Dan Foreman-Mackey

The Hack Day at AAS 225

January 18, 2015

At the last minute, my friend Adrian P-W and I found ourselves in charge of chairing the Hack Day at the 225th meeting of the American Astronomical Society that happened Jan 5–8 in Seattle. This is a job usually left to my charismatic PhD advisor, Hogg, who wrote blog posts about the previous AAS hack days (2013 and 2014; you can also find summary posts on Astrobetter). Hack days are a relatively new addition to this meeting and to the astronomical community but they are always fun and productive. This day was no exception!

Before the hack day, there were boards set up in the poster room at the meeting and a page on the Astrobetter wiki for people to list their hack ideas. There was also quite a bit of activity on the Twitter feed. Then, on the day, there were about 75 people in the room and roughly half had previously attended a hack day of some sort. The day started with hack pitches then after a quick coffee break, we got down to work. The main point of this hack day is to spend a full day focusing on a single project that you don't normally have time for but that you've been wanting to work on or learn about. The hacks are by no means required to be code-related but for practical and sociological reasons, they often are.

In the afternoon, after about five hours of hacking, we regrouped and everyone reported on what they had been up to. I don't have space to list all the awesome hacks but here are a few examples:

  1. Angus (Oxford/CfA) and Montet (Caltech) built a Twitter "troll-bot" that listened for people tweeting about air travel (optionally with the #academicnomad hash tag) and responded with an estimate of their carbon footprint. Unfortunately it looks like they have since been banned from Twitter for violating the terms of service. [code]

  2. Davenport (UW), Morris (UW), Fiore-Silvast (UW), DeStefano (UW), Holachek (ADS), and Bianco (NYU) extended Davenport et al. (2013)'s study of the gender distribution of question askers at the meeting. One key result from this year's analysis was that if the first question is asked by a woman, the gender distribution of subsequent questions is more representative of the meeting's overall gender distribution (60% M / 40% F) versus the average distribution of question askers over all sessions (80% M / 20% F). What causes of this difference? [code, data, and analysis]

  3. Partly to answer some of the questions raised by results like the above gender study in more detail, Schwamb (ASIAA), Salyk (NOAO), and Avestruz (Yale) developed a survey aimed at answering the question: "why do people ask questions?" Some results were in but a full analysis wasn't finished by the end of the hack day. I'm looking forward to hearing what they find (and I think they're still accepting responses)! [survey]

  4. Inspired by the visualizations created for this article (that she wrote), Ash (Columbia) figured out how to use Worldwide Telescope to render the constellations in 3D and view them from different angles. It turns out that constellations are just projection effects ;-). [video]

  5. For all of your music discovery needs, DJ Carly Sagan (NYC), Rice (CUNY/AMNH), and Lee (Harvard) started a collaborative playlist on Spotify for astronomy related tunes. It currently has 637 songs (two days of constant listening and Total Eclipse of the Heart is only on there about six times) from many different contributors. Add your favorite astrosongs now! [playlist, and instructions]

  6. To study gendered trends in publication practices and coauthorship, Bianco (NYU), Morehead (PSU), Wang (NYU) Li (Swarthmore), Senchyna (UW), and Lee (Harvard) downloaded about 5000 papers from ADS and estimated the genders of the first few authors using first names. Most of the hack day was spent scraping and managing the data (that's called "data science") so they didn't have time to come to any solid conclusions but they did produce an awesome dataset. [code]

  7. Rogers & Donaldson (MAST) collaborated to put together a prototype of a very lightweight visual interface to the MAST catalogs and imaging. The app is web-based but it is designed to used on mobile devices. I think that this will be useful for both research and teaching.

  8. VanderPlas (UW), Douglas (Columbia), Morton (Princeton), Peters (Drexel), and Hollowood (UCSC) worked together to develop a consistent API for time series analysis to be incorporated into the AstroML package. By the end of the hack day, they had implemented Lomb-Scargle and Supersmoother complete with unit tests and a full API specification. That's how I like to see code released! [code]

  9. Finkbeiner (CfA) hacked the IDL image-viewing platform, ATV, to show a movie of posterior samples instead of static figures. Everyone in the room gasped when he demonstrated that you can zoom, pan, etc. as you normally would while the samples continue to animate. How do we include this in papers now? Finkbeiner uses IDL and the classic email-me-to-get-the-code license but this is impressive enough work that I guess we'll have to forgive him. [code access]

A huge thanks goes to Kelle Cruz (CUNY/AMNH) and Meg Schwamb (ASIAA) for organizing the event, Debbie Kovalsky (AAS Meetings) for logistics, and our sponsors, Northrop Grumman and LSST, for the snacks and lunch. I'd also like to thank everyone who attended for making it such a fun day.

Please consider joining us for Hack Day at AAS 227 in sunny Florida! Programming expertise is not necessary and hacks do not even need to be code based. One example might be to make more knitted astronomy mascots at the Hack Days. If you have any questions about hack day, want to participate but have hesitations, please feel free to reach out to me, Kelle or any of the other organizers.

comments powered by Disqus