Jump to content

Wikipedia talk:Article Feedback Tool/Archive 2

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1Archive 2Archive 3

Office hours

Hey all; brief introduction, I'm Oliver Keyes and I'm the new Community Liason, Product Development, although I usually edit as User:Ironholds. You can find a general description of who I am, what I'm here to do and how to contact me on my user page, of course. One of the things I'm working on at the moment is getting editors to provide input on the Article Feedback Tool, mainly because we plan on blowing it out of the water and replacing it with something that actually provides useful data for editors.

As such, we'll be holding an Office Hours session with myself, Howie Fung and Fabrice Florin to discuss what the Foundation's ideas are and how editors can get involved. This will be in the usual channel at 19:00 UTC this Thursday; if you're interested in reforming the AFT, whether it's because you love it or hate it, I hope to see you there :). Okeyes (WMF) (talk) 22:50, 24 October 2011 (UTC)

it's pointless

My response to the questionnaire:

  • I don't really like the idea of WP's page ratings at all - they're too easily gamed by people with an agenda, or skewed by people with no knowledge of the subject. It's better to rely on other knowledgeable editors, and on talk page comments, to improve articles. Instead of this "Rate this page" box at the bottom with its useless star ratings it would be preferable to ask readers to post comments or questions on the talk page, with a note saying that anyone is invited to post there. I think most casual readers are afraid to post.

Milkunderwood (talk) 02:24, 25 October 2011 (UTC)

General comments

I would say do not throw the tool out with the bathwater yet. This type of feedback is to complement talk page comments. It is a pretty good idea overall, without a good implementation.

The problems I see are:

  • The software has many bugs. It is amazing how such a simple system can have so many bugs. But that is the current implementation. If I had to rewrite this, I could do it in one week, I promise, in one week and get it right with almost no bugs. So just set one good programmer there to rewrite it in a week after a better design has been finalized.
  • It is so easy to game it. But those problems do have solutions. A design is needed.
  • The idea that the system is "weight free" is a mistake. Anyone can login and game it. However, if experienced users are given more weight (an idea whose time is overdue) that problem can be remedied.
  • I suggest taking another look at Wiki-Watch approach. Granted that the developers of that system may have had an agenda, but the idea is independent of that. The approach of assigning weight to what someone says based on what they have done in the past is used there, and can be used as a model.
  • I think the eventual system should combine a Wiki-Watch type approach with this to have a "reliability assessment" system. Wikipedia has no such system and it is really needed.

However, regardless of the design, if the final implementation is going to be as buggy as this one, do not even bother with a design. So:

  • First get just one talented developer with one assistant. That is all you need.
  • Review Wiki-Watch and do some type of new design that assigns weights that would avoid the gaming of the system.
  • Submit the design for review before you start coding.

For the life of me, I do not know why writing such a simple piece of software has to be like pulling teeth. History2007 (talk) 15:14, 26 October 2011 (UTC)

Thanks for your comment :). Note that we're not throwing the idea out, just trying to make it more useful - I would honestly advise attending Office Hours (or I can send you the logs afterwards so you can take a look?) if you want to get an impression of what we're doing and see if you trust the designers doing it. We'll be moving it on-wiki after that; this is just an "introduction to..." session, as it were. Okeyes (WMF) (talk) 15:25, 26 October 2011 (UTC)
Re your other comment - the plan is office hours and the pages we're putting up to show the context. Okeyes (WMF) (talk) 15:28, 26 October 2011 (UTC)
I will try to attend, but it was easier to telegram my ideas beforehand. And design is one thing, implementation is another. So unless you get less buggy software after the fact, all the design is beside the point. Anyway, let us see what happens tomorrow. But think of it this way, look at ClueBot and how one talented developer put that together, singlehandedly. Now if all these social media consultants had been debating it, it would not have happened. So meetings and meetings are one thing, getting it working is another. Anyway, see you later. History2007 (talk) 15:49, 26 October 2011 (UTC)
We don't have any social media consultants that I'm aware of :P. The intention is not to get less buggy software, but software that is simply better at the job. There's a big gap between cluebot (which does a single job) and the AFT, which has to be all things to all people. Okeyes (WMF) (talk) 00:25, 31 October 2011 (UTC)

Problem Average ratings are locked at 5.0 after 28 votes

I have been watching the ratings on an article I wrote on the West Point Cadet Sword and the averaged was between 3.8 and 4.1 with 26 votes now it stuck on 5 stars in all 4 areas. It has been this way for more than a week, and I know that something does not add up to equal that type of average. Can you please tell me what you can correct this. Thank you.Andy2159 (talk) 01:07, 22 November 2011 (UTC) Added infoAndy2159 (talk) 20:32, 22 November 2011 (UTC)Andy2159 (talk) 23:26, 28 November 2011 (UTC)

I've been having a similar problem, but I think it's stuck on whatever ratings you put in for the article yourself. Also for me it's happening for articles that have as low as 3 votes. No way to tell until someone with some sort of programming background decides to drop in...

General statistics lesson

Hey, everybody. I assume you're here because you're interested (+/-) in the Article Feedback Tool. I have read through several comments about the system and I'm seeing some misunderstandings that seems to be causing a lot of confusion. It's basically a misunderstanding or lack of knowledge of surveys/statistics and how they work. I'm not going to be able to teach you much in a short time and if I tried, you might TL;DR so hard that your face would just melt right off.

Here's some terms that I'd like for people to understand with a little explanation from me. If you understand these concepts, you'll be much better suited to discuss this tool. Regardless, I hope you'll all join the office hours session tomorrow at 19:00 GMT/UTC in #wikimedia-office to discuss completely changing the system.OlYeller21Talktome 18:40, 26 October 2011 (UTC)

Collapsing to save space.

Survey

  • A survey is just a way of judging the feelings of a group of people. A statistic is used when you're judging the feelings of a group of people but can't survey every single person in the group.

In statistics, the "judging" part can be really complicated because it can use a wide range of tools and the "judgement" can be debated at great length. The more complicated the "feelings" you're trying to assess, the more problems you run in to. If you just say, "do you like this?", the answer can be a simple yes or no and the interpretation that a person has of that question is only about the word "like". If you say, "do you think the orange of that pumpkin would make a good hair color for my 9 year old's pony?" there's a lot of interpretation meaning that you could get a wide range of answers. The wider the range, the less sure you can be about how that person really feels.

Statistic

  • A statistic is a representation of a large group. In stats lingo, you would survey a "sample" of a larger "population".

For example, if you want to know if every person who buys McDonald's french fries on October 30th, liked their fries, you can't literally ask every single person. It's not feasible (and is called a Census). Instead, you can take what's called a "random sample" of the population. You then create a statistic by analyzing your findings (let's say you find out that 80% liked them). You can then use another method to determine if that statistic if significant or not which is know as "statistical significance".

Population

  • The "population" is the group that you want to get information from.

In our case, we want to know what Wikipedia readers think of the articles they're reading. All Wikipedia readers are then the population or more accurately, all literate, English reading, viewers of Wikipedia articles is our population. This is where it gets tricky because we don't seem to have this population definition hammered out very well. It seems like several people don't want ratings from people who have a strong opinion about the subject but that's a dangerous way to define a population.

Sample

  • The sample is a smaller group of the population (see above). They're meant to represent the larger group because we can't actually survey the whole population in our case.

"Sampling" (or taking a sample) is often taken "randomly" to try and get the best representation of the larger group. A person can't pick because they could consciously or subconsciously pick a bad representation because they've entered themselves as a variable in a larger equation. There are several different ways to take a sample and those methods are generally used for different types of situations and that's generally known as "sampling" (see below).

Sampling (taking a sample)

  • Sampling refers to the ways that a sample (see above) can be taken. Each method of sampling has its own strengths and weaknesses so their use depends on what the situation is.

We can't choose how we sample on Wikipedia because we can't choose which readers we will survey and we can't force a reader we chose to take the survey.

The way a sample is picked can be very controversial and is usually how people shoot holes in a stated statistic. For example, if CNN says, "75% of Americans eat dinner at 4PM in the afternoon", one could ask how they sampled the population (all Americans). CNN would then say, "We randomly sampled 5 Americans at 1PM by calling the AARP's list of members". So their "random sample" was 5 people over 55 who are probably retired (home at 1PM). That statistic is attempting to say that five retirees represent about 330 million Americans. I'm sure you understand at this point that their statistic certainly don't represent "all Americans" (our population) well.

In short, we can't chose our sample meaning our current method of sampling may make our statistic of little to no use.

Statistical significance

  • Statistical significance or significance level is used to express how well a statistic represents the population.

For example, determining the statistical significance of our findings is rather easy. It depenend on things like your population size (number of people that read the article), your sample size (the number of people who use the feedback tool), and how varied your data is (seems to be a big problem for online polling). We will, very easily, be able to calculate significance but the problem is that it still might not be useful data (it may not indicate how good or bad the article actually is) but that gets more into "correlation".

In short, we care about this because in general, the more people that rate an article, the more sure we can be about the feelings of all the people who read the article.


Correlation

  • Correlation refers to the relationship or dependence of two or more things.

For instance, the number of murders committed in the US rises in the summer months. Likewise, ice cream sales rise in the summer months. There is a strong correlation between the rise in ice cream sales and murders in the US. One could easily say, "Hey! Look at that! When people eat ice cream, they start killing each other!" This is a textbook case of how statistics are misinterpreted. There's a cause for both cases that hasn't been discussed; the heat. It's believed that heat contributes to murder rates and I think it's pretty obvious that people eat more ice cream when it's hot outside. When you focus to hard on the stats, you can easily lose sight of the big picture. This is noted with the saying, "Correlation does not imply causation".

This relates to the Article Feedback Tool in that 700,000 "Good" compared to 3 "Bad" votes on the Justin Beiber article may be because the Justin Beiber Official Fan Club did a drive to inflate that number or just because people thought "Good" meant that "Justin Beiber is good" instead of "the article is good". This is important because it means that our goal should and can never be to produce some sort of all telling coefficient or hard way of determining how good or bad an article is in any respect.

In short, even if the feedback tool is perfect, people are not which means that every statistic will need to be individually scrutinized by the article's editors.

Conclusion and discussion

We're not creating an article for a journal and we're not going to be peer reviewed by experts of the field. I'm not saying that we need to be super professional about how we gather data, analyze it, and use it. All I'm saying is that there are some tenants of surveying and statistics and understanding them will help us gather data that's actually useful. Knowing those terms above can at least plug you into a conversation about how to make that happen. OlYeller21Talktome 18:40, 26 October 2011 (UTC)

In case you're looking for my opinions, here they are
  • I think the goal of the tool is perfect. I can't think of a way to possibly improve upon the two points in the goal. This is a great start and means that we can make clear decisions further down the line.
  • The "star" portion won't work. As the sampling method is a "voluntary sample" (sample donated by interested parties) it has an inherent bias. Several online services are moving away from starring systems as users often have a strong bias that comes out in their votes in the form of the highest or lowest ratings. We should change to a thumbs up/down system.
    • This thumbs up/down system summarizes the user's feelings about the article in a very general way. As this is intentionally vague, a comment box could be added that records their comments on a "talk:<article>/feedback" page. It could make them a part of that recording by opening the standard edit screen which may show them how simple editing can be, therefore satisfying points 1 and 2 of our goal.
    • As I mentioned, the information is vague which would require editors to analyze votes over a given period. If there's an influx of votes in one direction, that sets off a flag/filter and the article is checked. No matter what, the data can't be a clear indication of the state of the article so changes in the data being collected will most likely be the most useful way to use the collected data.
    • The thumbs up/down method may be problematic based on its use elsewhere. On YouTube for instance, a thumb vote indicates if you like or don't like what's presented in the video. We don't want that. We'd get 10k thumbs up votes on Justin Beiber and Barack Obama would look like some sort of poll. While this may bring people to Wikipedia and satisfy point 2 of our goal (get people to participate on WP), it will directly contradict point 1 of our goal (getting feedback about the quality of articles). Relating what we want to the YouTube scenario, we want people to rate if the quality of the video is good (if the camera angle is good, if the picture is fuzzy, if the sound quality is good) and NOT the content of the video. We'd have to make sure that the person voting knows the difference between rating the subject of an article and rating the articles itself. That may be very difficult or impossible for any solution we come up with.
    • This doesn't mean that we have to remove the categories of ratings. I think that may solve the problem above. Drop the star system, keep/tune the categories, and bring in a thumbs up/down system. OlYeller21Talktome 19:18, 26 October 2011 (UTC)

I am sorry, but the blanket statement that The "star" portion won't work is not supported except by a personal opinion. Please see multidimensional scaling (that article needs help) as a possible strategy (among others) for quality assessment. But if there is one valid point in your long statement, it lies between the lines. I am getting the feeling that the "facebook generation" may not be ready to rate encyclopedia articles beyond an assessment of LiLo's current pose or Cherlie's current rehab. So maybe the whole thing is a waste of time after all. History2007 (talk) 20:06, 26 October 2011 (UTC)

Guessing you missed the line where I said, "In case you're looking for my opinions, here they are." I'm well aware of the concept of multidimensional scaling and use it almost every day to assess the feelings of target audiences in one of largest industries on the planet. That you think the whole process is damned from the start means that you're probably not going to be helpful in the discussion tomorrow. Sorry my statement was too long. Feel free to ignore it and my opinions if you feel it has no merit. OlYeller21Talktome 02:38, 27 October 2011 (UTC)
Actually your post made me change my position. Once I realized that much of the audience is the facebook crowd and used to that environment, I started to think that the process of getting them to rate an article on statistical analysis is futile. And given that from what we have seen the eventual system will have a few bugs (at least) the effort used should be used elsewhere. So your post will not be ignored. It was eye opening. History2007 (talk) 03:56, 27 October 2011 (UTC)
I regularly ignore the results of the tool as non-representational, although I view them for entertainment. I've seen excellent quality, well-developed and highly informative articles get low ratings with this tool; I've also seen shoddy, biased, and much too brief works get top ratings. I agree that the polled audience size is too small and there is no correction for bias, so the results appear highly questionable. Where it might be useful is if the results are rolled up to a higher level by combining the data across a large numbers of articles. For example, how well do the WikiProject article quality ratings match up with audience perceptions across the article space? What general categories get the highest/lowest ratings for their most critical articles? Where do we (as Wikipedia editors) need to spend more attention on essential articles? (Best to let an experienced statistician perform this type of analysis though.) &c. Regards, RJH (talk) 19:46, 11 December 2011 (UTC)

I might be out of place here and if I am feel free to delete my comment/suggestion. The star system seems to state that everything is fine, it is just a question of 'how many stars' fine. Has anyone thought of a 'minus' and 'plus' graph? From minus 5, to plus 5. Politis (talk) 12:46, 27 October 2011 (UTC)

Hey, Politis. Your input is certainly valuable so no one will be deleting it. Expanding the range from 5 to 10 won't show a "global" change meaning that the average would be about the same. It would show a little more detail but the detail is only usable/useful if there's an adequate number of ratings for an article which may or may not happen depending on the article. OlYeller21Talktome 20:51, 28 October 2011 (UTC)

History section

"Phase 2 began on TBD"

TBD? Surely the date is no longer in flux if events following it happened almost a year ago.

"In the fall of 2011"

Would that be the fall/autumn in the northern or southern hemisphere? Cheers.  Stepho  talk  07:21, 5 January 2012 (UTC)

Problem

I have been having a problem with the Article Feedback Tool for a while and I have not been able to find out why it is not showing up at the bottom of articles on my computer. I checked to see if it was disabled in the advanced options, but it's not disabled. Does anyone know what the problem is here? Thanks. Volcanoguy 16:50, 11 November 2011 (UTC)

Can you tell me an article that you think it should be showing up on, but isn't? --Jorm (WMF) (talk) 18:22, 11 November 2011 (UTC)
Mount Meager. I have been working on this article since August and the Article Feedback Tool hasn't been on any articles I have contributed to. It used to work in all articles then there was a problem with it (it was pretty much just a blank box) then it disappered in all articles. I went on a different computer to see if it works on there and there was no problems. Volcanoguy 11:31, 12 November 2011 (UTC)
I just checked the article logged-in and logged-out, both times the AFT shows correctly. Looks like there's a problem with how your account or how your computer is set up somehow. Did you try logging out but still using the same computer - that would isolate whether the problem is with your login or with your computer/browser? Wittylama 18:08, 15 November 2011 (UTC)
Yes I did check the article while not logged-in and it still dosen't show up. Volcanoguy 01:43, 16 November 2011 (UTC)
Are you running any script-blocking plug-ins (like NoScript)? If you don't allow mediawiki.org to run scripts, then it won't appear. WhatamIdoing (talk) 04:13, 16 November 2011 (UTC)
I don't have NoScript. Could the firewall have something to do with this? Volcanoguy 20:47, 21 November 2011 (UTC)
I am also having the same problem with IE 8 but not with Mozilla. I have a Danish keyboard and my IE is a Danish version with Danish screen messages. Could this be the problem? - Ipigott (talk) 11:22, 17 December 2011 (UTC)
Try updating your internet browser. I did and it is working for me now. Volcanoguy 20:05, 6 January 2012 (UTC)

Dunning–Kruger effect

I just did a quick ctrl+F to see if this was discussed already, and didn't find much here on this. So sorry if this has already been discussed - please feel free to direct me to the discussion if it's here already.

Anyway doesn't the Dunning–Kruger effect at least suggest that the check-box for "I am highly knowledgeable about this topic (optional)" is kind of useless? Does anyone have any feedback regarding if this is a useful thing to know when looking at the data? Actually, I'm not even sure how the data is reported and if this checkbox even does anything.216.31.211.11 (talk) 00:50, 4 January 2012 (UTC)

It flips a bit in the data file. I believe that the question has been dropped in the next version. WhatamIdoing (talk) 23:25, 11 January 2012 (UTC)

Just a reflection

I've often said in various places that although this construct is in principle an excellent idea, it's actual design and implementation are flawed. It's going to create a lot of biased results that will not reflect the true quality/usability of our articles. It appears that suggestions from outside the development team are not being given due consideration, and that the team has the intention of pushing it through come hell or high water, based on a broad WMF suggestion that some kind of external evaluation of articles is required. Discussions seem to be fractured and spread around many departments of the Foundation and its Wikipedias. I have been unable to keep track of it all, and I would dearly like to know what metrics are being prepared for the evaluation the gathered data. --Kudpung กุดผึ้ง (talk) 04:42, 17 July 2011 (UTC)

I second the opinion above. I don't question the good intentions of the WMF people but I am starting to think they are becoming Wikipedia's own worst enemy. "Beaucratic"-based thinking is starting to drive Wikipedia instead of the userbase itself. I don't like it. Jason Quinn (talk) 08:26, 20 July 2011 (UTC)
I think that the difference is that WMF defines "userbase" as including far more than experienced editors, whereas our power users primarily define "userbase" as "editors, especially editors just like me".
WMF says that the WMF user community includes almost five billion users. I'd guess that your user community only counts the tiny fraction of a percent of those users who regularly edit articles. WhatamIdoing (talk) 21:48, 23 July 2011 (UTC)
That "tiny fraction" is whose thinking garnered "five billion" users in the first place. They are not a demographic whose opinion should be overruled by a small cabal who think they know better just because they have a BA degree. (PS Whoever came up with the five billion user statistic is a lunatic. That's close to every human on the planet; so unless babies in third world countries have started using Wikipedia, it's impossible for that number to be correct.) Jason Quinn (talk) 17:42, 3 August 2011 (UTC)
Yes, it's a strange number, isn't it? I assume that they count a human who visits the English Wikipedia, Wiktionary, and Commons as three users, and every person who edits each of those as three editors. (That might well be the standard in the web world.) WhatamIdoing (talk) 17:14, 10 August 2011 (UTC)
I've not seen the number 5 billion; I know we're closer to 500,000,000. Our goal is to hit 1 billion by 2015. But even with 500,000,000 the editor community is fractional.--Jorm (WMF) (talk) 17:27, 10 August 2011 (UTC)
I remember seeing a claim of 4.8 billion, which caught my attention because it seemed so absurdly large. It was written out in digits (4,800,000,000), so perhaps the person simply lost track and pushed the '0' key too many times. WhatamIdoing (talk) 17:48, 12 August 2011 (UTC)
(Very) late addition: I think you're remember 480,000,000, which was the number of unique visitors in our literature for a while. :) Philippe Beaudette, Wikimedia Foundation (talk) 01:02, 9 September 2011 (UTC)
Just my reflection, but I thought Wikipedia was written for the benefit of the entire humanity, if possible with the contribution of the entire humanity, to be useful to the entire humanity. For free. But for close to 10 years since it existed, it hasn't asked humanity what it thinks about it, and has labled around half the internet users as "trolls" or "vandals". I think such a feature is highly late and probably won't change much, I don't see the results being publicly visible, and I think those privilledged to the complete results (including comments) will only consider (and even release) the positive results. I remember reading something similar happening in China. I also don't see any documentation or any published results. --Anime Addict AA (talk) 18:50, 28 January 2012 (UTC)

Ratings disappearing

I've several times seen the ratings from a page suddenly disappear and all the parameters go blank, as though it had never been rated. Does anyone know what causes this? SlimVirgin TALK|CONTRIBS 21:07, 17 August 2011 (UTC)

Does anyone know where I could ask about this? I've left a note on the mediawiki page, but no reply there either. It's not clear who is running this scheme. SlimVirgin TALK|CONTRIBS
Howdy! This has been brought to our attention and we are looking into it. Thanks for the report!--Jorm (WMF) (talk) 08:24, 28 August 2011 (UTC)

On the Sourdough page, there used to be at least 20-40 votes in all four categories, and the rankings were 4 or higher in each category. I'm guessing this was a month or two ago (it seems strange to have to go from memory, when any other page changes are recorded and thus can be investigated transparently by all). As of yesterday or thereabouts, I noted that there are no rankings for three of the categories, and one of the categories "Complete" has 22 votes with 1 (poor) as the result. So, the ratings appear to have been cracked, as I cannot see an intentional way to subtract existing votes, yet that appears to have happened. Gzuufy (talk) 19:45, 16 January 2012 (UTC)

Should I rate the page "as an expert"?

The new "Rate this page" box has me slightly confused over whether I even am an 'expert' on the subject at hand. On the subject I happen to be looking up this morning, Perlin and other noise generation, I find that I did study these in college, other types of noise functions and similar fields are hobbies, and my knowledge comes from external areas like textbooks. However, while I might be able to spot a gross error in descriptions or grammar, English is not my best language and I will often misread something. Or, speaking of other computer programming related subjects, having knowledge of a valid implementation of an algorithm may lead me to read a + as a -, or the other way around, leading me to approve an invalid article because of my reading mistake; when I am supposed to be knowledgeable in the subject matter. Or the page may have trustworthy sources, be complete and well-written, and still have such a minor grammatical mistake. Perhaps a 'How confident are you in this rating' may be needed, or a means to weight the opinions of those who have had revoked edits on the page. I have no real proposed solution yet; but I feel the matter does need to be discussed. exestential crisis detected near 66.207.88.49 (talk) 07:02, 21 July 2011 (UTC)

The question shouldn't even be there. It is confusing and is likely to be misused by many. Plus it doesn't even make sense for most articles. What's does "highly knowledgable" even mean for Justin Beiber, Snickers chocolate bars, disambiguation pages, and so on? Jason Quinn (talk) 07:30, 21 July 2011 (UTC) AMENDED Jason Quinn (talk) 04:24, 25 July 2011 (UTC)
Does Justin eat a lot of Snickers while disambiguating? If so, he's an expert. ;-) Jonathunder (talk) 15:11, 12 August 2011 (UTC)
I agree with this. Being "highly knowledgable" is absolutely subjective in the end and I wouldn't dare using it anywhere, since there's always going to be some possibly vital holes in knowledge in just a couple of days' worth of research. CuleX (Talk | Contribs) 13:02, 15 August 2011 (UTC)
"Highly knowledgeble" on Justin Beiber would probably mean you know him well, you're his friend, his manager or someone in the staff that has acces to direct information from him. I'm pretty sure those are exceptions though, as "knowledgeble" often refers to a field (chemistry (chemistry teacher etc), physics, and so on). Chosing popular culture articles from the more scientific types of articles this tool seems to be intended gives undue weight to one side. Just find a way to remove "highly knowledgeble" from popular culture articles and those where "high knowledge/expert" is subjective. Then again, the "expert" tag can be put on any type of article, so why shouldn't this be? Nobody's asked before who's an expert on popular culture articles when they tagged them with {{expert}}. --Anime Addict AA (talk) 15:53, 30 January 2012 (UTC)

Aesthetic suggestion

I think the box should be center aligned, with an option to collapse it in a similar manner to a navbox. That should give a much cleaner look for those who like this function but dislike how prominent it remains when they aren't using it. —WFC— 23:56, 21 July 2011 (UTC)

Article Feedback Tool should be deleted globally

These things are an eye-sore at Wikipedia. I note many editors and readers are not completing them. Lets call it a worthy cause, that needs to be ended and the templates deleted globally across Wikipedia. Thanks JunoBeach (talk) 15:36, 25 February 2012 (UTC)

Just out of interest, how have you assessed the proportion of readers that have not completed them? Also, what proportion of readers were you expecting to complete them, and why? --Demiurge1000 (talk) 23:02, 27 February 2012 (UTC)

WMF's "rate this page"

After answering a survey on an article, readers are invited to edit as part of WMF's strategy to recruit new editors.

I opened a discussion of the WMF's "rate this page" initiative to recruit editors from readers.

Such recruitment "surveys" are prohibited by the ethical code of public-opinion researchers.

Sincerely,  Kiefer.Wolfowitz 22:56, 27 February 2012 (UTC)

I will respond there. Okeyes (WMF) (talk) 23:26, 27 February 2012 (UTC)

Where do the ratings go?

When you rate a page, where do the ratings go? Allen (talk) 03:04, 7 March 2012 (UTC)

Replied on the WP:AFT5 talkpage :). Okeyes (WMF) (talk) 01:30, 8 March 2012 (UTC)


Missing Ratings

One is serious, the other not:

Why is there no category for "This page has been devalued by someone who is ignorant of the subject and has removed facts relevant to the article."?

and

"The writer of the most recent version of this article is just so damned ignorant of the subject that he/she is just plain full of shit."? -- Davidkevin (talk) 00:56, 14 March 2012 (UTC)

Have you seen WP:AFT5? Okeyes (WMF) (talk) 08:04, 14 March 2012 (UTC)

Accuracy?

I don't see any rating for the accuracy of an article. This would seem to be of prime importance, followed by trustworthiness of sources. I have seen plenty of articles, with loads of citations, which nevertheless contain incorrect information. — Preceding unsigned comment added by 74.92.174.105 (talk) 00:06, 4 April 2012 (UTC)

Banned Editors

I am a principal contributor to the article Metric system. A few weeks ago the article had high ratings after 168 comments. When I checked to day, the ratings had plummeted after just another 3 comments. Is there any way in which I can see why this has happened and in particular whether or not a banned editor had influenced the ratings. The editor in question had been very disruptive on this artcile and on related artciles, has been banned from Wikipedia indefinitely and appears to have made two attempts to disrupt things using WP:sockpuppet accounts. Martinvl (talk) 17:58, 21 April 2012 (UTC)

I'm afraid that's not really possible :(. We're moving away from ratings with WP:AFT5, which is much less gameable. I encourage you to give it a look and provide any comments you have :). Thanks! Okeyes (WMF) (talk) 18:15, 21 April 2012 (UTC)

Annoying Float

Notwithstanding the previous chap's impressive anti-liberal, anti-socialist, anti-leftwing, anti-bot, anti-etcetera rant, I would like to submit to your consideration the possibility that the "Improve this page" float that brought me here might be distracting if not downright annoying and from both usability and courtesy point of views, at a minimum, there should be an option to "close" the float and keep it closed at least for the rest of the user's session, same as the nagbanners that appear at the top during donation season.

For now, the only way of getting rid of the floats for me is blocking Wikipedia.org in NoScript, which causes loss of other functionality.

Ta much, tovarishchi.

P.S.: Better red than dead, hasta la victoria siempre, etc. — Preceding unsigned comment added by 88.16.184.39 (talk) 23:15, 21 April 2012 (UTC)

Thanks for the feedback :). The floating button is just a feature we're experimenting with to see what it does to quality and number of submissions; it is extremely unlikely that it will be made permanent. Thanks! Okeyes (WMF) (talk) 16:23, 23 April 2012 (UTC)

Results

Is the feedback submitted somewhere publicly visible? Kozuch (talk) 16:05, 2 May 2012 (UTC)

Yep; there's going to be a dedicated feedback page attached to each article. We're just finalising it up now, and will be showing it off in a special Office Hours session on Friday the 4th at 18:00 UTC in #wikimedia-office on IRC, if you want to come along and see :). Okeyes (WMF) (talk) 18:02, 2 May 2012 (UTC)
So where is this feedback page located? Is it a sub-page of the article? ~Adjwilley (talk) 20:47, 31 May 2012 (UTC)
Replied on the /Help page :). Okeyes (WMF) (talk) 21:17, 31 May 2012 (UTC)

IMPROVEMENT NEEDED!

Hi, Friends!

The feature "Rate this page" is an excellent idea. However, as it has only four attributes (Trustworthy, Objective, Complete, Well-written) and, into them, only that four evaluation degrees (I'd rather not rewrite them here ...), I wish to recommend you a renewed format about how this EXCELLENT IDEA might be improved very much, by creating: (1) MORE (how many? seven? ten?... what?...) attributes than only that four ones; (2) ten [it seems to be an optimal number] evaluation degrees instead of the only that four ones, so poor and inexpressive. More: ALL OF THEM MUST BE noun-typed, not adjective-typed...
If we desire an effective and confiable evaluation methodology for wikipedia production, then we must improve this feature "Rate this page". That is all, for now. Success!

If possible (since I picked out the project wiki), can answer me by e-mail (beremizcpa@gmail.com)? Beremiz Samir

Thank you!

Beremiz Samir — Preceding unsigned comment added by 200.101.69.132 (talk) 15:11, 7 May 2012 (UTC)

Sure; I'll reply via email :). Okeyes (WMF) (talk) 15:52, 7 May 2012 (UTC)
Sorry... Okeyes, but I'm not yet received any reply by e-mail. You can do this by placing there only the link redirecting to here. Thanks! Beremiz Samir. 189.11.234.28 (talk) 21:17, 12 May 2012 (UTC)
Huh; definitely sent! It should have the subject line "Wikipedia's Article Feedback Tool". Okeyes (WMF) (talk) 23:58, 12 May 2012 (UTC)
I don't understand what happened... Maybe spam detection. Can you resend (beremizcpa@gmail.com) this reply? I would thank you very much. Beremiz Samir. 200.101.69.132 (talk) 22:02, 17 May 2012 (UTC)

Percentage form...

...Percentage form is surely the best form of all for presenting the results of the evaluation for (those so few for now...) attributes. Beremiz Samir. 201.40.34.188 (talk) 15:55, 26 May 2012 (UTC)

We're actually moving away from a five-star ratings system, due to (obvious) limitations that you touched upon. Instead, we'll have a free text box, which you can read about here. Okeyes (WMF) (talk) 21:57, 26 May 2012 (UTC)
  • Well, the Wikipedia: Article Tools Comments / Version 5, as mentioned above, is important, meaningful and useful. Despite, whether or not increase the amount of current attributes (only four ...), and / or increasing the amount of current evaluation degrees (only four ...), both so obvious limitations , FreeTextBox is certainly an advance in meritum causae. This new approach, this new model for the classification of articles in wikipedia scope is certainly a desirable and promising combination (this is a happy marriage ...) between the objective and subjective methods. Anyhow, it should be remembered that the improvement already suggested for the objective part of the method must be implemented, namely, (1) ten attributes, (2) ten degrees of evaluation, and (3) presenting the results in percentage form. The adoption of this new model, combined and corrected, will result in a so better evaluation and rating criterion for articles, much more intuitive, and believable. I wish you success in this improvement! Beremiz Samir. P.S.: No answer came to me yet... 189.10.125.214 (talk) 19:35, 27 May 2012 (UTC)
Yes; given that it wasn't emailing properly, I left it here. It's a (much amended) version of the comment above. And the existing form is going to be entirely replaced with the free text box one, meaning that the adoption of 10 attributes isn't going to be necessary - the use of objective, quantified evaluation methods will be ending. It simply doesn't work. Okeyes (WMF) (talk) 20:11, 27 May 2012 (UTC)
  • Wikimedia project has shown a very remarkable achievement. In this context, English Wikipedia (about four million items!) Is, then, obviously, first, privileged position. This is great! As a source of research that is so valuable today, this encyclopedia deserves very naturally be endowed with a system of high-level assessment. This is what we want and hope will be implemented. Success! Beremiz Samir. 200.101.69.132 (talk) 15:07, 29 May 2012 (UTC)

"Please help improve this article"...?

Hi, wikifriends!
The new approach Please help improve this article You find what you were looking for? — as was shown in the article Jesus Christ — if adopted as a single solution will not solve the problem of meeting a good evaluation system, which certainly deserves this encyclopedia. So — and again — we insist on the adoption of a mixed system [(objective, with 10 attributes, 10 degrees of evaluation and a percentage of results) + (subjective, which may be the model presented in Jesus Christ)]. Please! Improve this rating system! Beremiz Samir. 200.101.69.132 (talk) 21:16, 30 May 2012 (UTC)

The objective model, as stated above, simply doesn't work. It doesn't matter how many attributes you have; the data is not useful to editors. I get that it may be useful to researchers, but the more complex and detailed we make the form the fewer responses we get. Okeyes (WMF) (talk) 01:01, 31 May 2012 (UTC)
  • Hmmm! I understand your point of view. In fact, it reflects your great zeal with the best functionality and significance of the evaluation system to be used in this wikipedia. I confess that in some ways, I agree with his assertions. However, due to precaution (for the reasons set out above), must also disagree at another measure, the mere adoption of only the new model. Hence, it occurs to me the following idea: how about adopting a hybrid model that (1) presents the first "Help Improve...", but in addition, within the scope of the first level of evaluation, this after being opened, (2) provide the reader an objective evaluation, along the lines suggested, as an optional improvement ("if you want to further improve this article ..."), for research purposes, as you have pointed out? How about this? Beremiz Samir. 200.101.69.132 (talk) 15:40, 31 May 2012 (UTC)
    Well, there are two problems with that, one from a research perspective, one from the perspective of what this tool set out to do.
    On the research front, doing it this way (an objective set of criteria following a subjective free-text box) is going to generate very little data. We already know that the free text box is generating fewer responses than the objective criteria; adding an extra hoop for people to jump through will winnow it down even more. The resulting information is not necessarily going to be very representative or very large as a sample.
    More importantly, the initial purpose of this software was twofold; to provide actionable feedback for editors, and to provide an on-ramp for readers to become editors. That's why there is a "call to action" inviting readers to edit after they've submitted feedback. Sticking an extra stage in before this call to action would dramatically reduce the number of people passing through to sign up, undermining one of the core rationales behind this software in the first place.
    Now, I've seen some really interesting things come out of the objective data we already have; User:Protonk has produced some great stuff. When we can fit additional benefits in, it's important (and awesome!) to do so - but not at the cost of compromising the software as a whole, particularly when the actual benefits are nebulous. Okeyes (WMF) (talk) 18:45, 31 May 2012 (UTC)
  • I'm sorry for wikipedia, Okeyes! I tried as I could! I will fight no more for this idea. I have understood your explanations, though not entirely agree with them. I must say this: I do not believe in the good (deserved) results provided by only this criterion "FreeTextBox". On the aspect of generating some feedback as well, this is relative, my friend. That depends on what you want: quality or quantity? It seems that wikiteam did not choose quality. I 'm really sorry. Anyway, I wish the best success to you. And to wikipedia, of course. Never more. Beremiz Samir. 200.101.69.132 (talk) 21:28, 31 May 2012 (UTC)
    No, this is choosing quality. As explained above, objective criteria provide absolutely no actionable information for editors. Okeyes (WMF) (talk) 22:32, 31 May 2012 (UTC)

Page ratings

I wonder whether this page explaining the tool could explain why and when the ratings reset. I watch a stable page with very few minor edits during the last couple of months. A few days ago, it had 49 ratings or so. Now it has 16. Does the tool reset regularly after a certain amount of time, regardless of whether it's been substantially edited? I've seen other editors baffled by this too. Cynwolfe (talk) 22:18, 5 June 2012 (UTC)

How to report vandalism

How do we report obvious vandalism to the ratings? See High five. Can the ratings be reset back to zero? Green Cardamom (talk) 03:19, 6 June 2012 (UTC)

The ratings auto-reset every (I believe) 30 days :). There's no way to manually go in and tweak them. We're developing a new version of the tool which should offer more useful advice, as well as more fine-grained control over feedback; if you've got the time, please take a look at the project :). Okeyes (WMF) (talk) 03:21, 6 June 2012 (UTC)
It used to be 30 edits rather than 30 days. Has that changed? WhatamIdoing (talk) 14:57, 25 June 2012 (UTC)
I'll take your word for it; I remembered 30 was involved, that's about it :P. Okeyes (WMF) (talk) 15:58, 25 June 2012 (UTC)

Use of AFT on other websites

Hi, this is Qzekrom from Uncyclopedia. We're interested in creating our own version of the Article Feedback Tool and would like to know, is the code available under a GNU Public License? 68.173.113.106 (talk) 00:16, 27 June 2012 (UTC)

All our work is; see mw:Extension:ArticleFeedback. Okeyes (WMF) (talk) 03:14, 29 June 2012 (UTC)

Intented use

Reading the feedback on articles that is in the news, almost exclusive all comments are in forum style, discussing the topic and not the article. Is this an intended use, and if not, what plans are there to get the users to use the tool as it is intended to be used. Belorn (talk) 14:33, 20 August 2012 (UTC)

I think this is always going to be the case, unfortunately, but I'm very pleased to see it used so productively on smaller articles (while containing junk, the centralised page is always a useful read). I'm not sure how we'd go around altering users' actions on this front - do you have any ideas? Worst-case scenario, we're building a feature to turn it off on protected pages (no sense allowing abuse through a second mechanism if editing has to be turned off) which may calm things. Okeyes (WMF) (talk) 14:35, 20 August 2012 (UTC)
One way could be to allow forum chatter, but to have user mark their comments as "Impression of the article", "Suggested new content to the article", "Issues with sources", "About the topic" and so on. This might make user voluntary leave the channels about editing free from chatter and put opinion pieces and forum comments in their own space. We would also not need to hunt down and delete pure opinionated/forum style comments, which would lower work load. I also like your suggestion around protected pages. Belorn (talk) 22:39, 20 August 2012 (UTC)

dumb question

This may be a dumb question, but where does the feedback from AFT go? Is it visible somewhere, or are there automatic rating/rankings pages? ⇒SWATJester Son of the Defender 10:10, 1 September 2012 (UTC)

The "out of five stars" kind, or the "helpful/unhelpful, then provide a comment" kind? Okeyes (WMF) (talk) 23:55, 1 September 2012 (UTC)
When you send in your ratings(I don't know if you can before you send it in) there will be an option of seing how it has been rated, called "View page Ratings". --Saederup92 (talk) 07:38, 13 September 2012 (UTC)
Special:ArticleFeedbackv5 for all feedback posts, Special:ArticleFeedbackv5Watchlist for posts on pages in your watchlist and Special:ArticleFeedbackv5/Cat for posts on individual pages (we need more photos of cute cats, it seems). jonkerz ♠talk 00:02, 14 September 2012 (UTC)
Sorry, both the star rating, and the comment/helpful kind. Basically was wondering if there was a way to see the rating of a page before having rated it yourself, or to see the list of top rated pages. ⇒SWATJester Shoot Blues! 06:54, 14 September 2012 (UTC)
I'm still confused, I'm afraid; you're referring to two different bits of software :). The star ratings only appear in the "old" version - to my knowledge, there's no mechanism for doing that, but we're going to be replacing that with the comment/helpful kind in Octoberish anyway. Okeyes (WMF) (talk) 07:19, 15 September 2012 (UTC)

Three tool gripes

Now that I've bothered to click the Feedback from my watched pages link ... what an excellent tool I find. Kudos. Three gripes, though. 1) The details page (e.g. [1]) does not have a nice bold link to the article using the article title as anchor. Instead it has a View article anchor. As wikipedians we're conditioned to look for the article title, not an anchor saying "View article" (a formulation of words which might as well be invisible.) As with the entries on [2], let's have an article titled anchor in the details page, please. 2) When marking some feedback as Resolved, incredibly, wikilinks cannot be used. WTF? See the feedback at this example. 3) It'd be good to be able to edit my Resolved comments after I've first committed them. thanks --Tagishsimon (talk) 02:47, 13 September 2012 (UTC)

Oh, I could go on. Sort by Article would be an invaluable aid to viewing a list of watchlisted article feedback. Is there a features request page? --Tagishsimon (talk) 02:49, 13 September 2012 (UTC)
You should post your suggestions (or a link to this section) on the Wikipedia talk:Article Feedback Tool/Version 5 because you are talking about the AFTv5, the new version being developed. --10:44, 13 September 2012 (UTC), Utar (talk)
Thanks Utar; much appreciated. I'll move the suggestions; consider this thread closed. --Tagishsimon (talk) 11:16, 13 September 2012 (UTC)

Caption

The screenshot should have some kind of caption so that it's clear that this is an image of what the rating thing looks like, not a rating box for this article itself. — Preceding unsigned comment added by Asmeurer (talk • contribs) 23:10, 21 July 2011 (UTC)

"Article Feedback page not enabled for this page."

Right now, Special:ArticleFeedbackv5 shows only "Article Feedback page not enabled for this page." What is broken? Chris857 (talk) 13:53, 21 September 2012 (UTC)

Have you ticked the "Don't show the article feedback...." button in your preferences? Under "appearance". Okeyes (WMF) (talk) 11:24, 9 October 2012 (UTC)

What does WHF mean?

Under in development there is a section saying: "WHF is working on the next version (Version 5) of the tool." What does WHF mean? --Rougieux (talk) 15:09, 21 October 2012 (UTC)

I think it's WMF for Wikimedia Foundation, so I changed it. --Rougieux (talk) 15:23, 21 October 2012 (UTC)
Thanks! Yes, we have a probleh with that character. It's an institutional thing; nobody knows exactly why. Okeyes (WMF) (talk) 20:09, 22 October 2012 (UTC)

Article message box/template

"...where the developers track of the current issues." keep track? Respectfully, Tiyang (talk) 09:54, 22 October 2012 (UTC)

Oops ;p. Where's this? Okeyes (WMF) (talk) 20:09, 22 October 2012 (UTC)
In the Article Message Box (AMB)/Template at the top of the project page. Respectfully, Tiyang (talk) 05:44, 24 October 2012 (UTC)
Oops. It is at the top of this page (talk page). Sorry about that. Respectfully, Tiyang (talk) 05:51, 24 October 2012 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:Article_Feedback_Tool/Archive_2&oldid=1137353788"