Better science for future us

“Better science for future us” is science that is more efficient, reproducible, open, inclusive, and kind. There are growing examples of better science in environmental and Earth science, and beyond, including from the Ocean Health Index team: Our path to better science in less time using open data science tools (Lowndes et al. 2017).

Slides

Better science for future us

Video stories

Our Openscapes journey towards improving the Alaska Marine Mammal Stock Assessment Report process (19 min; slides). A SAR story from the Alaska Fisheries Science Center (AFSC) Marine Mammal Laboratory (MML) Stock Assessment Report team - Brian Fadely, Josh London, Nancy Young, Tony Orr, Erin Richmond, Rod Towell, Millie Brower. Presentation to Fall 2024 NMFS Openscapes Champions Cohorts.

Better science for future us (22 min). Julie Lowndes shares part of her path to open science with the Ocean Health Index and Openscapes Champions success stories from 3 NOAA Fisheries teams 1) at the end of their cohort; 2) 2 months after their cohort; 3) several years after their cohort.

First Forays into the Cloud (19 min), by Aronne Merrelli to the 2024 Cohort of NASA Openscapes Champions. Aronne is Assistant Research Scientist at the University of Michigan College of Engineering (Climate and Space Sciences and Engineering) and a 2023 NASA Openscapes Champion.

My path to open science (7 min), by Stefanie Butland, Openscapes Team Member

These slides were contributed by guest instructors.

Empowering transformational science - Dr. Chelle Gentemann

Here we also introduce the Pathways concept that teams will develop throughout the Champions program. The Pathway is based on Table 1 in Lowndes et al. 2017, and helps teams deliberately identify data workflow practices and next steps to facilitate efficiency and open practices in terms of reproduciblity, collaboration, communication, and culture.

Pathways to better science in less time

Figure 1 of Lowndes et al. 2017 shows that open data science tools increased the ease of reproducibility and the ease of collaboration for the Ocean Health Index (OHI) team. But it was not the tools alone - it was the process the team co-created and prioritized.

Create space

Creating space means committing to synchronous collaboration convenings to learn and teach each other together.

A critical first part of this was prioritizing time (which included getting buy-in, showcasing). Then, this meant that the team could focus time on:

  • Getting comfortable talking about data/workflows
  • Building trust (to share imperfect work)
  • Recognizing that what we invest incrementally will have large dividends in the future
How we did it: OHI Team

The OHI team started having regular Seaside Chats 1x/week where they discussed filenaming, code review, standard operating procedures and documentation, and much more. We prioritized regular social hours, that we scheduled during work hours so that everyone could participate no matter their other outside-of-work responsibilities.

Create place

Creating place is critical for asynchronous collaboration. It is a place for code, shared practices, resources, conversations. Critically, this involves making sure that everyone on the team is comfortable contributing through these channels. This means both with the technology, and the culture of the team.

Asynchronous collaboration often requires some form of version control so the team can understand what documentation, data, code, graph, etc., is current. Specific places depend on the tools and platforms used in a given organization or research group. These can include GitHub Organizations, Repositories, Issues, Projects; Google Drive Folders, Docs, Spreadsheets, Slides, Calendars; Slack Organizations and Channels; Teams, Sharepoint; JupyterHubs; etc.

Find the common

Through creating space and place, teams will find the common workflows, tools, skills that they already have and need to do their work.

Documentation was a key part of this. And, writing documentation “for nobody” is very hard, and it’s a huge task. We prioritized documentation based on Onboarding and Offboarding: for our future selves first, and then future us.

How we did it: OHI Team

The OHI team asked, “how can we make sure everyone can participate as they need to?” Not everyone on the team needed to learn to code, or to use GitHub in the same way. We introduced new software sparingly, and helped each other learn. This included initial setup as well as follow up and practice. We leveraged existing habits & resources - within and beyond our team. Open communities were a big part of this learning. Through this we were able to distinguish data preparation (tidying) as distinct from our science, and make this actionable by shifting to smaller modular code to combine for different reports/audiences.

Shifting incrementally

Shifting workflows takes time, particularly because it is most often done while also meeting existing deliverables and deadlines. It requires changing behaviors and habits, which takes time, and is messy, and really depends on the trust built with the team.

How we did it: OHI Team

The OHI shift to using shared Google Docs, R and GitHub was motivated by necessity, reimagined by possibility and community, and done incrementally. It was an investment over years, but the enduring payoff has been huge.

Some key points to discuss from Our path to better science in less time using open data science tools (Lowndes et al. 2017): reproducibility & communication enabled by open tooling; and shared practices are useful beyond shared projects.

Ocean Health Index (OHI) team’s transition of code and shifting communication to GitHub

If you’re interested in more overview of the OHI setup, see this 2017 talk (25 mins): OHI Better science in less time or this 2021 Plenary at Better Science for Future Us at the inaugural Society for Open, Reliable, and Transparent Ecology and Evolutionary Biology SORTEE conference (30 minutes) (video).

Reproducibility & communication enabled by open tooling

RMarkdown/Quarto to reimagine data analysis and communication. RMarkdown/Quarto combines analyses & figures together, rendered to your reporting output of choice.

How we did it: OHI Team

An example: http://ohi-science.org/betterscienceinlesstime

Shared workflows not only useful for shared projects

It’s about increasing efficiency and reproducibility and open science. But it is also about increasing participation and inclusion. Consider how inclusion and equity show up in your daily practices. How you work and onboard others to your projects is a DEIA issue.

How we did it: OHI Team
  • OHI team: we identified as a team & prioritized helping each other
    • We work on many different projects
    • Use same workflows, share feedback, can think together across projects
  • Shared conventions reduce friction & cognitive load
    • Common ground, easier to talk about, easier to ask for help
    • You don’t need to design everything from scratch

Impact of shifting to open science

Here are a few examples to showcase what is possible and being done in environmental science.

How we did it: OHI Team

Ocean Health Index has produced annual reports for 12 years (2012-2024) and counting! There are many impacts of this, but at a glance:

  • In 2012, it took 30 people 4 years and several $M to complete the 1st annual OHI report. It would not continued if it costed so much time and $ each time.
  • In 2024, it takes 3 masters students in 3 months and $200K.

This is possible because of the incremental investment to make it reproducible, efficient, documented - focus on onboarding. Impacts mean that students and team can focus on new questions, making sense of results, and applications from this, not the assessment itself. And, it accelerates the rate at which other teams can get to this cost savings, since OHI is an example that it’s possible, a working open example.

Further resources

Not so standard deviation podcast

Parker & Peng, http://nssdeviations.com. Great discussions about data concepts and “in the wild”. Start with Episode 9: Spreadsheet drama

Practical computing for biologists

Haddock & Dunn, http://practicalcomputing.org. Software & computing concepts already on your computer. Start with Chapter 2: Regular expressions