Grants talk:IdeaLab/Article ranking
What questions to ask to rank articles?
(There was no way to provide a topic when I came from the Feedback button on the Grants tab... I added the topic manually.)
I do agree with the general idea of recognizing which articles are better, but the current description is too vague to endorse. What does "better" mean? Does it mean articles that get few corrections for inaccuracies or NPV problems? Maybe it means something about articles that are often targeted for vandalism, but which are well protected? Maybe it means articles that are well written? It might be articles that attract a lot of readers? Maybe citations and links back to Wikipedia should be counted, and the best articles are the ones that are most cited? I think it has to be considered as a multidimensional concept, not a simple 5-star system. (Also I dislike the implicit criticism when 1 or 2 stars are given.)
I do see one possible way to implement the multidimensional rating system. Ask readers of the article, perhaps with a little box that allows readers to record their reactions. I actually think it should be done in two questions. The first question would offer two or four dimensions and ask the reader to pick how to rate the article, and the second question would be asking for a rating on that dimension. What would happen with that approach is that it would first expose the aspects of the article that are impressing people and the degree of their reaction, but after Wikipedia has enough data for that dimension, the article would stop offering it, substituting other dimensions to find out what else matters for that article. For example, many people might rate a particular article as quite "interesting", so Wikipedia would stop offering that dimension and instead ask them if they want to rate the article on the "helpful" or "clear" dimensions. (There are problems with this opt-in approach, however.)
In relation to the EPR idea that I submitted (https://meta.wikimedia.org/wiki/Grants:IdeaLab/Multidimensional_EPR_(Earned_Public_Reputation)), I actually think this idea is the other side of that coin, and the dimensions of the articles should be related to matching dimensions of contributors. However the articles are evaluated, the positive evaluations should also be reflected in the EPRs of the contributors who wrote the articles. For favorable evaluations, it might be hard to assign specific credit, but it can be distributed among the contributors in proportion to their contributions to the articles. For unfavorable evaluations, for example if there is a factual error or an NPV violation, it may be easier to focus the responsibility on a specific author who would risk some reputational damage. Shanen (talk) 17:38, 29 July 2018 (UTC)