Meta discussion: reviews and ratings
Let's use this issue to analyse design issues around reviews and ratings, and scope out what work might be required to resolve them.
Goals
Reviews and ratings are intended to play a few different roles:
- A source of data:
- Can be used to provide recommendations (such as in "top rated" sections)
- A means to sort sets of apps, either in the categories or in search results
- A way to judge quality of an app before you try it
- A way for users to get more information about an app before they try it
- A way for developers to interact with users and get feedback
Issues
Current issues with reviews and ratings:
- Low ratings and negative reviews because of issues with particular packaging formats
- Outdated reviews (based on a years old version of an app)
- Crossover between different sources (ie bad ratings for a Fedora package gets attributed to a flathub app)
Examples
Specific examples of where reviews and ratings aren't performing well.
Potential steps to improve the situation
The key thing here is matching actions to issues, and scoping them out to see how much work they'd be.
- Allow developers to flag reviews and have it taken out of consideration in calculating the rating:
- Relates to an old version
- Relates to a bug that has been fixed
- Relates to a packaging issue
- Allow developers to reply to comments
- Establish a feedback channel for developers in the first place - do they even see ratings, currently?
- Show reviews and ratings in Flathub, allow users to add reviews there
- Always show the number of ratings in Software, alongside the average star rating
- Ignore ratings over a certain age
- Encourage users to rate the apps they use, somehow (not sure what the best way to do this would be)
Alternatives to reviews and ratings
- Popularity measures: "Most frequently downloaded..."