Recent changes being tested on the real-world translations
In our continuous road towards making Duolingo better, we've recently been testing a few changes to the real-world translations. Many of you have noticed and posted discussions about them, which we've read very carefully :) Thank you! Here's a rundown of what we're testing and why:
A new color scheme in the document page. There are two reasons for this change: (1) There was an inconsistency with the text colors between the two different translation views. In the "original" view, gray text meant the sentence was untranslated, whereas in the "translation" (side-by-side) view, gray meant the sentence needed to be checked. We decided to unify the colors between the two. (2) We wanted to also display the four different states for each sentence: untranslated, needs to be checked by the community, it is both translated and checked, and it has been done by you. We originally tried to display these four states with different variants of blue, but many people didn't like the ocean of blue, so we now have a scheme that's very similar to how it used to be, but with "untranslated" sentences in light blue text.
"Locking" sentences once enough people have checked them. The idea behind this change is to prevent edit wars, and it works by requiring changes to be approved by the person who submitted the translation that got enough votes to be locked. A few people complained that we were being too aggressive with locking sentences, so we now only lock sentences once there have been at least 3 versions. We're also going to test making it always possible to edit if you're at a higher tier than whoever edited last, but that test is not yet live.
It is important to mention two things about the locking system: (1) we don't just lock down a sentence after 3 edits. The translation must have at least three edits, but it also needs to have enough upvotes from people at high enough tiers. (2) If a locked translation gets enough downvotes, it becomes unlocked again.
Downvotes. There has been concern about the translator tiers and downvotes. In particular, many people think making each downvote count for 5 upvotes in the calculation of your tier is too harsh. We're now testing two new things: (1) getting rid of the downvotes entirely, and (2) making downvotes only count if they come from people in a tier strictly higher than yours (this last condition is not yet live but will be soon).
As always, we're grateful for all your feedback, and we're spending quite a bit of time thinking up new ways to create the best possible translation experience for everyone.
Thanks Luis. I am still not a big fan of the sentence lock-down system. I think we should always be free to change each other's translations if we feel strongly enough about the need for change. Many of the so-called 'edit wars' are in fact quite interesting debates, where you can learn a lot by seeing where different people are coming from and their different global perspectives, which I really enjoy. However, if we have to have the lockdown, a few suggestions that might be useful.
1) If you are the person who uploads the article, you should be able to change it at any time. Many of us take a lot of time and effort to find commons articles that are interesting to us, and to the community. Often these are in our professional area of interest. As a scientist and wildlife photographer, I often upload articles about animals, meteorology and natural hazards, and have the specialised vocabulary to translate them well. I take pride in these articles that I upload and do my best to ensure, with the great help of the Duo-community, that they are the as well translated as can be.
However, I would always welcome anyone to come in and have a go at changing a sentence if they feel strongly about it. I have learned heaps by good translators coming into these articles and providing new insights that I completely missed when I was translating or simply fixing my typos or errors. I don't really want to be the sentence lock down person and getting heaps of emails that someone wants to change something. Just let them change it. If I don't agree, I can just change it back. If they provide a comment why they changed something, I will always provide a comment back if I change it again or I will accept their change. That develops a good debate.
Also I often revise my translations 2-3 times to get them into the best shape, and by reviewing other user comments, changes and inputs. This may take up to a week, as I am very busy. So I don't want to come back to revise an article I uploaded to find that I am blocked from doing it.
So please allow the uploader of document to be able to edit any document they have uploaded at any time, so you get the best possible translation, consistent documents and a track of accountability.
2) If we have to have the lockdown system, I would agree that a higher tier can change at any time, but I would also add that where two people have equal tiers, then they should also be allowed to change each other's at anytime as well.
3) Sorry, but not a big fan of down-votes either, and if they could be dispensed with, I think it would be better. I can see some of the merits of having down-votes, but I think on balance, it serves to demotivate a lot of well-meaning people who are trying their best to learn a language, and can also be used as a weapon in edit wars. Also if you keep down votes, please have a system whereby a person who has given a down vote can reconsider and remove the down vote, if they and the down voted person have reached an understanding or agreement about a translation. If it were a temporary thing that could be removed or undone upon more discussion and reflection, down votes might be more useful. That way the down vote could simply flag an issue, but be removed when the issue was fixed in a mutually agreed manner. Often the best translations of difficult passages involve the input of 2-3 people, with all of us learning a lot in the process.
Anyway just my thoughts as a long-term Duolingo user and translator. Thanks for your consideration!
I think you are confusing debate, "To engage in a formal discussion or argument", with coercion. In a debate you argue about points, and eventually settle on an agreement, or part ways. In immersion you don't argue points, you enforce(submit an edit) whatever you wish, and put a comment if you so desire, even if it is completely irrelevant.
So until duolingo has a better way to discuss an article, it will not be a discussion, it will be just people forcing their ideas on others, who may dispute it and reject it firmly by forcing once again their view, leading to what we affectionately call an edit-war.
My experience on Duolingo which has involved about 100+ full translations across all 6 languages has been incredibly opposite to your assertion. Many times a final translation has in fact been a complete compromise between myself and 2-3 other translators. Jmcliste, the prolific French translator and I have had some excellent debates and discussions full of wit and humour with which I have learned a lot from his excellent skills. ThoughtfulDiva has recently been coming into the Italian articles I upload and has been wonderful in helping me get some better nuance and clarity in more difficult or obtuse passages. There are many of these examples. I regularly send thank you notes to people on their stream to thank them for their help. Other users kindly send me nice notes when I help them as well. In all my time on Duolingo, I can only think of a few times when I have been in a true unhelpful debate or so-called edit war. Sure they happen once in a while. The 100+ positive learning experiences far outweigh the few troublesome ones, and no system is 100% perfect anyway. Plus, I always feel that the Duo Staff work hard to improve the program constantly, and take user feedback and issues seriously, and I find that very encouraging.
I think the idea of sentence lockdown is horrendous though, it completely undermines my faith in the whole system. No translation can every be definitively correct, and it isn't worth punishing edit-warriors this way if it also means cheating people out of their chance to offer their interpretation. As a preventative strategy, it is about as subtle and fair as indiscriminate carpet bombing. "Well, we took out the women's refuge, the kindergarten and the hospital, but hey at least we got our guy!". You can't punish the whole class because one kid won't behave.
I tend to think of a translation in the same way I think of cheese, it's only worth eating after it has had time to mature. Sometimes fresh eyes are better than tired eyes. Sometimes even the best of translators can rush through, and not do the most ideal justice to the material. A good translation does require a certain amount of care, background knowledge and fluency, none of which is offered by this horrible new arbitrary rule.
For the sake of repeating my opinion over and over again, what is really needed is Communication! None of this hit and run down voting or silent editing should be allowed. If you are going to change an existing translation that is perfectly good, being courteous and writing the reason it needed to be changed for everyone else should be a REQUIREMENT, so that the truth can be learned by all. Trial by fire and all that. Those that can't back up their reasoning won't be able to be stubborn anymore. Even if this special requirement was reserved for sentences that have had more than 3 edits, it is still a million times better than this horrible indiscriminate new system.
Do you really want someone to spell out in words of one syllable 'you can't spell/type/construct a proper sentence' or 'you are a malaprop' on the occasions when your typing goes to pot? I don't. If someone had a legitimate reason for changing something it's usually blindingly obvious when you look at their revision - and if they didn't and were just substituting a synonym to accumulate points, that is obvious too.
Well...yes I do. The whole point is to make it tedious to edit the document for no good reason. I don't really care if a comment makes me look stupid, although I doubt everyone would be so rude like that. Even if the comment is two words like 'fix typo', it doesn't matter. The point is to log the motivation for every change. Then you can look at someone's history and say "well, he seems to do nothing but go through every article with a thesaurus and change everything", and theoretically do something about it.
When you look at version control systems for software, it is pretty much a sacred law that every change to an existing code base is backed up by an explanation. It actually works very effectively and can be very helpful tracing a problem, or dissecting the history of an individual file.
It's a purely psychological system. If you have to summarise or explain your changes, you're forced to confront the question of whether you have done a good job or not. It's subtle, which is why I am spelling it out :)
@BarbaraMorris, sorry I can't reply to your comment directly, but I don't think I agree with your position. I don't think anyone who finds communication and team spirit difficult should have authority on the translations, any more than I think people with shaky hands should become doctors, people with poor eyesight should try and be snooker champions or people with bad manners should be customer service representatives.
Besides that, you could imagine it would be easy to have simple canned multiple choice options for editing, with 'other' as an option for anyone who wants to be verbose with their reasoning. Combine that with my other suggestion, of only interrogating people on sentences that have had a lot of different translations, and the whole thing would be barely noticeable. Like speed bumps around a pre-school.
You have to stop people who are blitzing through without using their common sense, even at the expense of losing the interest of a few users. Quality will always be better than quantity in the long run, consider how bad it would be if Duo got a reputation as a translation service that gives results scarcely better than 'Engrish'. Once a reputation is tarnished, it is difficult or impossible to salvage it.
No it wasn't a direct answer to what you said, I am just citing it as the reason I am arguing for it. Currently the system makes it quite easy to get into bullheaded fights over trivial things, but fighting only happens when two parties don't understand each others position fully.
I can only really explain it like this: not long ago, my good friend slapped me round the head out of nowhere while we were having a conversation. Normally I would find that rather irritating, but it turns out I hadn't noticed a horsefly was about to take a nice bite out of my face, thus ruining the rest of my day.
Of course if he hadn't apologised and explained himself right away, I would have thought "What the hell, are you completely mental", and made a resolution to put his toothbrush in something unpleasant. Instead I was glad my friend hit me in the face, and I was glad he explained himself. What Duo has right now is a situation where everyone is walking around hitting each other in the face, and no one is justifying the reason. Hence the friction.
I don't agree that we should be forced or even recommended to put a comment for everything.
Making it tedious to edit for no good reason also makes it tedious to edit for good reason. Those who are gaming the system will just type some random junk, "askasdad", and those who are translating in good faith will waste their time explaining the obvious, "punctuation", "spelling" ec.
Replying to chilvence's comment beginning "sorry I can't reply to your comment directly".
I don't understand the part of your comment where you say you don't think "anyone who finds communication and team spirit difficult should ahve authority on the translations". Is that in response to something I said? I don't think I suggested that anyone should have authority on the translations; for sure I don't think anyone should have authority except duolingo.
I just don't think putting an enforced comment in place is going to prevent anything. People already have to click a couple of times to make a change; what's an extra click to select a reason?
I agree that duolingo should make every effort to ensure they get the best possible translations. But I think they are making every effort, whether or not we agree with all the details.
Half the time people use "Sounds better" and "So-and-so is done in such-and-such a way in English" which, in my book, are not proper justifications for making any type of change. Realistically, then, the explanations would have to be filtered too if we were looking for purely linguistic reasons for making edits. The fact is that not every single edit can be justified in the same way to every user, and way too many opportunities exist for edits that are stylistic in nature and not essential to the meaning of the translation to be made. Examples include differences in British and American spellings, punctuation usage, and straying from a literal translation in order to achieve a translation that makes a more standardized use of the English language. People are going to differ on these topics, and their reasons for doing so aren't necessarily cut and dried.
The new locking system (it's still being tested, I know) has the potential to end all possibility of debate. If sentences get locked after the third edit, the person doing that third edit wins.
I think that any attempt to programmatically reduce edit wars will be a demotivator, especially for the folks that Birdexplorers describes who are currently motivated to reach the best translation.
Are true edit wars that prevalent, where one person continually reverts to their own bad translation? Prevalent enough to risk driving away the well-behaved translators?
This is good news, but there problem is when someone is in the highest tier currently, which as far as I've seen is currently level 7, then the person becomes untouchable. No-one will be able to edit/change the stubborn user's translation if the user decides to reject or ignore all attempts to improve the wrong sentence.
In my opinion, something else needs to be thought of for this scenario.
@Luis, now that makes it effective, the only other issue I can see with that, is with old or "completed" articles, such as this one:
The very fist sentence/title, is incorrect and inconsistent, Jean is a French name that could theoretically be translated into John. But for the purposes of consistency, this just doesn't make sense. Most of the document contains Jean, and only two occurrences are John. If this was locked, and users continued in the tag of war, then it would never end.
The title alone has 19 upvotes, and the article has been deemed complete. So this is like fighting against the tide. We seriously need some form of consistency control, or some form of requesting a widespread peer review.
Something I suggested here(http://www.duolingo.com/#/comment/1093740) could correct this obvious mistake.
The alternative is for users to setup immersion gangs, and start downvoting the articles the group disagrees with, offering new translation and upvoting that themselves.
Is there a way to see what tier other user's have reached? I see my own, but of others I can only see what language level they are at. I agree with Salxandra... the other day I saw the translation of a dude who is at level 25 and it was hilarious. From a Wiki article he had translated that "there had been a shop that catered to horses". (The Italian expression "a cavallo di" means "between" and has nothing to do with horses = cavalli). Lol.
Thanks, that makes sense.
Although there must be some people at high tier levels who are upvoting without really paying attention. Here's one translation: "It's a chance of deep philosophy that he endorses to hit the forcefulness that he tries to give to his words". There is no full stop at the end of the sentence, and aside from what the original Spanish means, this sentence just doesn't make any sense.
I see that it does indeed have 3 upvotes; the translator who wrote that translation is at level 6 - what level would those upvoters have to be at to endorse this sentence enough to lock it?
The scenario of poor quality translations from higher level translators happens all the time. I have seen it countless times in Italian, and, less often, in French. Yesteday I was looking back at an Italian translation from a level 5, which was said to be about 90% done, where almost every sentence was badly translated and some seriously wrong. These had, thank goodness, no upvotes, and the wrong translations were not locked, but they were marked as 'checked by the community'. I have come to the conclusion that the tier levels are almost worthless, and the algorithm for calling something checked seriously flawed. As is, by extension, a lock-down algorithm that relies on tier levels.
Alipaulam, I didn't think that a translation could be marked as "checked by the community" with no upvotes...?
EDIT: Ok, I just uploaded an article and did some translation which nobody has looked at yet and it shows as 32% translated and checked on the immersion homepage. I believe it's because certain short or common phrases/ sentences (e.g. in Wikipedia articles) are automatically marked as checked.
[Answer to Alipaulam's comment beginning with "The scenario of poor quality translations from higher"]
yes, that's why I think downvotes are useful (for Duo to make a better evaluation of users' translation skill) and think --- as you think, it seems --- that they shouldn't be thrown away...
Thoughtdiva - Or maybe this happens if your level is high enough.
The ones I looked at weren't all simple or standard. The sentence that first caught my eye was so long I decided agaist reproducing it here. But it was undoubtedly wrong; among other mistakes, 'sconsigliare' was translated as 'recommend' (with no 'not' in sight). In my book that's a 'looks wrong' error coming from a level 5, though forgivable from a beginner.
It was also a very old translation, and the user appears to have left, so maybe the Duo algorithm just gave up hope that anyone would come along and check it!
Alipaulam - interesting thoughts. I guess I can't explain it, but maybe it does have to do with tier levels...I don't believe in "looks wrong", although I have been tempted sometimes. For example, one 6-word sentence was translated perfectly by a tier 7, then changed by a tier 1 to lower case for the person's name, an obvious spelling mistake, as well as changing the meaning to gibberish...but yes, we can be more forgiving towards beginner translators as they are still getting used to the system.
To answer your question, yes I believe so, and I described a few threads(1) that highlight the problem clearly. Besides, if the feature that enables higher tiers being able to change the document is implemented, it will resolve this issue for committed users such as Birdexplorers, since they will get a lot of votes for their work and always be in higher tiers.
This has been a problem that has created countless threads, and it has annoyed Duolingo staff for enough time to actually do something about it. So I guess it is prevalent, since Duolingo would not spend considerable time to implement a feature to address an insignificant problem. I think Birdexplorers is one of the more patient and caring users, so maybe Birdexplorers' method works well in Birdexplorers' case.
But it doesn't work well for others, and like the user stated, the documents uploaded by Birdexplorers are mostly domain specific. People shy away from touching articles they hardly understand, so that may be the reason. The locking system is good in my opinion, now the only thing that is needed is something to counterbalance it, and prevent abuse like the up-tiers being able to unblock it, or enough downvotes unblocking it.
How many votes is enough for a sentence to be locked down? I just saw a locked sentence with only four votes. Unfortunately, the translation was clearly wrong. If it wasn't locked, I would just edit it. But since it was locked, I didn't take the trouble... That is one of the downsides of locking sentences...
I'm not saying this is relevant to your case(s), it's just something that crossed my mind while reading this thread. If a sentence is incorrect and locked, and the person who "controls" the sentence no longer uses Duo, or is in the hospital, or is on vacation for a month, etc... then the sentence is doomed to sit there incorrect, possibly forever.
There should be a work around of some sort. If a sentence is locked after 3-4 upvotes, then surely 1 downvote (that is equal to 5 upvotes) should unlock the sentence?
Jaira, I can't remember now why I wanted to use downvotes that day, but generally it's either because of a clear mistake on something easy, or gobbledegook, by someone at a higher tier than me; or somewhat vindictively because someone has replaced a good translation of mine with something worse. As I've said, though, I don't use them at all any more. Vindictiveness is a bad reason anyway, and can rebound on you if you're the one at the lower tier. In the first case, I'll just edit; in the second case, I often let them go by altogether if there's the risk of getting a downvote.
@ BarbaraMorris The "how sad" was in response to Alipaulum who had apparently been included in a trial without downvotes but which were later reinstated. I had hoped the trial would lead to the abolition of the downvotes. As for getting updates it seems we're all stumbling around in the dark. Bits and pieces picked up on threads are all I get. If anyone knows how to get info on what's actual policy, what's being tested etc please pass it on.
I would at least expect to see my version replace the locked one if it were accepted; if not, why bother? And a notification of refusal, with reasons, would help prevent me making further poor suggestions. (Or it might just make me real mad if I still disagreed with the reasons given) !
Feedback on down votes
- I think the down-votes should be equal to the up-vote
- I think we need to know what caused a down-vote so we can learn
- I think people with higher tiers should also be able to be down-voted in some way by lower tiered people ... just because a person has a high tier does not actually mean they provide excellent translations (maybe 4 down-votes by lower tiered people = 1 actual down vote)
- I like the idea of down votes, it puts in some level of accountability
It needs to be clear to anyone reading the immersion section that they are penalising others every time they hit that button rather than making their minor revision. At the moment there may be no common understanding of the effect. Someone seems to be getting slap-happy with it today.
It doesn't help anyone to assess anything if it isn't given the same weight and used in the same way by everyone. If some regard it as a last resort if someone posts gibberish, and others think it's the routine response to something they think would look more the way THEY want it if they imposed a US date format on it(!!!), then there is no parity and the downvotes mean nothing beyond being a way of undermining others.
More on Down voting
- It seems that it might be done accidentally
- because of the huge impact it makes it intimidating to correct a tier higher than you
- because of the huge impact it can be used as a weapon
- Getting down voted with no reason given is discouraging
- High tiers can be obtained by brute volume effort not necessarily skill
- The reasoning for having a high down vote impact compared to up-vote is very hazy (not clear) to me
- current system does not seem fair
- make down-voting and up-voting the same impact
- When doing an down-vote
- make it clear there is a point impact on the person that is being down-voted
- a reason has to be given for down-vote
- the person being down-voted gets notification with article and the reason (helps the learning process)
- I don't necessarily need to know who did the down-voting just the reason and article
- Higher tier users can be down-voted by lower tier users .. maybe at a lower ration 3 down-votes = 1 actual down-vote impact
Firstly, can Duo implement some kind of "locking-down" etiquette? I came across a locked sentence discussion page, and then spent unnecessary time looking in the general discussion pages to find out why. Perhaps the moderator who locks the sentence can add a quick reason why the page should be locked.
Downvoting in Immersion. I HATE this feature. In my mind, Immersion is supposed to encourage people to attempt translating sentences from a language they are still learning, with feedback from people more fluent in the language (including those people who are also still learning English). It may be great for those people on higher translation levels, but it is a huge disincentive for those at the bottom. As a level 1 translator, I hesitate to translate more complex sentences because I know that translators can downvote me if they disagree with my translation. I'm happy to receive constructive criticism, but I have found that people editing the sentences rarely discuss why they have done so. In addition to not understanding why my translation is incorrect, it then means that I have to correctly translate another 5 (?) sentences to make up for what could be a minor error.
I have seen sentences edited for no other reason than because another translator preferred to use a different synonym, or preferred to alter the order of the sentence slightly (even though the initial one was satisfactory). It would be a shame if they had downvoted the first translator, but I'm sure that happens.
What really makes me cranky about the downvote button though is that sometimes the higher level translators are wrong. In the last week, I was downvoted, twice, not because of my poor translation (from French to English), but because they disagreed with my English grammar.
I may not be proficient in French, but I can guarantee you that my English is pretty good. In both instances above, I contacted the translator, and provided a link to an English grammar site explaining why I was in fact, correct.
So here we have a situation where a higher level translator with a poorer understanding of English grammar can downvote a lower level translator, who has translated the sentence to English correctly.
I found both experiences unfair, discouraging, disappointing and it made me quite angry, because I feel like I am being penalised for their lack of knowledge.
I was downvoted on another sentence - I couldn't make any sense of the French original, and after spending considerable time trying to do so, I managed to find what did make sense on the English Wikipedia site, so I did copy and paste that, and referenced that in my comments. Some small-minded little man downvoted it and told me he was reporting me "for abuse" because I copied and pasted, and didn't translate.
What particularly made me cranky about this was that I originally acknowledged where I found the interpretation, and also, that there are no official Duo guidelines for translating. Sure, there is that unofficial guideline (which isn't easy to find, I might add), but they aren't exactly "the rules" then, are they, so how do you report someone for "abuse" on that basis? I am pretty certain that the French wikipedia page was a translation from the English wiki page (and yes, I did look at the historical edits on both pages), and upon checking a third site, felt that the English wiki page was factually correct.
So this small-minded, higher level translator clearly felt that it was better to translate to English, and render the English sentence as meaningless and incorrect, rather than translate into something that is factually correct. If the aim is to correctly interpret it, disregarding factual accuracy, or regardless of whether or not it actually makes sense in English, he could have achieved that without downvoting it.
I don't get it, but I can say this. My time is precious. If I find that I am going to be unfairly downvoted, or reported for perceived "abuses", I'm not sure that I'm going to feel inclined to pursue translating. And if I, as a native English speaker feel this way, I wonder how discouraged those for whom English is not their first language are.
Okay, I'll try this again. (I just typed a relatively long reply, only to have it inexplicably erased when I pressed "Post". Remind me to ALWAYS copy what I've typed before hitting "Post"....)
This is really disheartening and the reason I agree that down-voting people is completely unhelpful. It simply discourages people from continuing. I am in the fortunate position of being at a fairly high tier so don't know if people have wanted to downvote my translations. But I have received notes after my translations have been corrected, saying "In English, we don't write/say/spell..." Granted, I don't know everything there is to know about English, but it is my first language and I'm relatively proficient and relatively articulate.
In the case of an article that turns out to be clearly one that has been translated from English (how's that for a nice awkward construction?), I immediately stop translating it and down-vote the article itself. I try to remember to put a note in the discussion area explaining why I have down-voted it.
As you say, time is precious and there seems to be little sense in translating into English something that is already written in English.
@ejm_ehterwork, in case you were wondering if a moderator deleted your post, I just checked up on that. There is no sign of it. If another mod had deleted it, I would still be able to see it. I'm sorry it disappeared right when you posted it. :( I keep doing that somehow too. I think I am pressing something with the palm of my hand. I'm on a laptop so I'm almost certain that for me at least, it has something to do with my scroll box. But, it can be frustrating to put a lot of effort into something that gets erased as soon as it's finished >_<. So, don't forget to copy!! ;)
As for articles that have an English version floating around out there already, it isn't anything to worry about. Most articles posted to Immersion are just for practice unless someone has uploaded it for a nonproject or if it has been marked as "Commercial" (The articles marked Commercial fund Duolingo.)
I hope this helps ^_^
Thank you, Usagiboy7. I hadn't thought that my comments were being deleted on purpose. (I should have made that clear.) It's something to do with a.) my connection b.) DuoL site c.) Chrome and/or d.) my overactive mouse somehow hitting cancel as I head it over towards "Post".
(Hah!! This time, I remembered to copy first!)
edit: of course, I had no problem posting....
It's wonderful to know that a person who spends the necessary time to research and get things correct is working in Duo. The system does have its imperfections and injustice. What's also frustrating is that if you try to contact admin about important matters, they send you a pre-fab fix for a common glitch that has nothing to do with your question. They obviously didn't even read the question I sent them, and although I responded to their email asking them to do so, they never got back to me.
I think it would be useful to keep downvotes as an indication of there being something seriously wrong with a translation, that perhaps you don't have enoughknowledge to edit, or which someone persists in posting despite being told that it's wrong, but to remove their impact on people's tiers. So vindictive or ignorant downvotes don't affect your tier. This is reverting to the system much as it was before tiers.
But without the negative penalties, and some form of judgement of translations, you have to wonder what the point of the tier system is anyway?
I agree with comment above that 'downvotes' being used as an indication of there being something seriously wrong with a translation, that perhaps you don't have enough knowledge to edit,"
I think the 'Down Vote' should be re-named to "Is something wrong here?" button.
This then allows problem areas to be highlighted and flagged for debate and discussion.within a translation.
I don't think this should carry any penalty.
As for translators completing volumes of work and then rising up through the levels willy-nilly then perhaps a 'gateway to the next level' short test could be set. This would give a sense of achievement and would give others greater faith in the competence of those awarded the higher levels.
OK. Here are some other suggestions. I don't feel strongly about these; I'm just brainstorming.
If you keep down-votes, require an explanation for the negative vote in the comments. If someone changed my translation, and thought it was horrid enough to vote it down as well, I think I'd accept it readily if they offered me a polite explanation.
Give the person who gets a down-vote a chance to respond and to wipe the negative vote clean. If three other people read the changes and vote against the down-vote, it goes away.
Here's an example: I get a negative vote from Person X. I can check a "protest" link, offer a comment, and ask other people to offer their opinion and vote.
To keep this from being about popularity, limit the number of times one person can vote to suspend another's down-vote to - let's say - once a month.
If Duolingo is going to depend on users translating articles to bring in income, then it's important to keep it fun and sane.
Users trashing other users in comments are a big part of the reason people have negative opinions about news sites, for instance. I'd hate to see that happen here.
That is great brainstorming, mreadercit! I have complained about downvotes having five times the weight as upvotes, but if the system were set up as you suggest above, the downvotes could become a valuable learning tool.
I think the most important thing that should be done is that when you press that little X button a dialog box should open that says "Do you know you can edit a sentence without downvoting the current translation? If you really think this translation does not confer the meaning of the original, please provide a reason for your downvote here."
I know, but it turns into such a faff every time you want to slightly modify what you typed to express it as well or as politely as possible, I end up having to go through the whole rigmarole half a dozen times for the sake of once sentence. It would be nice to be able to just edit.
Yes, I agree completely.
And the circumvention I gave only works for ordinary edits. For locked translations, you can only see your suggested edit afterwards. You can't see the comment you added.
(I really really hope that they scrap the sentence locking entirely.)
I like the new colour schemes, it makes it easier to read the document, personally, I'd prefer if it was customizable, but this scheme works well for me. I also like that you used our community lingo for excessive revisions to a document(Edit-wars), and I personally call the participants Edit-Warriors.
I support the idea locking sentences, this makes it more collaborative rather than forced edits. The downside, is that sometimes people may not check their email, a notification in duolingo would be good. In addition, if the author is stubborn and doesn't want to edit it, then I think there should be some sort of request review button, to mitigate disputes, and alert higher tiered users to come and review the sentence. Also I think there needs to be some way to finish up old articles with edit wars, maybe showing them up more often.
Oh the lovely downvotes, I completely adore them. If for nothing else other than the fact that it creates a lot of interesting debates around the subject. Taking them away will not help because artificially inflating someone's rank based solely on perception is a very dangerous thing, as history as shown with many incompetent leaders.
I'm a firm believer that a person's ability does not always keep rising, and in fact in some cases does diminish, e.g. old age, not using duolingo for months, and so on. The Tier Level would not accurately reflect that, in addition to the fact that there is no other empirical assessment (test) of Tier levels, other than user's perceptions.
I would also be interested to know how much of the Duolingo population actually uses immersion, this is another thing to ponder when evaluating these changes.
Edit: I also think that if they remove downvotes, they should also remove tier levels, or perhaps call it something else. Rank without certification, is like a balloon with hot air, and I fear it will meet the same fate as Icarus.
yay i'm on the test group. no more downvotes!
i don't think it's a bad thing that in the immersion you can only go up, it's pretty much like skill points or the gold coins in duo.
But i don't mind if some1 in higher tier can downvote tho it still should have the same weight as upvote x1. and i was wondering if you can base that only on someones tier lvl, i mean- what about his tree.
for example if somone knows something important from higher branch of the tree he might have better translations i believe. it shouldn't be based just on tier lvl.
As far as locking goes, what about having the person to whom the sentence is looked be required to either accept or reject changes (with comment) within a certain number of days, or have the sentence unlock again.
Additionally, rather than have it lock after the third edit, have it lock after a certain number of upvotes from a sufficiently high translation tier.
I agree about the time limit. Although I wonder what would happen if more than one person had submitted changes for approval. I saw it suggested somewhere that the first person who had submitted a change should get their edit automatically put in. But they shouldn't really get the lock immediately because they don't have enough upvotes. But then how would anyone know about the other potential changes?
About locking after a certain number of upvotes, Luis said this above: "The translation must have at least three edits, but it also needs to have enough upvotes from people at high enough tiers."
Thanks for all these improvements and tests to make Duo better and better. And for the announcement/keeping us posted!
Question: with the downvotes desappearing, does it act retroactively? Will we go up in tier level counting that we should have level up sooner if we haven been downvoted and so some of the upvotes from users in a strictly higher level (at the time of the vote) should count as an upvote for our actual level? [Don't know if I'm clear ;)].
About downvotes, let see what you're statistics say to be better, but I think it risks to biased with only upvotes.
Indeed. But a system in which you can only goes up, well it seems to me that it misses something. But, as said, let see what the Duo's statistic at the end of this period of test will discover to be the best of the two methods for them to be able to evaluate the more accurately the users (which is, IMO, the goal of tier level: a tool for Duo, not for users).
Elgohn, the "looks wrong" button IS the downvote. When you push that button, five points get detracted from the score of the last translator or editor. I was suggesting that a warning dialog box could pop out when you push that button -- which says something like "Do you know you can edit a sentence without downvoting the current translation? If you really think this translation does not confer the meaning of the original, please provide a reason for your downvote here."
I must be on a test then, because my downvotes are not listed anywhere and my explanation of the tier system reads thus: " You advance to the next tier when sentence translations last edited by you receive a total of 100 net upvotes from other learners in your tier (or higher). Each "Looks good" rating counts as 1 net upvote, and each "Looks wrong" rating takes away 0 upvotes."
Does that clarify my comment?
It's good that you're going over the "locking" issues. I would suggest, though, some way to go around a locked translation (a report of some kind?).
Also, I don't think 5 downvotes is too harsh at all. You cannot drop tiers, right? And bad translations (too such a degree that they merit a downvote) should count pretty heavy. If you translate well you will hardly ever see a downvote.
edit: I should also lift up one problem with locking and downvotes. Now that locking is enabled, sending in a suggested revision will not register as you having gone over that particular sentence/word. Thus leading to you having to, also, either upvote or downvote the sentence in order to be able to move on and finish the article.
I think the problem is that a downvote is just to easy to do, without offering any kind of explanation or communication. It's like kicking someone in the back of the heels as you walk behind them. I think the downvote is most likely to be used by people who think the translation is wrong, but can't even clearly say why - which is a very weak position, and in that case, their opinion should be taken with a tablespoon or two of salt, instead of relied on for a final verdict.
I think the whole system would be smoother if at the very most, a few downvotes was enough to 'unlock' an already locked translation. We don't need sticks to beat each other with, we just need to be able to communicate properly, and to that end, being able punish other translators just creates an atmosphere of passive aggressive.... take a word beginning with B.
Yes, that is also an insight into the problem. I'd hope people use the downvote most often out of ignorance (they think they should use it before editing), but it could also occasionally become a weapon in the hands of aggressive, jealous or vindicative people. :(
Adesva, I don't agree with you about the 5x penalty for downvotes. Right at the beginning of my "immersion experience" I had a problem with the highlighting. To explain better, the highlighting in the original text corresponded to other paragraphs in my translation. I got two downvotes because of this. In the end I probably had spent an hour or even two of my time and my points were 1/100 in tier one. I knew my translation was excellent and I couldn't see exactly what had been downvoted or by whom. I was so discouraged I thought I'd quit translating. Also, some people don't read the recommendations and simply press the X ("this doesn't look right") button because they think they need to do so before editing. For what reason should a downvote be weighted more than an upvote?
I am finding that the color scheme nudges me to be more thoughtful when going over the translations of other people. Sometimes it's easy to get into a rhythm while reading, but the different colors are almost like traffic lights. I suppose every sentence should be read very carefully, but in a volunteer situation, it's more realistic to have a system that flags sentences that she be read more carefully than others.
In principle I really like that idea, but I think it also could be gamed, given the disparity in how many people look at different translations, and at different parts of translations.I have seen some people already gaming upvotes, by just going in and doing the first sentence of a lot of translations, and picking up upvotes from people searching something to translate, before they get bored. By contrast if you do a translation all the way through, you can come back and find that no-one has looked at your stuff at the end of a piece even after several weeks.
Redoing the initial tiers based on the average of the historical data could at least start people off more evenly and avoid the current problem of people with undeserved mega-high tiers being basically unassailable.
I think it's too bad the tier levels were exposed to us, opening the system up to gaming. I guess it might attract more people to translation. But I wonder whether most of the former non-translators will just invent better and better ways to grab tier points without actually doing any meaningful translation.
Downvotes only from people in Tiers higher than you = ALL THE +10's!! :D Now all we need is a gradiant up vote scale, where people with Tier's lower than you get less than 1 point for upvotes. Dessamator had a great alternative idea though too involving captcha tests. Anyone have that link?
And locked translations should involve a number of approvals from T4 and up.
Translating isn't just a word for word substitution. Sometimes it's quite difficult to understand what the original sentence means. Sometimes I will come across a sentence that has already been translated, but it is obviously wrong or just sounds very unnatural in English. The problem is, sometimes I can't figure out a better translation. So, what are the options? I don't want to say it looks good (because it doesn't), I don't want to edit it (because I know I can't improve it), so that leaves the looks wrong option. Usually I just don't do anything and just skip that sentence. But I do think there should be a looks wrong option (or something similar), only so that it may raise a flag and someone else who is better can notice it and translate it properly. I don't think It shouldn't count negatively towards the person who originally translated it (although some are really bad, like the person just copied and pasted a Google translation).
You have a point there Brg. There should be a way to flag a sentence that you believe needs work but don't feel you can do well yourself... or even for just coming back to if you don't have time. At any rate I think it should be compulsory to write an explanation for your flagging. The former downvote system only penalized the translator or editor without even letting him or her know where or why.
Congrads on reaching Tier2. Sorry it was marred by frustration. All we're asking for is to learn what was wrong not who marked it "looks wrong'. My practice of going right to the home page stream and proposing changes has brought a variety of results:''thanks didn't know that'' being most common. Others showed me why both were acceptable and still others showed me I was wrong in which cases I came out knowing more. Oh, there was one : ''Mind your own business.'' Takes all kinds which includes the anonymous people who hit 'looks wrong". One query do we know if DL monitors the downvotes?
Correcting a translation does not mean someone originally translated it incorrectly, because there is more than one way to translate a sentence. Some of translations are simply better than others, and editing can help achieve the best translation. Wrong translations are simply wrong, not correct, not even less desirably correct. So, we mark those translations with the "It looks wrong" button. Meanwhile we can use edit button to fix both the less desirable translations and the wrong translations.
On the wider subject of things that need to change in immersion, it would be helpful not to be repeatedly told that very mainstream articles from as anodyne a source as Wikipedia on major figures in history or artistic works 'may be offensive' and have them rejected - I've seen far worse in the comments and even in the lessons than in anything I've ever tried to upload yet I still find material rejected.
This happens to me on occasion, and I simply write to DuoLingo Support explaining the problem and providing the URL for the link that I want to upload, and usually in a few days, they fix it so I can upload the article and translate it. Support is always quite helpful whenever I have any problems or issues on the site, which I have always found to be a very nice feature, as it shows they take their users seriously.
Personally, black and grey are easier on my eyes than green. I preferred it during the brief period when my completed translations were grey than green. Reading through an entire article in green after I've translated it is just nauseating to me. I realize I'm in the minority here, but I felt much more like clicking on blue sentences to turn them to gray than I feel like clicking on gray sentences to turn them to green.
I find the Real World practice really fun. One thing I must say, though, is that I think I got some downvotes because my translations' patent structures differ a lot. Maybe, that makes some people think they are wrong. Sooo, from now on I'll comment about the translation techniques I'll be using, that'll help people interested in translating well. :)
Downvotes should exist, but how do we make sure people don't do it just because they're mean xD
I really love DuoLingo. Thanks.
The patent structure is how ideas are written or said. Take this short example: Vous aller avoir un bébé = You are going to be a daddy. Can you spot the word "bébé" in English? Of course not. Patent structures in both languages are different, but they describe the same fact, in other words, the same "Underlying structure" or meaning. In the example above, I used a "modulation" to translate the same idea. Cheers!
I always copy/paste the good translation as an edit, and add a comment saying "Reverting to DonaldDuck's translation" (and sometimes additional comments about why).
They used to give the original translator the credit. I don't know if this was the reason for the change, but one problem with MinnieMouse reverting to DonaldDuck's translation is that if was not actually a better translation, it makes DonaldDuck look like he is trying to push his bad translation.
I'd like to see both people get credit (or blame), something like "DonaldDuck, reverted by MinnieMouse". And maybe for both people to get credit for the upvotes and downvotes too.
There is a problem I encounter from time to time. When I want to revert to my version I want then just to add a comment explaining my choice and I cannot save it. It cannot be saved unless I change sth in the translation. And I do not want to do this. Does anybody have this same problem as I do? Can this be resolved in any way? Thank you.
Haven't managed to read all posts thoroughly yet but at first glance I'm iimpressed by the tone. So many opinions, and often differing opinions, but all with the aim of producing a well done translation while improving your language skills. No, grasping for credits or names in lights. My hat is off to you (y'll as they say). Thank you DL for this initiative, it still needs polishing, but will be a major benefit to your already amazing site.
Many thanks for correcting the "y'all" misspelling. Is my face red! I lived in the south for a time but could never bring myself to use it [or learn how to spell it :-)]. Thanks also for the link "youse' is a real sore spot in NYese but I never realized how widespread it was. P.S. I've corrected the above spelling.
Yeah, I didn't know how widespread it was until I saw the link also. I was just curious, because it popped in my head when I responded, so I looked it up to see if it was a contraction or what.
No worries. I mispell stuff all the time. So bad even auto-correct tells me that I'm on my own. I've lived in the south for way too long and have even had a discussion of why "ya'll" is not the proper contraction for that. It seemed silly to argue at the time, but it started with them telling me I was wrong. I don't mind being wrong in silence, but out loud, I want a reason. Turns out it was me that wrote "ya'll" ("see it once,, and assume everyone does it that way" type of thing). Now, I'm convinced "y'all" is more correct. Especially since it is in the dictionary (albeit online). Is "albeit" short for "all be it"? Guess I'll look that up next.
@TerraZ :-) :-) Why the smiley faces? As I was typing the previous post my auto-correct kept printing "you'll' (I'm not saying that's what caused the error. I'm very capable of doing it on my own) but I'm smiling that you're having similar problems. Auto-correct really goes berserk on the German and Italian. It keeps us on our toes, though. BTW since it's clear you want things to be in order, you have a tiny error in your second post, second paragraph, second line.
That very error of writing an incorrect homophone of a common word is the type of thing that makes me think that the writer is somehow not well educated or doesn't read much. Ones like "there and their", "too, to, two", it's and its, etc. Then one day I was typing along and I noticed I had made a mistake of that nature myself. I stared at the screen in disbelief. WEIRD! But it happened. Lol. I really try to never be judgmental but it's probably a part of human nature.
The translation feature of Duolingo was called "Immersion". Everyone was automatically a translator. Duolingo removed the Immersion feature a couple of years ago.
It was the -best- part of Duolingo. I always learned so much while I was translating "real" articles, and it was fun collaborating on the translations with random other people who happened to be translating the same article. I really miss Immersion ... -sigh -