Fixing immersion translation ratings
Three things should happen: 1. We should be able to rate/review previous translations 2. We should be able to delete (or mark for deletion) identical translations 3. Document ratings should "fall down" on their own over time (like Reddit)
Let's look at the first two points: 1. We should be able to rate/review previous translations 2. We should be able to delete (or mark for deletion) identical translations
For instance, look at most popular article to be translated into English, "Harry Potter" (from the Spanish Wikipedia page). It's listed as 108/422 done (apparently some sentences are invisible, but that's a separate discussion).
When you click into the article at https://www.duolingo.com/translation/45514bdfd9d3dc8bc10cdf69bd16fadb you'll notice that every sentence has a number of prior translations, and most of them are the same. For instance, the single sentence "Ilustrador" has been translated as "Illustrator" 34 times. Some were people disagreeing with Illustrator/Illustrators (since there are multiple illustrators listed), but many times people edited the translation to say exactly what it said before. This is ridiculous.
- Document ratings should "fall down" on their own over time (like Reddit)
I put in a few documents, translated several sentences, got a few upvotes. The document that I discussed above is the "Harry Potter (the series)" article from the Spanish Wikipedia. I uploaded the "Harry Potter (the character)" article from the Spanish Wikipedia, translated a few sentences, and got many more upvotes.
Obviously the smart thing to do would be to go to a Harry Potter article and "retranslate" every sentence with what's already there in hopes of getting some quick and easy upvotes to raise my rank so that I can more easily rake in the xp later by translating what I really want to translate. This is ridiculous.
The Harry Potter article has well over 3,000 upvotes. It's quite likely that nothing else will ever come close, and even though the article isn't "done", every visible sentence has been retranslated many, many times. The popularity of the article as a whole should degrade over time so that other articles can have a chance to become popular in turn, to prompt more documents to be uploaded for translation, and so that people don't focus on one super popular article for easy upvotes to get xp before they go translate real documents.
Degrading over time to get rid of the partial ones sounds like a very good idea.
I thought that the existing system already refused to accept an identical translation?