User:Purodha/Quality Feedback

From Wikimedia Quality
There seems no way yet for organized delivery of suggestions.
That is why this one is kept on a user page for the time being.
Feel free to amend or comment.

Some websites provide a simple user feedback mechanism about the usefulness of a specific web page. While some mwf projects allow pages to be voted into the Excellent or Recommended Reading class, usually by other editors, there is little attention paid to Good, Acceptable, or Poor ratings; nor are readers actually encouraged to mass-feedback. This suggestion is for casual readers not using talk-pages, who do not want to write text, but likely do have an opinion, or a feeling, whether they have been helped or not.


We can provide a simple multiple choice feedback form to anonymously collect quality ratings for articles. Ratings collectd so far can be shown so as to encourage others to voice their opinion, and can be hidden so as not provoke bias. Rendering, and configurations details should be configurable per project.

What can / should be asked?

There are to areas of interest: How much of an article was read before the readers left? What was his feeling, why did he stop reading?

  1. Proportion read:
    • read only 1st sentence or two.
    • read introduction only.
    • read specific, targetted section.
    • read some arbitrary parts.
    • read much or most.
    • read everything.
  2. Left with feeling:
    • Happy, found everything I was looking for, and even more.
    • Just right, my questions are answered.
    • Acceptable, partially helpful, leaving open questions, could be improved.
    • I did not understand enough.
    • Was actually looking for something else.
    • Don't know.

Possibly, there could be a distinction between presentation quality and usefulness for me - I for one consider it not helpful enough to warrant asking users a third question.

This does not collect to feedback on missing pages, see bugs 5466, 7969 and Extension:DidYouMean for suggestions getting feedback on those.


While I recommend accumulating individual feedbacks into few numeric values per page as quality measurement at the moment they are recieved, several complications need to be carefully thought about:

  1. Ageing.
    • Older feedbacks become less relevant every time, a page is edited.
    • Huge edits may obsolete previous quality figures.
    • Small edits, such as rewording an introduction from highly scientific to everydays language, or adding one-sentence explanations of foreign or academic words, may as well make a big change.
  2. Amount read versus felt happiness.
    • There is a big diffence between: "I read the introduction, I am happy, all questions answered", and: "I read the introduction and left, I understood nothing" as the latter bears some insecurity as to what a reader might have understood had he had the stubbornness to read the entire text.
    • There is quite a difference between: "I read the first sentence, then I knew I was looking for something else, and left", and: "I had to read the entire article so as to find out it was not about the subject matter I had been looking for", and: "I was looking for something else. I saw this right from the starting sentences, but I read the whole article anyways, because it was most interesting to me, and very well done". The latter two will be hard to distinguish from the draft multiple choice selections above.


I think, this kind of quality feedback mechanism is suited for implementation as a MediaWiki Extension.


Other web sites using simple quality feedback:

  • Google Ads has a simple Was this information helpful? Yes / No question.
  • CIOB asks for ranks 1…5 plus optional text.
  • QYPE allows two independent votes about articles: helpful, and well written, plus letting authors award 0 to 5 stars to the places, they write about.