More transparency in the feedback process
I'm currently making my way through the last stretch of the French tree and find myself running into frequent errors: missing equivalent translations, poorly chosen primary translations, a few sentences here and there that are just malformed English. No biggie, it's part of the gradual improvement process for this site.
The problem I'm having is with providing feedback itself. The frustration I'm feeling is that it seems like it's never seen by anyone, leaving me in this half-way state of wondering if I should bother and wanting to help improve things, y'know?
I gather the designers and developers have their work queues full, but maybe the future features bucket could include some way of replying with "Yes, we got your feedback", then "We reviewed it and have implemented a change" or "No no, we think this sentence is fine".
Hi! At Duolingo, we have a team of amazing language experts. They spend a lot of time carefully reviewing all the feedback we get. We make sure to correct answers that are in error. If one of your pieces of feedback is accepted you get an email response from us mentioning that. We do review all of the submissions and really appreciate the time you take to send it to us. All this feedback helps make Duolingo even better. Thank you!
The only messages I have ever got were about my translations suggested through "My translation should be accepted" option. I've never got any answers about my multiple choice questions comments or suggestions to change the sentence itself. I reported quite a few sentences that were discussed in comments and found wrong or unnatural by native speakers. I've never got any answers on this. I can't believe 100% of these suggestions were wrong.
It would be cool to receive feedback even about wrong suggestions or those that can't be implemented soon.
Another frustrating thing are several-months-old discussions of some definite problems of the course that receive no comments from the team. I don't mean the team should read all the discussions, but at least the top-voted ones in every skill.
We definitely do read all the feedback. It is a huge amount of work and our language team spends a lot of time looking through these reports. We also read almost every single discussion thread. We jump in when we need to, but really like seeing the community in there discussing.
Is there any way that the community could help prioritize the feedback before it's reviewed by Duolingo staff? I gather that the false-positive reports slow things down a great deal.
Maybe users could help sift through reports for lessons they've completed and either agree or disagree with the feedback, which could help reduce the junk reports staff need to sift through.
Dunno, just spit-balling here.
(Actually, it looks like Olimo made this exact suggestion 8 months back: http://duolingo.com/#/comment/39100)
Would it be possible to provide some kind of stats counter per question, even a relatively hidden tooltip would suffice. Something that shows something like "In the past week, #X% of users got this question right. There have been #Y reports made for this question, #Z of which are currently pending review."
As Olimo says below, sometimes very valid suggestions get absolutely no response. I've been getting sentences like "How to do before it be too late?" and "How worth is a kilogram of vegetables?". If I'm wrong that those sentences are malformed, then we're not speaking the same language. ;)
I completely empathize with VancouverBrad's position regarding feedback. I've done a fair bit of Italian on Duolingo and totally love it. It is such a great feeling of contribution, as well as personal achievement, to have a suggestion accepted and that can act as a positive reward which is highly conducive to learning. My own hit-rate seems to have been about one a week for the last three months :)
However, I hate to thing of the times I have sent in poor suggestions, even if that is part of the learning process, and then when I think that everyone else is doing the same! That can lead to a reluctance to send in a report. If the Duolingo team are truly wading through all this then a simple "No" is all we need by way of feedback. Firstly this would help us remember the issue reported. As it is, the report just seemingly goes off into the ether of cyberspace. Secondly, it would help us to think again and figure it out. Thirdly, it would stop us thinking that we were being ignored which is clearly not the case as you point out.
Presumably, at some point, Duolingo will become so good that reporting potential errors will no longer be required. Until it reaches that point it would be of immense benefit to provide negative feedback to those of us who are struggling with the tension between what we do and do not know and what is perhaps a reportable problem.
I've been getting quite a lot of feedback lately about alternative sentences. I'm personally not very interested in getting reports about the times my suggestions were wrong, but most of all I really don't think Duolingo should spend time on this. I'm sure they would get lots and lots of answers back and it would easily get very inefficient. Would you really accept "No no, we think this sentence is fine". for an answer? Lots of people wouldn't, and that would just create more work.
Interesting comment about the later parts of the French tree being error prone. I have found this in the Italian tree as well. It is very odd that someone fluent in two languages should choose poor translations, particularly ones that are "malformed" English. Since it is odd maybe it is not true. Possibly Duo constructs sentences templates and the program instantiates the templates from a range of words/phrases of the correct grammatical sort. As programs are stupid and languages full of exceptions to structures, it fouls up from time to time. Especially when the sentence type is more complicated. Also to fix such a foul up may not be easy. As I proceed - alas slowly because I am old and have no gift for languages - I come to think that the machine component of the lessons is larger than one might expect (Duo is very coy about this). So in sum: if Duo don't fix a poor example it may be that it is actually very difficult to do so.