What do you think about alternative ways to earn BOINC credits ?
In our interview session with CSG (Citizen Science Grid) admin Travis Desell, he asked us:
- what we think about when there'd be another way to get credits in his BOINC project ?
In short the idea was like:
- BOINC users can watch videos/images and do certain tasks on them and get BOINC credits rewarded
Here is direct feedback of Travis after reading this post:
- "I don't want to directly award credits for watching video/classifying images. I think that would screw with the BOINC ecosystem too much. Rather I was thinking about giving some kind of bonus to credit earned through computing (5%-10%) for users who watch videos and classify images."
Here are some quick excerpts from the paper:
"there are too many images for ecologists to manually inspect in a reasonable amount of time, with over two-million images currently collected and an estimated two-million more to collect each summer."
"who wish to help the processing of data, to help catalog video for a number of sub-projects with the goal of producing machine learning algorithms to automatically detect different species and their behaviors."
"it is therefore the intention of the authors to keep a log of any such events where respondents are given an external reward for providing data and comparing the bounding boxes with those of citizen scientists who are providing their time and effort without external factors."
What do you guys think about this novel method to get BOINC credits ?
The audio file with the interview session and some initial feedbacks (how is this checked against bots, cheating, ...) about it are available soon here
2017 Feb 14: add Travis' feedback above
I posted on the Zooniverse forums a few years ago about earning GRC for the work done, but the admins had considered points systems before and felt that gamification would lead to poor results; they want people doing the science that want to do it, not as some kind of Mechanical Turk. I dont necessarily agree, but humans are more fallable than computers, and in a race for riches may start taking short cuts I suppose...
With enough redundancy to make sure you sort out invalid results I think this is a good way to earn some GRC. You always have to weight the results against the benefits. If you gain nothing but the "prestige" of doing this manual work I don't think they would have a very large user base. I've looked at their Wildlife@home project and it's a positive project, but it didn't give me personally anything to sit and watch this for an hour.
If I had gained anything, a small token in GRC for example, it would give me a reason to help out, and the admin doesn't have to pay this out of his own pocket since it's the Gridcoin network that rewards it. The admin can of course rain on the users that does the work to.
Im on your side, Im just saying we dont want to spam projects with bad results.
But a Mechanical Turk on Gridcoin would be cool.
It certainly would be pretty cool to distribute mechanical turk tasks through BOINC.
Would it be possible to distribute GRC to existing mechanical turk websites though? Perhaps that could be a way to get GRC into the hands of many users & perhaps even accepted as payment for tasks on said websites?
It's a shame that they don't want this to occur, it could potentially drive their volunteer rate up significantly.
Its a while ago, they could have changed their opnions.
We should ask them again then? :D
I had participated in various zoouniverse projects. It could be great to be able to link this to Gridcoin.
Besides, I am sure that a lot of ZooUniverse fans will be great at BOINC and Gridcoin, and vice versa.
There is already an anti-cheat system like boinc tasks. I do not see what administrators can lose by setting up this system.
It may be worth it to ask again !
Think it's a good idea, a parallel program would be best, with the existing methods used, until all or most of the wrinkles can be removed. So you can verify work, but you don't become the only one doing it.
I like the idea, because in the past i watched the videos of his CSG project Wildlife@Home. Through the rewarded boinc credit there could be the incentive for some persons to watch this videos and gain some GRC.
I think that it's fine to assign RAC to manual processes like this, if it is verifiable/non-cheatable (perhaps captcha inbetween manual tasks).
This pretty much enables rewarding manual citizen science work.
I don't know about a 5% modifier to active RAC though (if that was the idea), I think fixed RAC rewards would be better.
linking to someone's post on reddit:
"There are many opportunities for citizen scientists to assist in the analysis of the huge amounts of data collected"
see here