Is there any point in rating a translation after it reaches 100% ?
I've noticed that once a sentence reaches 100% complete and the "best" translation has been selected, no further ratings seem to be considered. In other words, we can be stuck with a truly awful translation and no amount of down voting can get rid of it. Have a look at the translation of "maladie criblée", if you want an example. And if further votes aren't going to make a difference, what's the point in others rating these sentences?
Sorry - I haven't explained this well, have I?
Have a look at the first sentence of "Attention à la maladie criblée". This is 100% complete, and has been for 2 weeks. The final translation selected then was: "Attention to the crippling disease", which is, to put it politely, a sentence with "Many mistakes". None of the available sentences was correct either.
I picked about 5 at random, and fired off some "Suggest edits" in the hope that someone would accept my translation of "Shot hole disease" as correct, since I could not add the correct translation myself. I note that one of these edited sentences now has 5 "Perfects" and 2 "Some Mistakes", whereas the final translation is still there with 2 "Perfects" (which is presumably how it made it to final translation status in the first place), 2 "Some Mistakes" and 2 "Many Mistakes". So why doesn't the sentence with 5 "Perfects" (all awarded since the final translation was chosen) now replace the one with "Many Mistakes"?
As I said above, there seems to be no point in "Suggest edit" or "Rate translations" if it is not going to change anything, and the original final translation sits there like a stubborn donkey, refusing to get out of the way.
In another article, one of my own contributions is now sitting in final translation, 100% completed and contains a typo. Even though I have edited this myself and I have received a couple of suggested edits as well, nothing will shift it. So it seems not only is the new translation model flawed for people who like to do the whole article, it is fundamentally broken too.
(Like this editor - I can't seem to insert paragraph breaks any more! [Although I've since been back and edited it OK.])
There are lots of oddities in the "100% completed" translations.
@ 1km: In the article on "maladie criblée" I couldn't find any "100% completed" translations which could no longer be rated but it is not as easy to do now. Hovering the cursor under the window in which the current best translation appears brings up the possibilities of “suggest edit” and “rate translations”.
I translated one more sentence, and a "current best translation" appeared, which did not contain any mistakes but I didn't think sounded as natural as mine. If it were someone with whom I had contact I would have considered suggesting an edit and I thought of pinching one of his words to edit mine, but couldn't be bothered. I could not view other translations of the same sentence. I miss being able to do that; I learned a lot from alternative good translations.
Is the new system moving towards gathering a certain number of translations then depending on edit suggestions to polish them up?
@Laurenipsum: I have received a couple of messages, in my "stream" ,suggesting that I edit a translation. I edited one. In both cases I was pleased that someone bothered to improve something and I had a friendly exchange of notes about the one I did not change.
I see the same. You (1km) have some good translations and I have to say the ones that bother me the most are the ones from that robot thing. They sit there at 100% and look as if they are from a different sentence. I like trying my hand at translations though I really like seeing what other have written compared to mine. Shift plus return gives you the ability to break up your paragraphs.
I've seen a lot of crap as 100% correct. Maybe "hive" translation just doesn't work, and expert translators really do have expertise. For example, your "shot hole disease" - I'd never heard of it (native English speaker) and sure as heck had never heard of it in French!!! 95% of people would be shooting in the dark, and failing, to translate that. Another example is that people don't seem to understand that definite articles in Spanish and apparently French often should be omitted in English. I could go on, but I won't.
@Persikov - I've never heard of it either. The only difference is that some of us look up specialist terms, rather than provide literal garbage when we don't know the answer. It may have escaped some peoples notice, but the DuoBot is perfectly capable of that without their "help". And with this new system of translation, these people have trashed the place and locked the doors before anyone with a brain has the chance to provide something better than the DuoBot. Sorry for the rant - I'm just getting frustrated waiting for translations to be "fixed".
I don't really know about the algorithm and personally I certainly try to avoid literal garbage. But some things just require some kind of chain or command or hierarchy. Like, many French articles use the present to talk about historical events; apparently that's OK in French. But the final translations jump all over the place, switching from past to present to past to present ad neasuem. There's nowhere that just says "PEOPLE!! USE PAST IN THIS ARTICLE!!" There's not enough sample size to converge on a consistent tense, hence the choppy final product.
Anyway, isn't translating a personal tool for expanding your language abilities? $%·/" the others or the final product if it's not good enough.
I agree with Persikov that there need to be some guiding principles for situations where there's a choice of "correct" translations, such as the tense issue he describes. It's maddening to try and conform, for instance writing in the historical present, only to find they've suddently switched to the more appropriate English past tense. etc etc