First parts are now done and deployed.

https://community.metabrainz.org/t/gsoc-2017-add-rating-system-in-critiquebrainz/239782/3

I finally got around to reviewing this and I have no problems with bitmap's suggestion ("mostly just that a review can be a rating with no text, which solves the conflict problem") as per our meeting in IRC.

More discussion on IRC: https://chatlogs.metabrainz.org/brainzbot/metabrainz/2017-03-27/?msg=3867086&page=4

I also created a topic for discussion on the MetaBrainz Community website: https://community.metabrainz.org/t/rating-system-in-critiquebrainz/226343.

I agree with having two tickets and I agree starting with something simple makes sense. Only thing about Michael's suggestion I'm not convinced about is the "only the most recent review can be empty" bit, but that's an implementation detail I guess

Don't think Michael's suggestion is that wild actually. We can allow multiple reviews and treat each review as either a score or text on their own or together. This would require to actually have support for multiple reviews in the first place (CB-187). It's not exactly a dependency for this ticket, but it will be for CB-245.

I can see how it would be nice to have a more flexible rating system that is customisable by the person who is actually rating things. This, however, seems like a pretty big project. Perhaps this ticket can be split into two:

1. a simple 5-star rating feature to get started and give people something
2. a more advanced rating system that would allow people to create their scales, etc.

Well, IMO a review rating can be much more interesting than just a "rate this out of 5". We could rate things by different scales (https://www.ece.uic.edu/~fmanriqu/weblet/vg.html as a games example, but I've seen this done as performance quality and recording/sound quality as two separate scales in classical music). We could have a larger scale (rating things out of five makes sense when rating a lot of things quickly, but for more detailed reviews, 1-10 or even 1-100 might make sense).

It makes sense anyway, IMO, that the review + rating are kinda "frozen" together. The user might later decide "actually, the album is kinda boring with repeated listens", but that shouldn't change the rating on the original review. And 13 year old me's five star review of Lord of the Rings shouldn't suddenly get a three star rating because 28 year old me thinks that actually, they could have cut half of the filler - but 28 year old me should be able to write a second, more critical review too, because both are valid points of view that will apply to different people (if I'm a teenager, I'll care about what the teenager thought, not what he's saying as an old dude!).

Alternative (possibly wild) suggestion if you want to allow multiple reviews: All ratings are reviews, optionally with text attached. But only the most recent review can be empty - if you write another review with text, it replaces the most recent empty review.

Something about this is strange to me. It definitely makes sense that a rating and review should be related, but I don't get what this "other" additional rating represents (and how to communicate that in the UI). Is it the user's "current" rating in case they changed their mind since the review?

If we kept a single rating, I would maybe just prevent the user from changing their rating via the API if a review is attached. Tell them they wrote a review, and make them do it from the review editor.

I'm not opposed to this ticket either, but I want to understand what this "other" rating means. How will they be distinguished in the UI? Will both be used for calculating the "overall" rating, or just one?