Ready, set, TEACH.

The semester starts in one week’s time. Before the break, I was furiously working, working, working to get the Open Science and Reproducible Research course at least skeletonized. This week will be all about sorting out tasks I need to get done before my trial by fire   first time teaching an entirely new course begins. My plan is to make this course VERY discussion based and open form (hahaha of course it’s open form)- essentially hit the students with reading materials which we will discuss together in class, and activities we will do together, and have some time in each class period devoted to supported work on our real open data set. The class is small and the students are pretty much all known to me so I think conversation should come fairly easily.

The folks at the Mozilla Science Lab have had us fellows work on a number of exercises to help us focus on our respective projects. Since my stream-of-consciousness tends to come out best here on this blog, I’ve decided to hybridize these assignments with blog posts. So to start- here’s some reflection on what I’m doing and why.

Challenge

Science does not end up in the hands of the people that need it. Within academic science, data and analytical techniques are not shared freely, making it difficult or impossible for other scientists to reproduce or build on their work. People working outside western academic science have even more trouble, because they typically do not have access to academic publications, the end points of much academic research. This means the people that need science most- the ones making decisions that affect human health, livelihoods, and the environment, do not have access to the information produced by scientists to help solve these problems. This problem can largely be solved by academic science moving to an ‘open’ model, where scientists use the tools and connectivity available to them through the internet to document and share all steps of the scientific process. However, academic science lacks the infrastructure to train scientists to learn to use these tools, and lacks a regulatory or reward structure that makes it appear worthwhile to change their established approaches.

Scope and scale

Closed practice is pervasive in academic science. At every level of rank and organization, the infrastructure is built to not particularly value open practice, and sometimes outright deter it. The culture of academic science re-enforces secrecy- I remember even as an undergrad, working as an assistant in a research lab, hearing conversations between grad students about their concerns that their work would be ‘scooped’ by others. There was an oral tradition where students passed down this message- that science was primarily an adversarial pursuit- you had to hold your cards close, lest your competitors use your data to solve their problems before you. These messages get reinforced as a student passes along through the pipeline and through the academic ranks. Frankly, high impact factor papers (typically in closed access journals) and grant funding are the currency of success in academia- there are few recognitions for inclusivity or reproducibility. Because of these incredibly dominant cultural aspects, I believe the key to changing the culture is through gentle shifts in regulation and the reward structure- and then aim for the bulk of the change to occur in early career scientists.

Refined problem statement

Closed practice hurts science while benefiting only a small subset of individual scientists. Open science can increase diversity and participation in science, while fostering the process of science itself by improving reproducibility, but requires training, advocacy, and a reward system.

Reflect

Most scientists agree that learning to use technology to improve the reproducibility of their work is a good thing, but there is a lot of pushback against open science in my field for two big reasons.

1. the learning curve associated with taking a whole new approach to science- it’s not trivial. With each step on the academic ladder, individual scientists have less time to spare, and approaches to problems are more and more ingrained.
2. There are risks to open practice, both perceived and real, and rewards can be difficult to quantify under conventional academic metrics. The cost: benefit ratio varies with field, career stage, institution, and many other factors.

The first factor, I feel, is fairly easily addressed. Academics are used to doing things that are hard. Offering training in open science early in their careers makes learning it less hard, and then they can follow the path as they grow as scientists. I’m less able to address the second point because these are real structural problems that are harder to overcome. I feel like we need to change the value system- how people are evaluated- in academia to tip the ratio on these cost-benefit analyses.

Brainstorm

I feel like other scientific fields are further along on the path to open science in some ways. For example, in more quantitative sciences like physics, for example, the slope of the learning curve towards using more technological approaches to science is damped, because a lot of that field is computational. Fields like physics are also known for massive collaborative projects (think the large hadron collider), meaning the expectation is already there that people working on an aspect of a problem will share their work so that collaborators can build on it. However, answering a lot of questions in organismal ecology is still possible with the work of a lone scientist, carefully making observations on her own, so open collaboration isn’t necessarily pushed as a part of a training program. I think one of the key factors in bringing open science to organismal ecology involve breaking down the hesitance towards technology I’ve observed among many people in my field. To do this, I think the best approach is to start small- show them simple, small steps they can take that make their lives easier or more efficient- be it better documenting their data, scripting an analysis so that it automatically processes observations from a new experiment, or making their contributions more easily integrated with a collaborator’s work.

Transformation

Open science has the potential to change both society and academia, for the better. It will place scientific evidence into the hands of people who need it most, from people working on more efficient agricultural systems in developing countries to people who want to learn better, evidence based ways to treat medical conditions. It will create an environment where scientists build on each other’s work, and can draw on the skills and ideas from the broader community.