF) (talk) 18:03, 16 January 2012 (UTC)

Hm, I see a lot of trash only in the stream...--Kozuch (talk) 11:39, 28 January 2012 (UTC)

I've recently experienced "disappearing votes" in V4 of this tool on a particular page, so I'm also concerned about the transparency of V5 of the tool. There doesn't appear to be any transparency trail for editors to follow, and it certainly prevents the third-party historical observation of "disappearing votes" and other curious phenomenon. This also seems unfair to those who edit pages: those who have all their changes recorded to a historical list of all page changes. As far as future versions of the tool, version 5, where written answers regarding article improvement may be solicited from readers, where are readers' answers going, who will have access to those answers, and what, if anything, will be done to provide transparency to all editors of particular pages? I would suggest that all the answers be pinned to a tab of the article page. However, this might be a redundancy problem, as the talk page is historically used for that purpose. Gzuufy (talk) 20:08, 16 January 2012 (UTC)

As the RfC explains above, all reader comments (minus those caught in the blacklist, abuse filter, or oversighted/hidden due to Issues) will be listed on a feedback page, both per-article and in a centralised location, which absolutely anyone can read (even blocked users, although they can't do anything but read). The "disappearing votes" with AFT4 is deliberate; ratings expire after 30 days. That was necessary because of the nature of the feedback being solicited - for version 5, we're not seeing it as a priority. Okeyes (WMF) (talk) 23:14, 16 January 2012 (UTC)

I have just run across this new feature that has been under discussion and development for a while. I found it on Newt Gingrich in the lower right hand side of the screen. And clicking on it brings up a pop-up version of what is already at the bottom of the page. Perhaps this has been suggested before, but I'd like to at least see the little "X" giving the viewer the option of clicking it away if they don't want to see it. It does tend to be an obstruction of text if the viewer just wants a clear line of sight from left to right, without having to scroll the page upwards just to keep moving past "Improve This Page". Just a thought.Maile66 (talk) 11:35, 20 January 2012 (UTC)

Yup; we've got that feature on the to-do list, although the testing (which is why it's currently live) ends in a couple of weeks, so it may not go up before then. Okeyes (WMF) (talk) 15:33, 20 January 2012 (UTC)

A user who wants to suggest a detailed improvement will likely be frustrated when they click "Improve this page". Once the popup appears, the page can no longer be navigated, and the popup window will likely block the view on one or more of the passages they may want to comment on (Like, "The article states X, but in reality Y", in which you want to be able to see how X is phrased in the article while typing the comment.)  --Lambiam 15:31, 20 January 2012 (UTC)

Yes; this is why the form can be dragged and dropped ;). I'll pass it on to the devs. Okeyes (WMF) (talk) 15:35, 20 January 2012 (UTC)

That such an affordance exists might be clearer if the strip where draggability gets activated is rendered as a visually distinct bar.  --Lambiam 21:02, 20 January 2012 (UTC)

An excellent point; I'll suggest it :). Okeyes (WMF) (talk) 09:55, 23 January 2012 (UTC)

The majority of my reading of Wikipedia is just browsing. Not particularly "looking" for anything. So I cant really answer the question. But without clicking 'Yes' or 'No', no grey questions appear in the text box prompting more general feedback than the first question indicated, so intially it seemed like the text box was purely for elaborating on the question "Did you find what you were looking for?".

Also, the 'Yes' button is already highlited blue before clicking anything. Seems odd. - Glen newell (talk) 05:05, 22 January 2012 (UTC)

So you would advice to add "Not sure" button there? And make either it or no button highlited? --09:52, 23 January 2012 (UTC), Utar (talk)

Glen, that's a really great point :). Tell me; do you edit, regularly, or are you regularly a reader? We've been looking to engage readers in the development more, so do give me a shout if you'd like to participate. Utar, loving the suggestions; I'll email all of it to the devs. Okeyes (WMF) (talk) 09:57, 23 January 2012 (UTC)

It seems now clear. When you can send empty textbox as feedback and when you were able to be not sure in FES why not to allow not sure feedbacks too, eh? --20:54, 23 January 2012 (UTC), Utar (talk)

P.S.: But "not sure feedbacks with empty textboxes" are not cool a bit. At least something should be there. — Preceding unsigned comment added by Utar (talk • contribs) 21:03, 23 January 2012 (UTC)

I'll try and clarify. The question, "Did you find what you were looking for?" seems, ummm, restricting. Just by finding that a wiki page actually exists for some subject, means i've kind of already "found what i'm looking for", so if I also feel something is missing, how do I answer?

The text box question, "Any suggestions for an improvement?" is a more open question. I would prefer that as the main question that appears. It allows people to mention if a section is confusingly written, or if a picture is too small, etc.

Has Option 1 been decided on already?

I've done a few edits, i'd say I am a novice. I thought this was the participating :-s ? Glen newell (talk) 05:08, 29 January 2012 (UTC)

Why add a floating window?! Why make things increasingly complicated? A wiki is simple, text is simple, keep it simple. Loads of windows and pop-ups, and menus, may look nice, may feel as development, but they more often than not are plain simple feature creep. I just gave up editing, looks like I need to give up reading too? Please, keep WP simple. Adding floating nag windows most likely will not help. Tangentially related: http://www.design.wrong.net/?p=17 - Nabla (talk) 18:28, 22 January 2012 (UTC)

• Hear hear. floating elements are annoying and I am disappointed to see them now on wikipedia. I keep my browsers clean of all popups and general annoyances so I first thought some website had got past my defences! I will be looking for a css style or userscript to disable it for myself. For what it's worth I gave this feedback: "Get rid of this Improve this page pop over. It's annoying and not what I expect from wikipedia. Use a static banner at the bottom instead." -84user (talk) 03:45, 23 January 2012 (UTC)

  As it happens, I agree with both of you; I can't stand floating elements, docked to the browser or not ;). This is just a (temporary) test to see if it works; we're also developing a dismissal button pretty soon. Okeyes (WMF) (talk) 09:58, 23 January 2012 (UTC)

One issue that I haven't seen much discussion on is what's going to happen to praise feedback. Although it's true that the most valuable feedback consists of specific questions, issues, or suggestions, I believe praise feedback (even when totally generic like "great article") could be extremely important in motivating editors to spend more time editing and maintaining articles, particularly for articles with only one active editor, who may otherwise feel isolated in their little corner of the wiki. Do others agree with this? If so, can I get some kind of assurance that this feedback won't be hidden or tossed out along with the spam and abuse? Dcoetzee 05:33, 28 January 2012 (UTC)

Agreed. When doing the "hand-coding", I've consistently marked "great, this helped with my homework" or "thank you" as helpful simply because it would be to me. I hope I've been doing this right, and that this is treated as, if not entirely useful for moving forward, constructive. sonia♫ 05:45, 28 January 2012 (UTC)

I agree and as far as I know there are plans to make sure this type of feedback is somehow used. Dougweller (talk) 17:28, 28 January 2012 (UTC)

There are :). So, when it comes to the hand-coding, we've basically been telling people to do their own thing in regards to whether praise is useful/not, because we were measuring what editors think, not what we would like editors to think ;). We're strongly of the opinion that praise is really, really important though, and shouldn't be thrown out. It's as simple as this: On Wikipedia, when you make a bad edit, you get a talkpage message. When you make a good edit, you get...nothing. No message, no positive reinforcement - no evidence that your edit was actually helpful. Anything we can do to try and even the scales in that regard is A Good Thing, because we can't keep going in an environment where people are shouted at if they screw up and ignored if they don't. Okeyes (WMF) (talk) 09:54, 29 January 2012 (UTC)

Clever. Just found one reader complaining about this. Nageh (talk) 14:53, 28 January 2012 (UTC)

Yeah; unfortunately we can't really do much with that :S. We selected a randomised pool of articles to test on, and we could've removed all protected articles (changing the validity of the sample), or we could've kept them. We chose to keep them, but the number keeps altering depending on what pages get protected/unprotected, so it's going to play a merry hell with some of our data. Still, I agree we should be fixing this; I'll drop our devs a note and see if they'd taken this into account. Okeyes (WMF) (talk) 10:00, 29 January 2012 (UTC)

Hello, I have asked the questions below on the village pump, and It was suggested I might receive answers here :

I have been doing a few minor edits in article HMS Titanic and found that the edit link's urls are extremely long with strings of characters including "action=clicktracking", "token=", and "articleFeedbackv5_click_tracking" . Could you help me find answers to the following questions :

• Is there a Help: page or a Wikipedia: page or a Mediawiki page where I could read further information about this?
• Is this "articleFeedbackv5_click_tracking" thing connected with Wikipedia:Article Feedback Tool ?
• Is my small edit understood by the tool as a "feedback" ? Then what are the consequences of a small edit ?
• Why do these long urls show up only on some articles but not on some others ?

Teofilo talk 18:54, 28 January 2012 (UTC)

• So, in order:

1. Not yet; the devs never informed me it'd display like this, so I haven't written one. I'm putting one together as we speak, and should have it up in 24 hours (or possibly a bit longer, depending on when the SF-based folks have time to review the draft text).
2. Yes, it is - long story short, several editors requested (and we thought it was a great idea) that we test the impact AFT5 has on editing activity. If it cuts down on the number of people who edit, because they can give feedback instead, it's A Bad Tool. So, what we did was ran clicktracking software that anonymously stores data on, basically, "where people go" in relation to pages where AFT5 is deployed.
3. It's not understood as feedback.
4. We're only testing the tool on a randomised sample of 0.6% of articles. Okeyes (WMF) (talk) 10:15, 29 January 2012 (UTC)

I am against this attitude of doing first and explaining later. It is another version of "hanging first and trial afterwards". I have decided to start a strike and stop editing Costa Concordia Disaster after finding this morning that the edit link urls on this article have been changed into those personal information collecting URLs. Also it is becoming uncomfortable to edit section 0 of an article. On a normal wiki article, to edit section 0, one copy-pastes the edit link of section 1 and changes "1" into "0". This is no longer possible in a reliable enough way, as the effect of changing the URL becomes obscure. Teofilo talk 17:42, 3 February 2012 (UTC)

My WP skills are not high but my request to WP is simple: please stop doing things that drive away VERY GOOD editors like Teofilo. We need his great work on Costa Concordia Disaster.SteveO1951 (talk) 21:21, 3 February 2012 (UTC)

I disagree with "If it cuts down on the number of people who edit, because they can give feedback instead, it's A Bad Tool". When I have responded using the feedback tool, it is usually because I can "smell" bias, or sense completeness (or not) etc but that does NOT mean that I personally am competent to try to edit the Article. I like being able to give feedback to alert/encourage more knowledgeable people to take a look and edit. Do you want "mere amateurs" (such as I am on many topics) hacking away at articles? I hope not.SteveO1951 (talk) 21:41, 3 February 2012 (UTC)

Oliver, when you say it is enabled on 0.6% of articles, isn't that the proportion that have the feedback tool enabled on them? Surely you need a control sample too... --Tango (talk) 14:08, 4 February 2012 (UTC)

I'll get the details from Dario; I've asked him to drop a note giving more information (he's the research organ-grinder, I'm the monkey). In the meantime, this might help. Teofilo, I'm sorry you feel this way; trust me, had I been informed that this was going to happen before editors brought it to my attention, I would have made sure people were asked in advance. In the future I plan to ask a lot more pointed questions around the research we do. Steve, that is merely a hypothesis that editors - particularly User:WereSpielChequers - asked us to check. We are neither endorsing or dismissing the hypothesis; this research is precisely to establish whether or not it is accurate. And actually, almost all of us are amateurs :P. Okeyes (WMF) (talk) 10:22, 5 February 2012 (UTC)

Why does the feedback box appear on every article? The feedback box is sometimes larger than the article itself, in the case of stubs. I think we should only have those on articles which people are actually watching, and not stubs which someone just created, then walked away. Otherwise, the feedback would not be used in stubs anyway. Just a suggestion. Agent 78787 (talk) 02:19, 29 January 2012 (UTC)

It doesn't; it appears on 0.6 percent of articles. And we could just put it on "big" articles, but that would rather undermine the point of the tool ;). There is going to be a centralised feedback page where all feedback can be seen, so feedback on poorly-watched articles being abandoned shouldn't be a problem. Okeyes (WMF) (talk) 10:17, 29 January 2012 (UTC)

While going through and coding responses, I came across this response (10296, posted on Republican Party presidential primaries, 2012):

Please no more jquery foolishness. I had to use ABP to prevent Jimmys personal appeals from changing the page layout and causing mis-clicks. You have huge comments boxes with stars, an obnoxious AJAX search and now this floating "improve this page" button. Enough already! Stop! Please stop! I've been using and eding WP since 2004, this is annoying. I also run an MW at work, with your CSS circa 2k8, much better. Finally, thest text box in this idiotic dialog doesn't scroll. FAIL. FAIL. FAIL.

—Tom Morris (talk) 21:08, 29 January 2012 (UTC)

Very xzibit. I'll pass it on ;). Okeyes (WMF) (talk) 21:11, 29 January 2012 (UTC)

Tom is very much right, IMO. I have been avoiding WP ever since it became a political party. I still read once in a while (search engines still point here, damn! :-). This floating thingie IS quite annoying (and I've said so a few lines above). I'll add that they are annoying and quite dangerous. The more thingies you add the more likely is that code breaks up on someone's browser. The net is an increasing compatibility hell. Maybe - maybe, BIG maybe - WP's code is up to (some) standard. [whose standard, by the way?] Maybe... But are all browsers in all versions? There are increasingly more sites full of automated crap. Say, Google Images started to show a nice wall of images. Nice? Yes, at first sight. What really happens is that the screen jumps up or down aimlessly *some* of the times I click an image. And they capture some keys, so now I can not search within the page's text hitting "." (dot) as I'm mostly used to. Do you really want to build layers and layers of crap? Yes, crap. You can ask for feedback in simple broadly compatible ways for sure. Why go for the fancy stuff? To look modern? Modern but unusable... makes sense. Hmmm.. does it? - Nabla (talk) 19:43, 4 February 2012 (UTC)

Nabla, your hypothesising isn't particularly helpful - and tom was merely passing feedback on ;p. I don't think anyone is saying the feedback box is unusable except you; certainly not the user Tom is quoting. Indeed, the issue he's reporting is a feature that the devs forgot to include (and are now including), not a bug due to browser compatibility as you seem to think. We have an extensive browser testing regime I can get you details on if you so desire; we don't support all browsers, no, because we find Netscape Navigator and Internet Explorer 3 aren't widely used. We do support all major browsers, and test through a heck of a lot of different versions and OS permutations. Okeyes (WMF) (talk) 10:29, 5 February 2012 (UTC)

«'Jimmys personal appeals [...] changing the page» and «this floating "improve this page" button. Enough already! Stop! Please stop!» refers to «a feature that the devs forgot to include»? Which? I am asking to keep the site simple and 'clean'. If you disagree, fine, but please, save us from silly replies, I am not asking to support IE3. - Nabla (talk) 02:50, 8 February 2012 (UTC)

• Well, when you post about compatibility hell and version support, I answer version compatibility questions ;). As I have said, the floating button is going away. If you read the comment tom is passing on, it says "Finally, thest text box in this idiotic dialog doesn't scroll" - that's the missing feature. Okeyes (WMF) (talk) 03:23, 8 February 2012 (UTC)

Ah! I admit I forgot about the last sentence, and focused on the start. Other than that, it is sad that you reply with contempt. Do you really think anyone is asking for IE3 compatibility? I bet not. So why the reply? You, maybe me too, this conversation, is what is wrong - or maybe the symptom of whatever is wrong - in WP. Your replies are those of an arrogant fool, though probably you are not one. I know I am much less patient and more straight to the point (you may think 'aggressive', maybe you are right...) in here, currently, than I normally am 'outside'. Think about that, not on silly flaoting thingies, and snappy half-funny replies. It is not the technical part that is wrong, it is the interaction. Focus on that. - Nabla (talk) 02:09, 9 February 2012 (UTC)

(talk page stalker) Hold on there. From this section, I don't see contempt at all. To the contrary, I see Okeyes replying to speculation with logics, knowledge, and facts. Please, let's not resort to attacks and whatnot. We're here to building an encyclopedia, and specifically on this page to help develop v5 of the feedback tool, something that from this section alone I can clearly see Okeyes focused on. I mean I understand that a bunch of popups and fanciness can be annoying and break on older browsers on whatnot, but I don't really see that happening here. Do you have any evidence to support your accusations and claims? If so, let's help debug/improve/fix the tool. Jessemv (talk) 02:29, 9 February 2012 (UTC)

• Thanks, Jessemv :). Nabla; we are focusing on interaction. Part of the benefit of this tool is humanising readers - making clear that they're not The Other, a group of at-best ambivalent people who occasionally vandalism. Another part is prompting readers to participate and edit - to help reduce the problem with Wikipedia's conservatism and increasing veteranism and burnout. That's interaction work. After this, we'll be focusing a lot of tools dedicated solely or mostly to interaction. I'm sorry if my comments came off as arrogant. Okeyes (WMF) (talk) 03:14, 9 February 2012 (UTC)

Right, thank the ones that praise you whatever you do. Thank the ones that attack on your behalf. They help. They improve interaction.

Okeyes, Your comments did not look arrogant, they were arrogant. You replied as if you already knew where I am coming from: I am a fool wishing all to stay as it were on the good old days of IE3 (or something like that). You focus on interaction? Do you? I took the time to come and talk to you. I asked to keep things simple. I know, I said this technical option was 'crap'. Not nice. But it happens that you agree it is... crappy... not the best and final option, you'll remove it, you say. So wait! We agree on something, a floating window is not good. But you went over that, ignored the points of contact, and *you* chose the confrontation: "your hypothesising isn't particularly helpful". So, you say, I am making up problems, and I am only disturbing. One comes with good intentions and gets pushed off. Yes, WP needs, badly, to improve interaction.

Allow me to try once more, if I drift too much into off-topic, don't reply, or say so.

I mostly agree with the main problems you point. Tools may help. Go for it. But I think the main problem is 'legal' - or 'social'. WP is simultaneously too fast paced and too slow. Say, go to the administrators noticeboard, and you'll see "broad consensus" forming within a couple of hours, and some enthusiastic (and probably well meaning) folks wanting to act upon that 'consensus'. An example on the other end, go to the Arbitration pages, and see how some problems are recurring, and take years not to be solved in any way (excepts blurbs of short terms consensus).

There is policy 'against' voting. Unfortunately the net effect is that we have voting-by-shouting-and-harrassing, the loudest and with more time to spend group wins. Test on implementing tools to 'vote' in some better way. (askng randomly for editors input as is done in RfC is an interesting idea). That could make a lot of a difference. - Nabla (talk) 13:07, 9 February 2012 (UTC)

• I don't see Jessemv's comments as "attacking"; they made no comment about you, simply tried to difuse you calling me an "arrogant fool" and asking for evidence...followed by you saying that my comments looked arrogant, they were arrogant. Hardly progress :) I didn't choose to have this disagreement; last time I checked it was you insulting me. I am grateful for the points you have brought up, and I will pass them on to the developers. Okeyes (WMF) (talk) 13:34, 9 February 2012 (UTC)

  I agree on the "social" front; unfortunately, handling social issues is done by the community, with assistance from the Community Department. If you're looking for help there, you're best off speaking to Maryana or Steven; that's more their area. Okeyes (WMF) (talk) 13:35, 9 February 2012 (UTC)

Jessemv accuses me of "resort[ing] to attacks and whatnot", and that is an personal attack. Jessemv accuses me of unsupported "accusations and claims", while I have made none, I alerted to a danger that exists elsewhere existing, that is a personal attack. If aggressive users that support you are good users, while others are bad users (me, in this case - and yes, I was aggressive, but mostly *after* you downtalked to me), then you are part of the problem. And you are making it worse, by praising the wrong people.

Would your initial reply be "I am grateful for the points you have brought up, and I will pass them on to the developers" this conversation would have been useful and pleasant for both, instead of a silly argument it turned into. The 'social' front is not handled "by the community", the social front is you, is me. When you talk arrogantly, as you did, you make WP a worse place. When I reply aggressively, as I have done, I make WP a worse place.

It is a pity... and a waste. - Nabla (talk) 16:45, 12 February 2012 (UTC)

I added most of the content to the TRW article, which is now rated "Incomprehensible" by five out of five readers. I noticed this soon after adding vandalism level 1 warnings on two users' talk pages (using my somewhat-inflammatory username — silly me!). This is not a request for action, but I wanted you to know in case the problem grows. Overjive (talk) 18:34, 11 February 2012 (UTC)

Thanks :). Well, we're moving from a star rating to a free text box, which is also somewhat less public - so as we transition from version 4 to version 5, problems should clear up on that front. Okeyes (WMF) (talk) 18:37, 11 February 2012 (UTC)

Somewhat ironically, this may actually be a situation where a proper feedback tool could be helpful. To a lot of people TRW is best remembered as a credit reporting service, something that the article mentions but very obscurely. That's what the IPs are trying to tell you, I think, but they don't know how to get the message across. Looie496 (talk) 19:50, 11 February 2012 (UTC)

True, it's on my to-do list. It's great to have feedback. Plus, you purged the old ratings so I can see the new ones. Thanks! Overjive (talk) 20:58, 11 February 2012 (UTC)

No problem! They purge every 30 days :). Okeyes (WMF) (talk) 02:52, 12 February 2012 (UTC)

Just drawing an edit war to everyone's attention. Sorry if this isn't the right venue. 06:45, 18 February 2012 (UTC) — Preceding unsigned comment added by Dougweller (talk • contribs)

Why do you want to provide more encouragement for readers to become editors. Some readers may become good editors but but they are almost certainly in the minority and there already is adequate encouragement. Getting feedback from readers may be of some value. RHB100 (talk) 19:27, 27 February 2012 (UTC)

Well, for one thing because the editor population is dropping consistently. What is the adequate encouragement? Okeyes (WMF) (talk) 20:04, 27 February 2012 (UTC)

Thank you for answering the question. I'll put any futere questions or comments on this topic on this page. I was not awate of the decline in editor population. The edit links before every section of every article along with the Help page provides passive encouragement to potential editors. One problem with more editors, however, is in my opinion the fact that before many pages there are templates which say something like, "This article's use of external links may not follow Wikipedia's policies or guidelines. Please improve this article by removing excessive or inappropriate external links, and converting useful links where appropriate into footnote references. (January 2012)". These type of templates are in my opinion a waste of time. The editors who place these templates typically do not give a single example or explanation of what they are talking about. If the decline in editor population leasds to fewer of these templates, productivity could go up.

I highly doubt that; a decline in editors would lead to fewer editors to do work. That isn't typically linked to an increase in productivity, unless the presence of templates is something that massively inhibits the desire of editors to contribute? Okeyes (WMF) (talk) 21:25, 27 February 2012 (UTC)

Just a note that AFT seems to have been mentioned at Wikipedia talk:Statistics today. --Demiurge1000 (talk) 23:05, 27 February 2012 (UTC)

Also at Wikipedia talk:WikiProject Statistics, along with various user talk pages and presumably some other places yet to be decided. --Demiurge1000 (talk) 23:12, 27 February 2012 (UTC)

Thanks for bringing this to our attention, Demiurge :). Okeyes (WMF) (talk) 23:26, 27 February 2012 (UTC)

Or that someone couldn't, say, add one? Please? — Isarra (talk) 01:55, 29 February 2012 (UTC)

We're adding one :). There'll be a "hit this close button and it goes away" feature as of next Thursday, assuming nothing breaks or explodes when we deploy MediaWiki 1.19. Okeyes (WMF) (talk) 11:24, 29 February 2012 (UTC)

Does it go away forever? This is important. — Isarra (talk) 17:59, 29 February 2012 (UTC)

Once you disable it, it won't come back. It's important to note that the first deployment of this feature will only work when you're logged in; TL;DR, we pushed for something based on cookies so readers could use it too, and found out it was too complicated to develop easily. We decided we should design a user prefs version first, because it can be deployed quicker, and work on a cookies option later if we even decide to use the current boxes. It's better for it to be ignorable by some users now and all later, than just by all later :). Okeyes (WMF) (talk) 18:24, 29 February 2012 (UTC)

Will it go away on all articles, or only on the one on which you hit the close button? Nikkimaria (talk) 21:15, 1 March 2012 (UTC)

All articles. Okeyes (WMF) (talk) 21:25, 1 March 2012 (UTC)

Okay, I assume it's possible to undo that in some way (say if you clicked it by accident)? It'd be great to allow dismissal on a page basis too, but I don't know how technically feasible that is. Nikkimaria (talk) 04:46, 2 March 2012 (UTC)

Yeah, it is; it'll be under the "preferences" tab, like the AFT4 dismiss button :). Okeyes (WMF) (talk) 13:40, 2 March 2012 (UTC)

Where is the feedback from new users showing up? I cannot find it anywhere. (I can read the aft feed on ~dartar on toolsever, but where is it on Wikipedia?) Addendum: The aft stream is chock full of vandalism, spam, and general trolling. I'd say that at most 1 in 10 were helpful. Reaper Eternal (talk) 15:35, 29 February 2012 (UTC)

Well, there are a couple of things to bear in mind:

1. The data is from all 3 tested versions, mixed in. The info we've got from the best-performing version, as reviewed by a pool of editors, say up to 60 percent is useful.
2. We're going to have abuse and spam filters currently lacking in the raw stream.

As things are now, the feedback isn't visible on enwiki - we're still doing final tests for the feedback page itself. Hopefully the public release will occur in the next few weeks :). Okeyes (WMF) (talk) 16:30, 29 February 2012 (UTC)

I hope you guys will use ClueBot NG's technology, because IMO it's one of the best spam catchers out there. Hopefully as people realize that the stream actually has a use they will start giving informative feedback. Jesse V. (talk) 16:34, 29 February 2012 (UTC)

Indeed! As far as I'm aware, ClueBot is closed source; we can't get at the heuristics :(. I am (hopefully) wrong. Okeyes (WMF) (talk) 16:53, 29 February 2012 (UTC)

Well perhaps they'd build a version for you. They'd be doing Wikipedia a huge favor after all. Jesse V. (talk) 21:27, 1 March 2012 (UTC)

That's true :). I'll give my boss(es) a poke and see if that's the direction they want to go in. One problem is that we're having to run a very tight schedule - the 1.19 release kind of bollocked everything up a bit - and so we may have to deploy without, and then wait to see if we need it to tackle the problem. Okeyes (WMF) (talk) 21:48, 1 March 2012 (UTC)

(edit conflict) OK, I just grabbed the top group of edits, and this is what I see.

Page title	Message posted	Assessment	Blockable via blacklist or edit filter?
Test (assessment)	STRESS IN RELATION TO YOUR EXAMINATION	Not enough information
Lionel Messi	His endorsement or contract with adidas	Not enough information
Cat	np it was about cats the compltery forgot bout demon animnals	Not enough information
Post/Redirect/Get	"(may be prevented by using JavaScript to disable the button after the first click)." -- link to an actual snippet.	Request for content
Memory and aging	choooooooooooooooooooooooooooooooooooooode\n	Vandalism	Y
Memory and aging	choooooooooode\n\\	Vandalism	Y
Bleed from Within (band)	add a band image	Request for content
Nitrogen	my mom	Vandalism	Y
Global Positioning System	gps receiver	Unhelpful
Michael Peto	Lots of great info here, but would like to see more on Peto's life before the war. Also pictures!	Request for content
Mariah Carey	I was looking for her ancestry and found it quickly!	Approval
Transformers: Dark of the Moon	Yes, yes, I found!!	Approval
2011 FIFA Club World Cup	yes I am excited about it	Forum post
Android (operating system)	I was looking for AFTv5, and I found it.	Test
Once Upon a Time (The Twilight Zone)	Not enough farts.	Vandalism
Android (operating system)	why you can' t write more?you can send your reply on my ID on yahoo that is (Redacted)	Personal information	Y
Cat	afehbhjmgebvsjd,mfvhznld.	Test or vandalism	Y
David Wallis Reeves	Testing - I'm the primary editor of this article, so this shouldn't be treated as real feedback.	Test
Naruto Shippūden 3D: The New Era	Need some Gameplay modes and list of characters, stages, attacks, and things like that.	Request for content
Sofia	do not forget yhe Brazilian president Dilma Russeff. Her father is Bulgarian	Request for content
Lionel Messi	HAVE YOU HAVE SEX	Vandalism
President of Israel	SIR YOUR ARTICLES ARE VERY GOOD REFERENCE FOR UNDERSTANDING WHAT I WANT TO KNOW . BUT I AM STILL NOT VERY CLEAR ABOUT THE TERM "SEX SCANDAL SO I WANT TO CLEAR MY DOUBTS ABOUT THIS TOPIC . I WOULD BE VERY THANKFUL TO YOU IF YOU GIVE ME INFORMATION ABOUT THIS TOPIC IN AN EASY LANGUAGE. \nTHANK YOU	Not an article issue
Android (operating system)	I would have expected this article to be more in-depth.	Request for content
Somatosensory system	k dot fuck u	Vandalism
Lomography	Diana F+ article	Not enough information
Cat	Concerning the age of Cats: Is 1 year = to 7 years like dogs ? I already know cats can live over 21 years. Do cats see in color or in black & white ?	Forum post
Stability and Growth Pact	simply stated and informative on a chronological basis, just what I wanted, excellent.	Approval
Sun Dance	Tell when it started\n	Request for content
Law & Order: Special Victims Unit (season 12)	Its not going to be the same please bring back.	Forum post
Date rape drug	What should I do to get tested for proof that I was given date rape drugs	Raised at ANI as a possible emergency
Single (music)	Love you guys!!!! Merry Christmas OR Happy Hanukkah:}	Forum post
Krampus	This is crazy how people believe in this wow.. lol i never ever heard of anything such as this... You find new things everyday thats fer sure.	Forum post
Wikipedia	everything sucks, wikipedia is stupid, and im better than you... im trolling your sorry butt.	Trolling
Tinto de verano	It helped me make a delicious drink	Forum post
Wikipedia	This made me wanna /wrists.	Meaningless
Krampus	Nothing :D	Meaningless
Global warming	Free of politics and representative of the scientific consensus	Approval
Wikipedia	To much information!	Not an article issue
North Korea	fuck you	Vandalism	Y
Wikipedia	nope	Meaningless / test
North Korea	interesting ; amazing	Meaningless
Christmas	It does not say in the bible what day or month Jesus was born. I do not celebrate Christmas and there is no Santa clause My christmas is every day becuase Jesus is with me and watching over me. Every day is a celebration for me. Not just on the 25th of december. Please show me that Jesus was born on the 25th of December with a bible verse.	Forum post
Izmir	thanks for the links those are useful.	Approval
Single (music)	the Lyrics for Rocket Man\n	Not enough information
North Korea	adf	Test
Robert E. Crowe	Would like to know ancestry of robert e crowe	Request for content
Single (music)	this is a great article!!!! it is very informative and very well organized and with great links!! thank you very much!	Approval

It seems that most of the edits that actually addressed an issue in the article were merely requests for more content. A lot of the problem seems to stem from forum posts, which are not blockable via the edit filter. Reaper Eternal (talk) 17:14, 29 February 2012 (UTC)

See my points about the different version testing :). Okeyes (WMF) (talk) 17:21, 29 February 2012 (UTC)

Cluebot NG is not closed source; see User:ClueBot_NG#Source_Code. InverseHypercube (talk) 21:34, 21 March 2012 (UTC)

If that is the case, I'm hoping that it will be used. It seems that it's the most reliable filter out there, and if it could be adapted to fit this tool I think that would work very well. Jesse V. (talk) 14:55, 4 April 2012 (UTC)

Well, Cluebot is an external dependency, and one we'd have to grant the right to revdelete things for it to be useful. That's a WP:BAG decision :). Okeyes (WMF) (talk) 18:41, 12 April 2012 (UTC)

Take a look at http://www.stormfront.org/forum/t870653-2/ -- encouraging members to vote down articles they don't like (well, they don't like much as it's all evidently written by "Liberal Zionist Jews". Dougweller (talk) 19:42, 6 March 2012 (UTC)

Why are you posting that link? Don't feed the trollsracists. Nageh (talk) 20:00, 6 March 2012 (UTC)

When you rate a page, where do the ratings go? Allen (talk) 03:06, 7 March 2012 (UTC)

Look a couple of sections above. Here's a shortcut. Jesse V. (talk) 03:23, 7 March 2012 (UTC)

I'm still confused about this. Can someone elaborate? Thank you. Allen (talk) 03:31, 7 March 2012 (UTC)

They will go to a dedicated feedback page for each article, which we are developing now. Okeyes (WMF) (talk) 04:09, 7 March 2012 (UTC)

I'm surprised we're still testing the floating feedback link. All the feedback I've seen regarding this feature has been vehemently negative. Kaldari (talk) 19:50, 21 March 2012 (UTC)

I've seen similarly negative feedback :(. At the moment it's in the we-want-to-test-the-impact-prominence-has stage; research is research. Unless it turns out that 90 percent of the feedback that comes through is pure awesome or whatever, I'm personally going to argue against using it when we deploy fully. Okeyes (WMF) (talk) 20:03, 21 March 2012 (UTC)

Thanks, Kaldari and Okeyes (WMF), your points are well taken and duly noted. I also heard similar concerns from other community members and we're taking it all in. Rest assured that we are only testing the floating button for research purposes, to measure feedback volume and quality for different link placements. The other placement we are testing this week is a small text link below the article title (option A), as described in our feature requirements page. We want to learn where we can add the most value with the least amount of friction, towards our goal to get more readers to participate productively on Wikipedia. Note that we're only going to test these two different feedback links for a couple weeks, until mid-April, and only on 0.4% of the encyclopedia. We'll share the data we collect with the community, so we can engage in a consensus-oriented discussion about whether any potential increase in quantity and quality of submitted feedback is worth the increased visibility. We think it's useful to have a data-informed conversation about this question, and that's why we're testing it. I for one can't wait to learn more. In the meantime, thanks for helping us think intelligently about this experiment. Fabrice Florin (talk) 07:14, 22 March 2012 (UTC)

• Sorry, I hate this thing. Can't see any use for it. But since it is here: Seems to me the questions you ask (trustworthy etc.) reflect our goals (writers' goals) rather than readers'. New ratings would be for (something like): understandable, well-organized, covers the topic, etc. And as for "Helpful"... can't think of any question less helpful. Good luck with this thing.Ling.Nut3 (talk) 15:01, 22 March 2012 (UTC)

  So, this is the version 5 page; have you seen precisely what we are doing with version 5? We're scrapping the idea of specific categories entirely :). Okeyes (WMF) (talk) 21:08, 22 March 2012 (UTC)

  I should vanish, then. As I said, I see no benefit in this (well, nothing that benefits the encyclopedia as it appears on the page; I do see very clear benefits for various participants). And I have already given the only suggestion I had. So good luck. Ling.Nut3 (talk) 01:13, 23 March 2012 (UTC)

  well Ling.Nut3, if you disagree with the current version, do you also disagree that there should be any means for readers to give feedaback? Is there feedback the could be solicited from readers that you feel would be meaningful? MathewTownsend (talk) 01:25, 23 March 2012 (UTC)

  Call me Ling. I now regret coming here, because my reply will only persuade you that I am either cranky or a crank, or both. :-) So here you go: Article writing is article writing. Some people can do it, and others cannot. Feedback widgets won't change that, and they won't change the way articles are constructed in the genre of English-language academic writing. The software could presumably be upgraded in various ways to make the interface easier to use, but we have paid IT personnel to worry about that, so the feedback widget again is redundant. The only benefits of this widget are 1) It may provide yet one more avenue for recruiting new editors, assuming that peoples' comments are replied to quickly and positively 2) It does give people something to do other than write articles or whack vandals. I would suggest that the Ambassador program is probably a more productive use of time for #1, although I concede that every avenue that adds even one new editor is a valid avenue. As for 2... well... to each his own. Good luck with this project... Ling.Nut3 (talk) 02:35, 23 March 2012 (UTC)

  I'm not Okeyes, but Ling I actually understand your feedback and to the contrary I neither see you as cranky or a crank. :D We want new users to edit, because Wikipedia always needs more help. I'm certainly looking forward to some of the feedback on some articles I'm working on, such as Folding@home. To me this tool will provide quick notes which I can then use to better the article. So it helps existing editors further improve their articles, while still recruiting new editors. I'll see how much time the feedback actually takes, but I think in the long run it will benefit more than it will waste people's time. I like being productive, except when I don't and for that there's Facebook. :D Jesse V. (talk) 05:16, 23 March 2012 (UTC)

I like the new link that appears at the top. Some feedback though; I suggest making the big X smaller :) it looks quite odd - maybe just do it in text, or consider using something like Font Awesome (also make it red, from a UX experience blue is a bit confusing for a "remove" link).

Also, it got me thinking back to ideas I had for sorting out the rather out-of-date theme & this might be a good time to consider fixing the userlinks and title banner as a "floating header" so it stays on the page as you scroll. Then the feedback link will be prominent for the whole article. --Errant ^(chat!) 19:40, 23 March 2012 (UTC)

Interesting ideas all :). I'll pass them back to the devs. To be honest, at the moment we're inundated with bugs, feature requests and a heck of a lot more - and this is only intended to be a test :(. Okeyes (WMF) (talk) 23:23, 23 March 2012 (UTC)

As a researcher, I was quite excited to see the Article Feedback tool and getting into the design of various studies that could use this new qualitative source of assessment about an article's quality. Finally, something more quantifiable, fine-grained (at scale), more finely time-varying, and from a larger range of diverse users than the FA/GA/etc. classifications that are wide categories and only assessed for once in a great while. The new tool only offering qualitative feedback makes this work more difficult. Sure, we can do some sentiment analysis, but it won't be nearly as good as quantitative ratings. I also suspect there will be less feedback because there's a higher barrier to making that contribution- it's a small difference, but in a "long tail." I'd really like to see a box that had both the quantitative and qualitative feedback; a user can submit one or both or neither. A/B/C testing might also be really useful to measure the quantity of feedback from just the qualitative vs. just the quantitative vs. both being displayed. I'm willing to help write up those results into a short paper. WBTtheFROG (talk) 16:41, 26 March 2012 (UTC)

If you are researching quality the new tool provides a much better measure than the old Article Feedback tool. It is sometimes quite disillusioning to see that really bad but specialist topics get rated close to 5 stars while much better but more popular topics get rated more according to "I like it", averaging at somewhere in the middle on the quality scale. So, thanks very much for the new tool, especially if you are a researcher! Nageh (talk) 16:51, 26 March 2012 (UTC)

The new tool provides a more detailed measure of quality that's helpful for evaluating any one particular article, but doing so will now be a manual process subject to the biases of whomever is converting from the qualitative to quantitative feedback. It will be much harder (a lot more time-consuming and less reliably accurate) to get quality ratings for a sample of, say, 5,000 articles, whereas with the quantitative feedback option you can just query the API and have it all in a database with one person in less than a day. You might need to apply some postprocessing to account for regression toward the mean and articles with little or no feedback, but that should still fit easily in the same day, with plenty of time to spare. Does the new tool have any quantitative aspect that I'm missing? WBTtheFROG (talk) 17:32, 26 March 2012 (UTC)

Hey; sorry for the delay in replying :). The only real quantitative aspect is "did find/did not find what they were looking for", but to be honest, enabling research isn't our main goal here. Our goal is to enable readers to submit feedback that editors can use to improve or inform content, and the old layout, as helpful as it was for quantitative research, simply didn't do that very well (for reasons explained on the project page). Okeyes (WMF) (talk) 18:18, 28 March 2012 (UTC)

The feedback tool box on the current version can be lowered by placing all of the elements on one longer line. Also, since categories are part of the article they should be above the feedback tool box. Is there a technical restriction in having to place it above the categories? -- Alan Liefting (talk - contribs) 01:16, 29 March 2012 (UTC)

It'd be messy; categories are considered part of the "page content". We'd have to parse content to work out where it goes each time someone loads a page :S. And when you say the current version - do you mean the one with the free text box, or the old, five-star-rating thing? Okeyes (WMF) (talk) 22:46, 29 March 2012 (UTC)

Hey all. My regular(ish) update on what's been happening with the new Article Feedback Tool.

Hand-coding

As previously mentioned, we're doing a big round of hand-coding to finalise testing :). I've been completedly bowled over by the response: we have 20 editors participating, some old and some new, which is a new record for this activity. Many thanks to everyone who has volunteered so far!

Coding should actively start on Saturday, when I'll be distributing individualised usernames and passwords to everyone. If you haven't spoken to me but would be interested in participating, either drop me a note on my talkpage or email okeyeswikimedia.org. If you have spoken to me, I'm very sorry for the delay :(. There were some toolserver database issues beyond our control (which I think the Signpost discussed) that messed with the tool.

New designs and office hours

Our awesome designers have been making some new logos for the feedback page :) Check out the oversighter view and the monitor view to get complete coverage; all opinions, comments and suggestions are welcome here :).

We've also been working on the Abuse Filter plugin for the tool; this will basically be the same as the existing system, only applied to comments. Because of that, we're obviously going to need slightly different filters, because different things will need to be blocked :). We're holding a special office hours session tomorrow at 22:00 UTC to discuss it. If you're a regex nut, existing abuse filter writer, or simply interested in the feedback tool and have suggestions, please do come along :).

I'm pretty sure that's it; if I've missed anything or you have any additional queries, don't hesitate to contact me! Thanks, Okeyes (WMF) (talk) 14:37, 4 April 2012 (UTC)

Hey all. So, we're going to be implementing a set of Abuse Filters for AFT5, to cut out the worst of the cruft. The provisional list can be found here; if anyone has any ideas about the existing filters, or wants to suggest new ones, please let me know here :). Okeyes (WMF) (talk) 18:45, 12 April 2012 (UTC)

I saw when doing the feedback that quite a bit of it had the word "porn" or another short thing. You could create a filter to not allow things under, say, 15 characters? ~ ⇒TomTomN00 @ 19:43, 12 April 2012 (UTC)

Agreed. It's difficult to get anything constructive under 15 characters. How about they need to have more than three words, and they can't use one (or several) words repeatedly? Jesse V. (talk) 20:38, 12 April 2012 (UTC)

I'll throw these ideas at the regex-meisters :). Okeyes (WMF) (talk) 04:59, 13 April 2012 (UTC)

Disagree. Whats wrong the comments "not helpful" or "excellent article" or "I liked this"> Agreed they are not detail critiques, but I caouldn't say they're not helpful DGG ( talk ) 21:51, 17 April 2012 (UTC)

• Hello everyone, and many thanks to Okeyes (WMF) for bringing us together here to help filter abuse on the article feedback tool! He will be leading the charge for this project, working with some our top filter authors and filter managers to create more filters for article feedback. In the meantime, here is our proposed plan for extending abuse filters for article feedback. Also, here are our recommended feedback guidelines, which list the type of feedback which could be encouraged or discouraged, with the community's support. Our hope is that we can all join forces in coming days to add more filters to this new tool, so that we can leverage the abuse filter to reduce the noise on the feedback pages when we deploy this tool more widely next month. Thanks in advance for any help you can offer towards this goal! Fabrice (BTW, I'm the new product manager for editor engagement at Wikimedia and responsible for the article feedback tool, among other projects.) Fabrice Florin (talk) 23:10, 12 April 2012 (UTC)

However, could filter the WHOLE thing for swear words. I've seen some responses as "this is a whole piece of sh(it)." (I put that word in brackets just in case!) Tomtomn00 (talk • contributions) 18:49, 13 April 2012 (UTC)

Define "whole thing"? What are we not filtering? Okeyes (WMF) (talk) 15:26, 17 April 2012 (UTC)

And why is the comment "this is a piece of shit" useless--some articles are. Even if not quite the case, such a comment can often indicate what the reader thinks is a POV problem with the article--sometimes a genuine POC problem. There's always a choice in studies like this of getting real feedback, or sanitized feedback only for the responses you want to encourage. DGG ( talk ) 21:51, 17 April 2012 (UTC)

I think it's important to distinguish between "developing a filter" and "using a filter". Ideally, what we're going to do is throw around ideas - all filters that could be useful. An example would be blocking swearwords. Then one of you friendly regex writers creates it and deploys it, with it set to take no action but instead to log when the filter has been hit. When we've done that and let it run for a bit, we can go through and easily tell if the filter is catching a lot of useful stuff. Obviously the test is not "does it ever catch useful stuff"; we cannot design for edge cases. Okeyes (WMF) (talk) 22:15, 17 April 2012 (UTC)

Hi. Are you getting much in non-English languages (eg Swedish, Russian or Chinese) and does that bother you? We are fairly tolerant about it in articles, or it comes alongside page blanking in many cases, or gets caught by other more specialised filters, or not caught at all. I also couldn't see any mention of Filter 231. I don't know if that's been mentioned before or gets caught by something else, but there's only so many letters you can string together to form a coherent word. -- zzuuzz ^(talk) 07:39, 14 April 2012 (UTC)

• 231 is a great suggestion! I know you're a filter-writer; are you willing/able to adapt it for AFT5? :). On the foreign language points, I'm honestly not sure. Is there really a realistic way to use regex to catch such things? Okeyes (WMF) (talk) 15:18, 17 April 2012 (UTC)

  I'm sure someone at MediaWiki:Titleblacklist can give you all the details about the available scripts. There is also Filter 346 which is quite effective, or Filter 271. -- zzuuzz ^(talk) 22:53, 17 April 2012 (UTC)

• Thanks everyone for their comments so far :). Just a note that I'll be largely AFK until Monday, at which point I'll respond and get all the info I can from other WMFers :). Okeyes (WMF) (talk) 08:36, 14 April 2012 (UTC)
• I confess I still find the above TLDR. I'd want to clarify, you're looking to implement some dedicated filters here, locally at this wiki. Is that right? If that's the case I think some thought needs to go into resource consumption and sharing filters. We are already running at the condition limit, and there's some quite high priorities for that processing time. -- zzuuzz ^(talk) 22:53, 17 April 2012 (UTC)

  That's what we're looking for :). If you know of a way to tweak the existing filters to run on feedback as well, do say so! By condition limit, you mean the number of things we can run/times we can run things before everything starts getting stupidly slow? Okeyes (WMF) (talk) 23:08, 17 April 2012 (UTC)

  The condition limit is a magic number on the EF home page which starts to stop catching proper vandalism when it goes above 2%, or something. You will see admins occasionally go through and disable a load of harmless filters with few hits or little point to get this number down. -- zzuuzz ^(talk) 07:54, 18 April 2012 (UTC)

• Zzuuzz, filters like AF231 are restricted to (article_namespace == 0), so only the mainspace. Once the SpecialPage for AFTv5 is rolled out, I suppose for those filters we want to also take in the Feedback pages, article_namespace could be set to <= 0? Ideally, we'd modify existing filters, and much the same standards for the mainspace could go for Feedback too. AGK [•] 00:24, 18 April 2012 (UTC)

  I am not at all familiar with AFT. The 'action == feedback' can be filtered it seems. I think there's a few things in the Special namespace, like 'action == block' which you'd have to watch out for. Yes I think adding feedback to the current filters would be preferable to standalone filters, but I can see the benefits for testing. -- zzuuzz ^(talk) 07:54, 18 April 2012 (UTC)

  Can we work with that, then? For testing purposes, we'll set up new filters to see if they're worth having by having an easy way to identify things like number of hits. If they aren't worth having, baleet. If they are, fit them into the existing filter when such a filter exists. Does that work? Okeyes (WMF) (talk) 03:28, 19 April 2012 (UTC)

• I mostly have a stream of questions that that the general detail doesn't address. While we know that most will use it well, there is a hardcore set of abusers (vandals, spammers, ...) out there looking for any and all weaknesses. [I am jaundiced through dealing with wiki-wide abuse on a regular basis.] Can you give feedback about …
  • the new options/filters that will be available At this point I do see that there is anything specific for feedback blocking at mw:Extension:AbuseFilter/RulesFormat.
  • clearly there would need to be some automated and automatic rate-limiting for all feedback as it will become the easiest means to abuse; and probably the quickest, so having as the default the ability to either rate-limit or session-limit any person or IP address to prevent the blunderbuss approach to havoc
  • whether there will be a specific filter that may be able to be used for feedback
  • will urls be allowed? and will these be checked against
    • global m:spam blacklist?
    • local blacklists?
    • title lists
  • How/will they manage wiki code and other code if inserted
  • Will there be an ability for users to flag abuse in feedback?
  • Is there an easy means to mass clean up any abuse in feedback?
  • Will feedback be able to be hidden and/or oversighted?
  • What ability or changes have been made to blocking to enable a quick response where abuse occurs (either from a logged-in account or from an IP address. Can this also be designed to be well-managed when the abuser moves on to the next wiki.
  • What defences are being built in for bot attacks which are becoming more prevalent. We seem to be providing yet opening another avenue for attack with only thin defences, existing known issues to which no clear defence is present and a new means to abuse.
  • I have a concern that as each wiki is individually maintained that the spammers and abusers are operating holistically that a per wiki defence is going to be just adding to the burden of admins, global admins and sysops.
  • Also how these are going to be checkuserable, and will block apply similarly or separately to Feedback

— billinghurst sDrewth (admin, cu, steward some here and/or other places) 08:01, 20 April 2012 (UTC)

• The Abuse Filter will allow for feedback-specific filters, and for feedback to be run through the existing filters if the existing filters are amended slightly to include action="feedback". Rate-limiting is included, as is the global spam blacklist, while all markup and other code is sanitised before input. Oversight and hide functions both exist; I'm confusing about what you mean by "ability or changes to blocking". If you mean "will blocking also block people from creating feedback", yes. Checkuser data will also be available. I'm afraid I don't quite get "what defences are being built in for bot attacks"; the rate-limit, really. Someone could mass-spam feedback at slightly below this limit, but it would be a very slow process and the person easily caught. Okeyes (WMF) (talk) 14:01, 20 April 2012 (UTC)

Possibly naive or outdated question: will these filters and filter changes count against the overall "rule-count" limit per edit in the filter software execution? If so, what other (perhaps mainspace) filters will be prevented from executing? Or will the feedback filters be run on a separate server so as not to impact the mainstream rules? Franamax (talk) 06:55, 24 April 2012 (UTC)

As far as I am aware, they will. We're working to defeat this, either by integrating them with other filters, increasing the rule-count limit (if it doesn't negatively impact on performance noticeably) or possibly another server - it's very early days. Okeyes (WMF) (talk) 16:02, 24 April 2012 (UTC)

So, as mentioned above my several users (thanks, User:zzuuzz!) we've got an issue with the condition limit. This is a limit built into the AbuseFilter extension that, as I understand it, limits the number of conditions that can be used overall; so when Filter X reads "if action=edit, check if added text contains "poop"", that's two conditions - the action=edit, and the checking. Adding a second filter with two conditions brings it up to four total, a third brings it up to six, and so on. AbuseFilter has a hard limit on the number of conditions to prevent ridiculous strain on the servers. This obviously limits what we can do with AFT5's implementation, because if we create an entirely new set of filters, we strain the condition limit substantially (if not exceed it).

Instead, what we're thinking is the following:

1. As an interim measure, AFT5 filters will be integrated with the existing ones if they have identical actions, or not used if they do not. So "if action=edit" would become "if action=edit or action=feedback", implementing both filters with only one extra condition (or no extra conditions. I'm not regex-fluent ;p). This will limit or prevent any strain on the condition filters. Obviously this has issues; we can't fine-tune things as much as we'd like, we have to present the same response each time, and it substantially impedes developing new, AFT5-specific filters.
2. Meanwhile, the Foundation team will be discussing whether or not to tell the contractor who built the Abuse Filter to start work on a plan he has to "group" filters, only executing some in some situations and others in other situations, to defeat the condition limit.

Does this sound good to people? Okeyes (WMF) (talk) 09:42, 30 April 2012 (UTC)

It does. --09:03, 1 May 2012 (UTC), Utar (talk)

Are these comments from readers going to be visible at the bottom of the article? SlimVirgin ^(talk) 22:47, 17 April 2012 (UTC)

No; there's a dedicated Special page for them (or will be) :). Okeyes (WMF) (talk) 23:07, 17 April 2012 (UTC)

Okay, thanks, that's a relief. I had visions of spending weeks on getting an article in shape, only to have someone add at the end of it, "This is a pile of crap." :) SlimVirgin ^(talk) 00:19, 18 April 2012 (UTC)

Thank you a lot for dropping the insanely stupid star ratings. The version 5 feedback tool is exactly what we need as Wikipedia authors. We want to know if an article is useful for the people and why. All we got from the previous versions were some brain-dead numbers. When looking at a single article these numbers are completely useless. What does it mean if for example “Complete” got a rating of 3.1 out of 5 stars in average? Nobody knows. We need reasons. We need to know why people think an article is incomplete. I really, really hope you will keep it as it is now. I beg you as a Wikipedia author. --TMg 17:10, 19 April 2012 (UTC)

Trust me, we have absolutely no plans to go back to the stars, for just that reason. But thank you for the enthusiasm! :). Do you think there is any chance the German community as a whole might find this tool more useful? It is often difficult for us to see which non-English communities will like our software. Okeyes (WMF) (talk) 18:10, 19 April 2012 (UTC)

This depends a lot on how and where the feedback will be stored and how we can retrieve it. Currently we are using the comment pages for this and its always sad to see it end like here or here where the comments are archived by a bot without anybody answering or even reading them. I think this will be the biggest challenge for you to find ways to make the feedback as useful as possible. It's definitely not a good idea to simply dump everything to the comment page like the WikiLove extension does (we all hate this impersonal nonsense, it's like spam). Here is a hint: The German community is working a lot with categories to organize and do quality management (this makes categories almost completely useless for the readers, but that's another story). I think it's important to provide tools that work on the category level, e.g. retrieve lists of most commented or most disliked articles in a category subtree. Also it should be very easy to drop useless comments (in a polite way to not affront the user who wrote the comment) without cluttering our watchlists or discussions. In short, I think it's all about filtering. Oh, and please don't do it like the LiquidThreads extension does, fragmenting discussions over millions of nested subpages nobody understands. --TMg 22:42, 19 April 2012 (UTC)

100% seconded. Nageh (talk) 23:25, 19 April 2012 (UTC)

Is the Foundation still spending resources on Liquid Threads? I agree with TMg, it's dreadful. Dougweller (talk) 06:46, 20 April 2012 (UTC)

The long answer is complex; the short answer is "no, we're wrapping it up, killing it stone dead and moving on". Unless the community suffers a bout of insanity and actively requests it, we have no interest in pushing it on to enwiki. Okeyes (WMF) (talk) 20:00, 23 April 2012 (UTC)

A few comments about the current version of Special:ArticleFeedbackv5:

• Please add a minimum length. Comments like "." or "o" should never be possible.

  On the minimum length front, we really can't add that sort of filtering for legal reasons - but we've got support for the abuse filter, so if you know any regex writers, they can easily stick a script in to do it. Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

  You can not add a minimum length of 2 characters because of legal reasons? What the ...? You are the developers. Add an option and set it to 2 by default. What's the problem with that? --TMg 14:34, 3 June 2012 (UTC)

  Hmn. We could probably build it into the software at its base level; what I meant was "we cannot simply go around writing abuse filters for legal reasons". And base character lengths aren't necessarily a useful filter - if I just want to indicate that the article did/did not contain what I wanted, I'm now forced to leave a comment by such a change. Okeyes (WMF) (talk) 15:26, 4 June 2012 (UTC)

• Is there a filter to block URL spam and typical spam garbage like HTML tags and BBCode? From my experience a single URL should be possible (if someone wants to post a possible reference, for example). Two or more URLs should not be possible.

  Not directly, although it is tied into the spam blacklist. HTML tags and suchlike shouldn't work, however. Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

• I find it a bit confusing to have two possibilities to rate "bad" comments and two possibilities to rate "good" comments. While scrolling the list I often hit "thumb down" and "flag as abuse" the same time. Isn't an "abuse" and a "not helpful comment" more or less the same? I know, you want us to flag spam but we can do the same by simply hitting "thumb down".

  Yeah, this is understandable. We should perhaps disable one set for people with access to the other. Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

  Okay, it's going on the features list for our next pass, which is (I'm afraid) in Autumn. Sorry about that; we've got a pretty severe backlog built up over here. Okeyes (WMF) (talk) 17:32, 5 June 2012 (UTC)

• What does "Feature" mean? Where is it featured and how? Why can everybody unfeature a featured comment? Where are the feature/unfeature comments stored? Why is there a comment field and why is it mandatory optional? Either make it required or remove it.

  Anyone can unfeature it? Since when? And they're stored on the feedback pages, they're just more highly weighted in certain views. The comments field is for a justification, and by definition, if it is mandatory it is required. Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

  Sorry, I mean why is it optional? Either make it required or remove it. --TMg 14:34, 3 June 2012 (UTC)

  Can you explain why? :). There are a lot of elements of Wikipedia that are optional - edit summaries, for example, which are a fairly close analogy. Okeyes (WMF) (talk) 15:24, 4 June 2012 (UTC)

• When viewing the "Details" for a single comment the link to the article is missing.

  Great point; I'll submit a bugzilla request for that. Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

• As said above this tool needs to be aware of the category tree. I consider this very important.

  Why and how? Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

  I'm sorry? This is how we work in the Wikipedia. We use categories in quality management, basically everywhere. For example, I want to see the feedback for all articles in the Category:Digital technology and all subcategories because this is the topic I'm working on. I don't care about the articles in the Category:Literature, for example. Therefore I want to filter by category. Simple. --TMg 14:34, 3 June 2012 (UTC)

  Yeah, I've been an editor for 6 years; I know how categories work. However, I disagree that, at least on enwiki, they're used for quality management - most people ignore them. I'm also not sure what the technical limitations would be on this - the ArticleFeedback table in the database is (to my knowledge) distinct from any notation of categories. I'll find out. Okeyes (WMF) (talk) 15:26, 4 June 2012 (UTC)

  It looks like it would be incredibly difficult technically, and essentially slow everything down to a crawl :(. Okeyes (WMF) (talk) 17:32, 5 June 2012 (UTC)

--TMg 16:01, 1 June 2012 (UTC)

Those are all excellent points! I also noticed that their feedback is not yet tied to their contributions. I'm hoping that will be implemented in later phases, since it hides accountability, reduces transparency, and makes abuse/spam much easier. Jesse V. (talk) 17:24, 1 June 2012 (UTC)

These are all great; I'll forward and look into them :). Okeyes (WMF) (talk) 17:54, 1 June 2012 (UTC)

Hello there Mr. Oliver

I have an iPhone 3GS (version 4.3.1 8G4, model MB715LL, on AT&T but using wifi at the time of the bug). I was reading the article on Chu Hsi in desktop mode and tried to click on the feedback stars at the bottom. When I clicked on the first one, the stats were highlighted in that field. When I clicked on the second area, the stars from the first unhighlighted and only the stars on the second of the four criteria were highlighted. This meant I couldn't add an imput for all four fields and therefore couldn't give feedback.

There is no reason to reply to me on the IP talk page. Thank you for your time.

Wei — Preceding unsigned comment added by 155.33.212.249 (talk) 20:27, 21 April 2012 (UTC)

Well, we're unfortunately not developing desktop software to support mobile hardware at the moment :(. I think it's inevitable that stuff gets broken. I'm afraid I can't promise we'll fix it, particularly since (as you see) we're moving away from having those four fields :). Okeyes (WMF) (talk) 20:02, 23 April 2012 (UTC)

I've been editing Wikipedia articles for a long time. I've noticed the "ratings" feature for awhile and it's easy enough to ignore; but I find the new "Improve this page" popup which I just saw at the bottom right of Water buffalo incident to be quite annoying. On top of this, it simply would not go away when I tried to click on the "close" X button. This is with the latest Firefox on the Mac. This tells me that this feature needs a lot more testing before inflicting it on users ... if you're going to do this at all. (In general, I find popups quite annoying, and I suspect I'm not alone. It's not obvious to me that Wikipedia needs to be doing this, and IMO it could have significant negative effects on the reputation of Wikipedia.) Benwing (talk) 04:59, 26 April 2012 (UTC)

Oh dear! The close button works fine for me; does absolutely nothing happen when you click the X? It's meant to show a popup that tells you how to disable it. As it happens, I agree it's frustrating; I think we will not be using that form. Okeyes (WMF) (talk) 14:13, 26 April 2012 (UTC)

Benwing may have popups blocked by the web broser. --06:43, 2 June 2012 (UTC), Utar (talk)

Certainly possible. Benwing, do you know if that's the case? Okeyes (WMF) (talk) 13:18, 2 June 2012 (UTC)

I'm using Firefox and it does indeed block some popups. Yes, when I saw that popup, clicking on the X did nothing at all. Right now on the water buffalo incident page I don't see the blue popup anymore, just an unobtrusive "Help improve this page" rectangle at the bottom of the text. Are there any pages that still have the Version 5 popup, so I can see what things look like now? Benwing (talk) 00:47, 6 June 2012 (UTC)

Hmm, that page still has the same "Article Feedback 5" category at the bottom. So I guess I'm just not seeing the popups in general any more. No indication that a popup is being blocked (usually when this happens, a pane appears at the top of the window indicating this). Benwing (talk) 00:52, 6 June 2012 (UTC)

I'm not anti this project at all, but it seems to me that getting people willing and able to fix articles is a bigger problem than knowing what is wrong with them, especially for less trafficked articles. I quite often drop comments on talk pages, and most of these comments, I flatter myself, suggest worthwhile improvements or point out valid concerns or inadequacies. However, very often these remain unactioned, and gain no response or interest from editors, even after months or years. How will this feedback tool be any different? 86.181.174.116 (talk) 01:49, 29 April 2012 (UTC)

Hopefully because we're building in a lot of functionality designed around people being able to see lots of feedback posts from multiple articles at once :). Rather than being limited to "people watching this one talkpage", comments will be aggregated across multiple feedback pages to expose posts to more eyes. Okeyes (WMF) (talk) 05:01, 29 April 2012 (UTC)

Was there a forum/RfC/vote/something in which we approved having a blue pop-up box appear on articles? Did I miss the community discussion? I get tons of RfC notifications in my watchlist header, but I don't remember this one.

Why should users have to go into their settings to remove that box? In fact, why should it appear on any registered user's view? Registered users know they can improve articles; that's why they registered. Kafziel ^{Complaint Department: Please take a number} 15:58, 8 May 2012 (UTC)

Well, to go through in order:

1. No, there was no vote on the box. The idea of testing this design and others like it is something that has been publicised on the AFT5 page and features requirements for months, now, and something we have discussed in the numerous (and widely advertised) office hours sessions we've held on the project. The project as a whole has also repeatedly been publicised in the Signpost.
2. Actually, a lot of users do not join to edit - look at the tens of thousands of accounts that never have. This is because either (a) they're not joining to edit at all (which we know to be the case) or (b) editing is very difficult and intimidates them. In the case of (a), this tool offers them an opportunity to contribute and highlights the fact that they can edit. In the case of (b), it offers them a way of contributing that isn't editing.
3. Personally, I find it incredibly annoying as well. We stuck it up to test a hypothesis we had about the relationship between prominence of the feedback form and quality of comments, and will hopefully be taking it down soon. Okeyes (WMF) (talk) 17:19, 8 May 2012 (UTC)

On point two, the OP suggests that the box be disabled by default for registered users, which I agree would be a great idea. After all, its primary purpose is to get input from readers, not people who are already registered users. On point one, this is a problem - while the box was certainly widely advertised, it was imposed. That would seem to be in conflict with our consensus model, no? Nikkimaria (talk) 18:07, 8 May 2012 (UTC)

Ping. Wasn't a rhetorical question. Nikkimaria (talk) 04:57, 14 May 2012 (UTC)

Hi. Yes, I'm surprised that this went through. When I saw this message I had to go find an article that had V5 on it, just so that I could see this blue box! I didn't hear of any vote, but I really don't think its a good idea. But I also agree with the OP about disabling it for registered users and have it appear for everyone else. Jesse V. (talk) 05:53, 14 May 2012 (UTC)

Thanks for reminding me, Nikkimaria! Sorry; I'm technically (very technically) on holiday this week. Alas, reality does not agree with me.

So, the consensus model is used for internal community decisions - technical changes to MediaWiki proper are normally understood to operate using a system of...I guess "informed imposition". We try and make sure everyone knows what is happening and work in changes to address their concerns, but the switch is ultimately at the WMF's end.

That being said, I totally dropped the ball on this one :(. It was discussed during office hours, but office hours != publicly, transparently and on-wiki, at least not to the same degree. I'm hoping to obliterate this ASAP, and have emailed our head of research to find out when he can safely remove it.

I am afraid I'll be on a plane for most of this evening and Tuesday (flying back home from North Carolina) but I'll try to address things as soon as I get back. Okeyes (WMF) (talk) 14:50, 14 May 2012 (UTC)

Well, have a good trip :-). I think the idea about having the tool appear only for unregistered users merits serious consideration. Perhaps you could bring that up with the WMF, once you're home. Nikkimaria (talk) 14:59, 14 May 2012 (UTC)

If we're getting responses from registered users (especially inexperienced registered users), then I don't think we should turn it off by default for those respondents.

I think we're going to have to write a formal information page up about the difference between the WMF changing the WMF's website vs the community changing the community's policies and practices. We have a large number of power users who seem to be suffering under the misapprehension that the volunteer editors are actually in charge of the WMF's website. WhatamIdoing (talk) 16:53, 14 May 2012 (UTC)

That essay seems to be a minority viewpoint. Anyways, given that the tool invites users to edit, and editors are already editing, it's not as helpful for that group. Perhaps limit it to non-autoconfirmed editors? Nikkimaria (talk) 17:48, 14 May 2012 (UTC)

YDOW is supposed to be (very slightly) amusing while identifying a problem with certain attitudes among a minority of users. I am proposing here a different, strictly formal page that outlines the territorial divide, e.g., that the WMF is wholly in charge of software updates, but the editorial community is wholly in charge of the MOS. WhatamIdoing (talk) 21:30, 14 May 2012 (UTC)

Okay, but this wouldn't be the place to propose that. Nikkimaria (talk) 22:49, 14 May 2012 (UTC)

(EC)WhatamIdoing, your logic is flawed: The fact that people are clicking on the boxes doesn't mean they wouldn't be editing anyway if the boxes weren't there. All us poor deluded volunteer editors have managed to muddle along pretty well, lo these many years, somehow figuring out how to edit without the encouragement of any popup boxes. Wikipedia isn't exactly falling to pieces for lack of participation.

On your other point... Yes, the community is generally asked—or at least notified— about highly visible technical changes to formatting. I was asking whether that happened in this case, Okeyes admits that the ball was dropped, which answered my question and (as far as I was concerned) settled the matter. Nobody was making it personal until you showed up. Kafziel ^{Complaint Department: Please take a number} 18:01, 14 May 2012 (UTC)

• There have been a variety of boxes, not all of which look like this one, but my point remains: if people are using it, then we need an actual reason for taking it away. I'd be able to "muddle along pretty well" if you took away the "Edit" tab at the top of the page, but I don't think we should do that, either, because people are actually using it.
• "Notified" is not the same as "in charge". There's a group over at VPT right now who seems to be under the misapprehension that they are "in charge", i.e., that the WMF actually requires permission from the community to make changes to the WMF's website (in that case, to the watchlist software).
• Since you believe that Wikipedia has plenty of participation, I invite you (sincerely) to make a habit of contradicting all of the "we're losing editors" panic posts at the Village Pumps and WP:AN. I could certainly use some help with that perennial task. I hear that Jimbo's talk page attracts quite a lot of worrying about editor retention, too. WhatamIdoing (talk) 21:30, 14 May 2012 (UTC)

• Obviously I meant that we are not just muddling along. We've used that edit tab, with no pop ups whatsoever, to turn this encyclopedia into one of the most successful sites on Earth. So far, so good.
• I don't follow VPT. I'm assuming you're talking about the bolding and the little green stars and all that other stuff that's been going on with the watchlists lately, which I noticed but didn't care about. Stuff like that doesn't bother me because it's not part of our readers' experience. If it annoys the editors but isn't noticeable to readers, then I don't care.
• I rarely pay attention to anything said about losing editors, for the very reason that it is a "perennial task". It's background noise. Navel gazing. It's pointless to argue either side (which is not to say I never try [1]). Editors come and go, people bitch and moan about it, and yet every year we get bigger and better. It's for that very reason that I say the blue pop up box (particularly inasmuch as it requires registered users to opt out, rather than opt in) is a solution in search of a problem. Looks like, so far, you're the only one here who disagrees. Whether in practice or in principle, I can't tell. Kafziel ^{Complaint Department: Please take a number} 22:53, 14 May 2012 (UTC)

• If it helps, I just got replies - we should be turning this off Thursday :). Okeyes (WMF) (talk) 12:45, 15 May 2012 (UTC)

Cool, thanks for the update. By the way, I actually do think it's a good improvement over the old star-based feedback system, I just don't think it should show up by default for autoconfirmed users (who, in my opinion, should be handed a big fat WP:SOFIXIT instead, ha ha). I've signed up for the newsletter so I'll be more informed in the future. Thanks! Kafziel ^{Complaint Department: Please take a number} 18:52, 15 May 2012 (UTC)

More precise targeting is a great idea; I'll look into it! Okeyes (WMF) (talk) 05:48, 16 May 2012 (UTC)