Respeted Sir, I think i finally found the work which is worth for me, Here is my Worker ID: A2M5862MTN7T4A, I have done around 7352 approved hits with 98.6% quality, I would be greatful if you could grant me he qualification to work on OCMP34. Looking forward to work with you. Thanks and Regards Yogendar
Sorry guys - the accounting team is working on it as fast as they can, I'm sure. I can't post until they're able to do their good work. Meanwhile, my data team and I are working out a new data upload/download strategy - I don't yet know if that will take us out of commission for another few days.
Hi everyone, I've gotten the go ahead to run several more batches, I'm just waiting on the accounting team. The good news is I should be able to run batches even while we work on the new upload/download strategy (behind the scenes stuff that shouldn't be noticeable to you at all). Thanks, ocmp34
Hey all! So while we're on board now for some time, I'm under a lot of pressure to lower costs. I'd really appreciate if you had any ideas for lowering costs (that don't involve paying you less!). Are there tools we could build that you could use to scrape data (remember that we have a whole dev team we could steal a little time from)? Is there information you need that would help you find more donor lists? The retrieval step is definitely the worst: so few organizations have donor lists, so it's expensive to look at them all. Is there a pattern to orgs that don't have donor lists, such that we might weed them out? What questions am I not even asking? On the one hand you may be thinking, why would we help ocmp34 if it means less work for us? Remember that I still have well over one million orgs that need researched, not to mention brand new tasks we have in the pipeline after our company's big spring deadline. And, once we get the new upload/download pipeline running, there should be a lot less pushing the big red button - that is, I'll be able to let loose huge chunks of tasks rather than the small batches I push now. I've got a few ideas the ocmp team and I have cooked up already, so hopefully there will be some very minor ease-of-work changes in the near future. tl;dr What can OCMP34's dev and data teams do to make your job easier, thus winning us better results and you better pay? Thanks, ocmp34
I can tell you that hospitals and medical organizations are the hardest/least likely to have the donor public. At least that is my personal experience. Maybe others have had more luck on them. The problem is that a lot of us do not like giving up on searching. You could always state to search for x minutes, and if unsuccessful to submit with an N/A. The problem is, with enough digging the donor list could probably be found, but the "enough digging" is what makes the pay to time ratio not worth it. I really do not know what else could be done to make these more efficient, thus making the pay less and allowing you all to hit budget.
I'm not sure this is true. We have done some in-house testing and have found that even if we search for an unreasonable amount of time (like an hour), you guys are 91% accurate - that is, you are already finding 91 donor lists for every 100 donor lists that are online (to the extent that our in-house researchers were able to find them). P.S. - that's an increase of more than 20% from before the curated workforce... you guys are amazing!
I feel like adding hot keys to the "Find Company Website" HITs would make it faster to do those. For instance, say you're looking up St. John's Hospital. So, you find the website, copy and paste the URL to the HIT page, but now you've got a list of 10-20 different categories, and you're supposed to categorize the organization. Since we know it's a hospital, we categorize it as a hospital. But instead of sitting there searching through the list to find that one little radio button, you could make it so there is an option to press a pre-defined hotkey (for the sake of my example, we'll say 'H' is the hotkey). So, you press 'H', it selects the "Hospital" radio button, and moves you onto the next organization to be found. It may only take a few seconds to find the organization type from the list of radio buttons, especially after doing them and getting the hang of them, but I feel it's worth mentioning that since we work for change here on MTurk, seconds are kind of like the equivalent (in my eyes, at least) to hours in the "real working world". Just like the pennies we work for, it all adds up at the end of the day. As far as the donor list retrievals, I can't say I can think of a way to speed those up, besides maybe what Turbs said, since that's much more complex than copy/paste a web address and categorize the website.
I am glad we are as accurate as we are! I think removing those that were scamming is going to save you money as well. Hopefully others with ideas will come in here. I like thetrolls idea of making the different types of companies into hotkeys.
Hi all, I've been doing a ton of research this morning and it looks like outside of this thread, ocmp34 is getting a pretty bad rap for underpaid tasks. I'm working to remedy this by automating what we can so we can raise the pay for the tasks we absolutely need to be completed by mturkers, so stay tuned - it's important to me not only to get good results but to be as fair as I can be. Thanks, ocmp34
Thanks a bunch for actually caring about that! It is rare to get a requester that cares if we are being paid a decent amount. Many think that we can live on $1 an hour. In today's world you can't even buy a loaf of bread for $1. So, at least I, appreciate that you care about paying us a decent wage. Thanks for keeping us informed!
I don't really agree with this. But the pay of each of the batches is very uneven for me, unless i'm doing something wrong. The main batch is about $5/H while the other 2 batches pay over 3x as much although short lived. I'd guess the people complaining about the pay are only doing the main batch. Perhaps it would be better to even the pay between the batches a little? But really i'm happy with the way it is already.
What if I separated the categorization (for the first task - find the homepage) into a separate task? This would speed up the HITs for which there's no homepage, which is over 50% of them. Or I could add that step to the retrieval task instead of the homepage task.
I do not know if I would toss it in the retrieval. That task is long enough as is. As its own you could probably do it. You could half the cost of the first then, maybe. It is quick enough to find a homepage. It is longer to wade through the different types of organizations there are and click the right bubble. Then again the second task would be the same thing as the first, unless you put the link to the website into the organization type hit.
Choosing Organization type takes 3x to 4x time than finding homepage. Last batch i have noticed category changes as per IRS. On the first hit i got blinked it going to be harder to select the exact organization class from this list. After several minutes returned that one and stopped working instead of brute forcing. Messaged two more turkers, they do came with the same reason. It would be easy to classify if you give at least 10 example organization names for each category or ~make an help thread separately to clear doubts. It's vital for us to provide more accurate organization type for upcoming 1M hits with ease and to eliminate felling being underpaid.
Hmm, I'll work on clarifying these categories. I'm biased here but I much preferred my categories to the IRS ones, which I feel are unnecessarily vague. I'll work on adding some examples to the instructions today. I've also separated the homepage task and the categorization task (the latter gives you the homepage found in the former). This should allow you to quickly run through the homepage task, netting you more $$, while also paying you a bit more for the categorization task.