Should be allowed to translate in both directions
Users learning French, for example, should be allowed to translate French sentences into English as well as English into French.
Or another example, people learning English from Spanish, should be able to translate Spanish into English as well as English into Spanish (only latter is allowed).
Both are important skills that would contribute to learning the language. Especially more relevant as more languages are added, as we'll get more native speakers of different languages involved. Native speakers are more likely to understand the original sentence. Could even compare the two groups of translators for any given sentence and see whether there is a difference in average quality.
Obviously, I'm talking about translating content around the web, and not the lessons, which already do this.
Ultimately it's a mutually beneficial thing. More translations for duolingo, more practice for users. There are bad translations on both sides, that's guaranteed in crowd-sourcing. But if the lessons are teaching you to translate both directions, there's no reason why the web translations should be any different.
However, presumably Duolingo has explored this option. Perhaps they discovered it tended to produce poor quality translations.
I take it you mean in the translations of real material? Because the lessons already have us translating in both directions.
With respect I disagree very strongly with this suggestion. A native speaker has the best feeling for nuance and what sounds most natural in their native language. A learner of another language certainly doesn't have this facility in the target language and even people who have a high degree of fluency won't still necessarily have the 'ear' for natural expression that a native speaker has. As recognition of this I am aware of a number of foreign writers who have a high degree of fluency in English but still employ a native speaker to translate their books into English. Nabokov is a counter-example but his facility with language was at genius level.
Isn't Duolingo already relying on crowdsourcing to take care of bad translations, though? I can't imagine that the quality of the translation pool as a whole would get that much worse, and in some ways it might even get better. Like Shim said, someone translating from their native language is less likely to misunderstand the original sentence, which is a problem I've noticed in a lot of learners' translations. A translation with perfect English grammar but incorrect information is just as bad as one with correct information and bad grammar, but hopefully by mixing the two we'd end up with a "best translation" with neither of these problems. Plus, there are already nonnative speakers using this site--there's nothing stopping someone who speaks English as a second language and wants to learn one of the others signing up. And of course, not all native English speakers have perfect grammar or are great writers. (And we're not exactly translating Nobel-worthy literature, either.) There might also be some way for Duolingo to allow this function and still protect their translation pool, such as keeping translations into nonnative languages from being voted "best translation," or limiting the proportion of sentences we're allowed to translate into our nonnative language.
I'd personally appreciate the ability to do this from a learning point of view. Given the penchant for immediate repetition and simplicity of the sentences, the standard lessons aren't enough practice for the "native -> learnt-language" learning direction. I wouldn't mind if these translations weren't used for clients, but it is incredibly useful for learners. Also, surely more data is always (somehow) better and can be used for something.
I think it would be helpful to have more questions in the lesson focusing on translating from our native language to our target language? Thus forcing learners to recall from memory (not multiple choice or sound recording) the words and their spellings. I find these questions to be on the low end of the distribution. I say this since we get a lot of practice doing the reverse in the "real world translation" section. Is there no place to offer feedback to the duolingo team?
Duolingo wants to monetize the translation that's why they want native speakers to translate sentences to their own language. When learning a language it's very important to practice translating sentences to the foreign language, but Duolingo is mainly about teaching how to understand sentences in the other language, and there is less emphasis on expressing yourself in the foreign language. This is because of the above mentioned business modell, I guess.