Immersion suggestion: Don't give rewards if the initial translation is the same as Duobot
There always seem to be a few people who systematically post the Duobot translation as their "translation", sometimes hundreds of sentences a day.
How about removing the incentive by not giving any XP when an initial translation exactly matches the Duobot translation? And maybe don't count "Looks Good" votes for that translation.
For short sentences, where it's more likely to match Duobot through our own efforts, the loss of XP would be quite small.
For long sentences, there might be the occasional case where someone happens to get the same result as the Duobot translation through their own work. But I don't think it would happen very often.
(I often check Duobot after I have done my own translation, and very occasionally I have exactly matched Duobot. But it's extremely rare that Duobot does a good job for anything more than a few words.)
I think the risk of well-behaved people losing a few XP and a few Looks Goods is worth it if it reduces the spamming of Immersion by idiots.
I worry that if we did this, spammers would just work around it by slightly changing the DuoBot translation. I think a better approach is for us to identify people who copy-paste DuoBot hundreds of times, and take away the XP they earned by doing that.
Lingot for creating the topic. :)
I would only modify as follows: have the algorithm flag such sentences for abuse@. The automated messages to abuse@ could have #hashtags that allow them to be treated differently than reports written by individual users.
Denying/rescinding points would be one option for DL. Sanctioning repeat or heavy offenders with suspensions or expulsions would be another.
Personally, I think that all sentences shouldn't grant any points to the author if they are exactly like Duobot. We are here to improve translations, not to amass points anyway.
If the sentence is the same as Duobot's one, then it means that the credit should go to Duobot, and not the second "entity" (person) who came up with it. At best that person should earn proofreading points (at worst nothing), and not points for creating a new translation, regardless of whether the translation is the first one or not.
Cool idea. My experience pretty much parallels yours.
And here's another, perhaps slightly more radical, idea. Lose the Duobot translation entirely. Make everyone translate from scratch.
Just a thought.
I've definitely been guilty of using Duobot as a crutch, to avoid looking stuff up in dictionaries or grammar books. The thought has occurred to me that I might learn more if I had to shovel the coal myself.
I got a good chuckle out of your post, btw.
I think it would be a shame to get rid of Duobot completely. I nearly always have a go myself first, unless I'm stuck, and then I look at the Bot to check. On not a few occasions, I have then changed a word or two in my version because it has actually come up with a better word.
And yesterday it came up with the absolutely correct translation of something I might never have got by myself - an translated extract from the US constitution, where its vocabulary was so odd that I looked up the original - and there it was. Now, I might have looked up the original anyway, but I might not, and that's exactly what machine translation is good at.
Sadly, however, these occasions are of course vastly outnumbered by those where the machine translation is gibberish.
I don't think it's a good idea to take away Duolingo translation. It's helped me a couple time before. Not to mention if you take away Duolingo translation, people would just use Google Translate, and I did have noticed people posting Google translations directly :-/ I love Duolingo immensely, but I have to say the immersion system has some serious flaws.