Musings on scientific discoveries, luck, and being prepared
I’ve been thinking about the relationship between luck and good preparation lately. We have published a paper on the discovery of three plus one small planets around a small star recently, and we basically did the work for that in one day. The way that worked out got me thinking.
From the outside, what happened was this: Data from the latest Kepler-K2 campaign was publicly released. My student Rob Wells looked through some light curves and found something that looked like a triple-planet system. We started writing a paper about it, and had it finished after a bit more than one day of full-steam-ahead work. We submitted it to the journal (MNRAS Letters), and the arXiv, in the early evening. Got some helpful and thorough comments from the reviewer, edited the paper draft, got it accepted, done. From the outside, I imagine that this looks like being quite lucky.
From the inside of our group, there was actually much more work involved. My student had spent a lot of time over the past half year to get a data crunching pipeline working, which did quite a number of things (detrending the telescope data, doing a rough search for transits fully automatically, producing image files that allow the human user to quickly browse through stuff and sort for things that look like interesting systems). I had previously worked on spectral energy distributions of stars for another project (the YSOVAR project and my paper in the project series), and had some experience with proper MCMC fitting of transits from yet another project (my work on X-ray transits). All this preparation meant that we could directly jump into the relevant analysis for the paper and write it up with a collaborative online latex editor really quickly. We had to determine the stellar spectral type, produce proper transit fits (which eventually revealed the fourth planet candidate), and perform a stability analysis for the spectrum (in which our colleague Chris already had some experience, so he got that up and running really quickly). A lot of this came down to having all of the necessary experience in our small team. It would have been even better if we had all of the necessary tools and code somewhere easily accessible in one place – that’s what we have done now with a shared code repository for our group, so that other discoveries should be publishable quite quickly from now on. I’m really happy about this.
Another thing I like about how all of this worked out: My student came to me with his interesting dataset on a Friday afternoon. We did some of the work for the paper on that Friday afternoon, then all went home for a work-free weekend, and came back on Monday to put a full 8 hours of work into the project. Submitted to arXiv just before the daily deadline (18:00 in our time zone). And everybody got to go home on time. I was a little bit worried that maybe lots of other teams would be working through the weekend to publish interesting results from that K2 data release. But it turned out that we were the first people to publish anything from that campaign at all, and the next papers (about different star systems) came out more than 2 weeks after ours. This means that, even in the exoplanet field, one can do one’s work during normal work hours, Monday to Friday, and not be scooped left and right. This makes me much happier about recommending the pursuit of an academic career to my students.