The phoenix renews her youth
only when she is burnt, burnt alive, burnt down
to hot and flocculent ash.
Then the small stirring of a new small bub in the nest
with strands of down like floating ash
shows that she is renewing her youth like the eagle,
The science team has been using the unofficial name “Phoenix” to refer to the newest version (3.0, but who’s counting?) of the Milky Way Project (MWP). One year ago, we took the project offline, running out of data after collecting more than 3 million classifications made by tens of thousands of Zooniverse volunteers. It was the first time since our initial launch in December 2010 that MWP was not among the ever-expanding list of active Zooniverse projects.
Those of you who had followed and participated in MWP for all those years may have been wondering if we were ever coming back…and to be honest, I wondered, too! There were a couple of major challenges to overcome if Milky Way Project was to “rise from the ashes” and fly again.
First, we had an enormous backlog from three years’ worth of classification data to sift through before we could determine the future direction of the project. Our first data release, published in 2012 and presenting catalogs of over 5,000 bubbles, was just the proverbial tip of the iceberg, the first year of classification data. We always knew we could improve upon our bubble catalog, and we also wanted to study some of the other classes of interesting objects identified by MWP volunteers, but as is always the case in science, you have far more good ideas to pursue than time, money, and personnel to pursue them with.
The second challenge exacerbated the first. Our project lead, Dr. Robert Simpson, resigned from his position at the Zooniverse in March 2015 to pursue a fabulous new career at Google. Rob asked me to succeed him as project lead, and I accepted this responsibility somewhat reluctantly. Not for any lack of enthusiasm for MWP—but because Rob’s were very large shoes to fill! From the beginning, my role as “science guru” had been to provide high-level science direction and interface design ideas for MWP, produce the infrared images used in the project, and assist the science team with the interpretation of the resulting data. I knew little of the technical details powering the (ever-evolving) Zooniverse “back-end,” and I relied upon Rob to crunch through the database of classifications to provide catalogs of objects for science analysis.
Getting myself up to speed on everything I would need to know to lead all aspects of MWP was a daunting task, on top of my regular duties as an Assistant Professor of Physics and Astronomy at Cal Poly Pomona (juggling multiple research projects at once, teaching classes, mentoring students, and all that good stuff) and my most important job, husband and father of two girls (my younger daughter was born two months before I officially took the reins of MWP—what timing!).
MWP: Phoenix would therefore never have come about without the exceptional energy, dedication, and talents of Tharindu K. Jayasinghe Arachchilage. Tharindu transferred to Cal Poly Pomona from a junior college in his native Sri Lanka in Fall 2015, and joined my research group soon thereafter. Since Tharindu was already an experienced Python coder and familiar with database management, I handed him the set of 2 million unanalyzed MWP classifications, pointed him to Rob’s original paper, and asked him to see what he could do with them…
In just a few short months (all the while carrying a full load of tough courses required of undergraduate Physics majors, and working two other jobs on the side), Tharindu had his codes developed to the point that he could query our database of classifications, pull out different kinds of objects (for example, bubbles versus star clusters), and identify “clusters” where multiple MWP volunteers classified the same kind of object in the same location. We were in business! I hired Tharindu as a full-time MWP researcher over this past summer. We developed the new interface using Zooniverse’s amazing Project Builder, ran my old image-making codes to produce 77,000 new jpeg images (the largest set to date, and it overwhelmed the Project Builder’s uploader until Tharindu found a workaround), and continued to test and refine Tharindu’s cluster-finding algorithms. For that last task we were greatly assisted by two other students, Don Dixon from Cal Poly Pomona and Jose Velasco from Citrus College, who spent many hours visually reviewing the new MWP bubbles overlaid on the original science images, comparing them to other bubble catalogs (including the original MWP bubbles, of course) to see if we were missing real bubbles, or finding fake bubbles (we call them “fubbles”, and you’ll be hearing more about them later) that are just random, “smoke-ring” patterns in the images.
To accomplish our immediate science goals, we need 2 million new classifications within the next 6 months—so Tharindu will have sufficient time to analyze them and write up the catalog paper before he graduates. We need all the help we can get.
If you are just discovering the Milky Way Project, welcome! If you are a return volunteer, welcome back, and thank you so very much for your continued help. Happy classifying!
—Dr. Matthew S. Povich, Upland, CA USA
Next time: What new science do we hope to get from the new MWP?