The masterful Brooke Simmons (of Galaxy Zoo fame) kindly shared a script of hers that takes the classifications from a project and groups them based on strings of classifications that appear to have been done by the same person. When a volunteer is logged in, the user name gets associated with his or her string. But the awesome thing about Brooke’s script is that it allows us to get a sense of what volunteers who are not logged in are doing.
I ran Brooke’s script on Season Spotter and found that in addition to the 2,136 registered volunteers who have tried out the Questions part of the project, approximately 3,436 additional not-logged-in volunteers have tried it, too. For the Image Marking part of the project, 1,300 registered volunteers have tried it, and about an equal number (1,303) of additional unregistered volunteers have also tried it. Pretty neat.
What’s also cool is that I can take the output from Brooke’s script and make the same sort of visualizations I made a couple weeks ago. In these visualizations (called ‘treemaps’), each square represents a single volunteer. And the size of each square grows with the number of classifications each volunteer has done. So if one square takes up 1% of the whole treemap, then the volunteer who is associated with that square has done 1% of all the classifications. The colors assigned to each square are random, except that I colored all the not-logged-in volunteers a dark gray so that they stand out. Here’s what we get for Season Spotter Questions:
A couple things really stand out. First of all, I was really amazed by how many classifications some of our unregistered volunteers are doing! My assumption has always been that most not-logged-in volunteers come to the project for the first time, try it out a little, and then leave. But that’s not true for a handful of dedicated, but anonymous, volunteers. For Season Spotter Questions, the #5 and #6 contributors are both unregistered, and there’s a regular occurrence of unregistered volunteers at all contribution levels.
It makes me wonder a bit about why people don’t register. Is it too much of a burden? Do unregistered volunteers simply have a preference for as much anonymity as possible? If you are one of these unregistered volunteers, please do take a moment to comment (anonymously, of course) at the bottom of this post and tell me why you prefer to contribute as an unregistered volunteer.
The second thing I notice in this treemap, is the pattern in the bottom right corner. The squares are sorted such that the volunteers doing the fewest number of classifications (1) are all tucked together in the bottom right. I’ve also grouped together unregistered volunteers. So reading from the bottom right of the treemap towards the center, you see a group of registered 1-classification volunteers. Then a band of unregistered 1-classification volunteers. Then a band of 2-classification registered volunteers. Then a band of 2-classification unregistered volunteers. And so forth. It’s notable that the last dark gray band is quite large. It represents those unregistered volunteers who have done 5 classifications. And then the pattern seems to disappear.
What’s going on? Well, after an unregistered volunteer has completed 5 classifications, Zooniverse pops up a box asking the contributor to register or sign in. The volunteer can also choose to close the box and continue on unregistered. But it seems that doesn’t happen so often. Instead that pop-up box is either encouraging people to register. Or it’s causing people to say to themselves, “okay, I tried that project. I think I’m done here.” I don’t have data to determine which of these two things is happening, though.
When I create a visualization of the Image Marking side of the project, I get a similar treemap. You can see that 3 of the 5 top contributors are unregistered(!), and that unregistered contributors are represented at all contribution levels. You can also see the band-of-5 pattern here, too.