Workers feel this requester pays well
edirewrite Wage History
Top Worker Reviews
AfterDarkMark Average Pace
Ed-Judge the relevance of a machine-translated document summary (2019-01-08) - $0.30
$10.00 / hour
00:01:48 / completion time
ProsCan be done faster. Did a bunch, so not really sure what my average is. READ the cons. Might save ya some trouble that I almost had. T
hese are good tasks though. Master's. Certainly pay fairly/ well. Speed up once you get the hang of it, though some tricky ones to work out.
Helps if you can skim quickly or you might be slower than I was at these. Again, not sure on my average time. Probably did some in under a minute. Some in over 2 minutes, more at first.
ConsBE CAREFUL! Within this same batch (there was a different 20 cent one up at the same time, but just talking about this one), the instructions can vary hit to hit!!!
There was a set of orange instructions or pink instructions on the first page of each hit at the bottom, but within the same batch they change frequently! And don't go simply based off the word at the top (which is colored orange or pink) as sometimes that stays pink, even though the instructions at the bottom have changed to orange.
Here's why that's so important. For the orange ones, we were looking for the exact phrase or a synonym in the same form - example the requester gave is that lavatory is okay for bathroom, if that's the word, but bathrooms (plural) is not. Or Purchased is okay for bought (cause both past tense/ synonyms), but buy is not... So could only mark the scale if the word was exact or a synonym in the exact form. In this version, if submarines was in the document and the word was submarine, then none found should be marked.
For the pink instructions, the form did not matter. So, if the word was navel, then belly button, belly buttons, navels were okay. Or if the word was buy, the bought, buys, purchased, etc are all okay. So any form of the word should lead to us marking the main scale and not the none found (or whatever it says) bubble.
That's really tricky and sneaky (though not intentionally I don't believe). I didn't notice until I had done quite a few, so am actually expecting some rejections possibly. Hopefully not, but some of the questions (think like one or 2 our of the 6 that are in each hit are quality control questions - usually easy to tell which ones - and I fear reading the instructions wrong or not realizing they were changing a bunch may have led me to screw some up).
This might not make sense, until you actually do the hit, but it's really important. So used to batch instructions staying the same throughout that I didn't reread the instructions every task or was just mindlessly skipping to the 2nd page where the actual task begins. Guess it's a good lesson to be more careful if I do get the expected rejects.
klee Average Pace
Evaluate summaries of reviews - $0.80
$14.12 / hour
00:03:24 / completion time
- HIT Rejected
Cons1 out of 5 of my submissions was rejected with a message "Thank you for your time and effort. However, your submission contains incorrect answers to sub-tasks to which answers are known in advance."
I was very thorough in making sure I marked the ones that were obvious with the correct response. I suppose I could have missed one but I really took my time with these hits to try and make sure I didn't. So perhaps they weren't as obvious as I thought and these rejections are subjective. I'm not going to keep doing these, even though the pay is ok it's just not worth the risk.
johnxyz Proficient Worker
Rating fluency of English questions - $3.00
$26.67 / hour
00:06:45 / completion time
- HIT Rejected
ProsAll inside frame
ConsCan be subjective
Some questions had no answer that made sense, but you are forced to pick one
REJECTED: "Failed control sample: selected output that was exactly the same as original as being the most dissimilar." It's not fair because all questions must be answered, and thus we have to pick one which is dissimilar (even if they are exactly the same.)