Pen’s reflection blog post:

We started the discussion around what makes a good booklet/guide on doing good science. We agreed that there are many broad good practices (“Good science is open science, closed science is not good science!”), such as open source code, open access publications, and open data. However, it becomes difficult if you want to gear the guide to specific scientific fields. The concerns of an astronomer might be different from a biologist, even the tools they use are different. So if the guide is too specific, then it might not get scientists in other disciplines interested. Eventually, we came up with a bullet point list of a rough outline for a broad (not discipline specific) guide below.

Brigitta

had a great idea which is a “case studies” section. This puts a lot of the practices into context and show that open science can be good/successful science, contrary to what some people might say.

As part of our wide ranging discussion, we talked about the institutional and cultural issues around doing science. The scientific method is fundamentally about building on past work and making your results for others to build on. Don’t forget that as scientists, we are always “standing on the shoulders of [other scientists]”! Unfortunately, the realities of academia today (tenure review, credit attribution, the publishing process (e.g. Elsevier!!!), etc.) often incentive us to do closed science. How do we solve this fundamental betrayal of science? This is a big problem that will take more than one person to solve. However, it is important for all of us to remember that we are ultimately responsible to science not politics and keep that in mind when we make decisions. For example, climate change is a huge global issue that will take an equally huge cooperative effort to solve, and it has taken decades and we are still not close to solving it. But that doesn’t mean we shouldn’t work hard on it! It is also true for advocating open science.

That said, for making a guide to open science, talking about such deep issues might lose a reader’s interest because they might want to see specific things they can do, rather than a deep discussion. How do you strike a good trade-off between the two? This is a question that we leave to the reader of this Etherpad. Finally, I propose that while the big problems can be overwhelming, one thing we can do is to think about the baby steps we can take towards more honest/responsible science. Maybe every month you can pick one small good practice to put into action in your research workflow.

Outline for general guide for doing open science and good practice

What is open science? (general)

open literature review

  • https://pad.okfn.org/p/opendiscovery-mozfest

cultural/political questions

  • Science doesn’t operate in a vacuum
  • Practices/norms in different research areas?
  • how can you implemenet new ways of working?
  • Problems with attribution/credit

Publishing open access

  • What does open access really mean -> definition.
  • Maybe talk about copyright and licenses here?

Open data

  • What does open data really mean -> definition.
  • Licenses.

  • Where to submit data

    • Repositories

  • Long term storage/retrieval issues

    • Proprietary formats both in hardward and software

    • Paper degrades

    • Solutions?

Tools for doing open science

  • forced/planned obsolescence

    • what if Google Drive or GitHub shuts down?

  • “Open source” replacements for commonly used proprietary tools Very general list of collaborative office tools:

Some things for which I have sadly not found good replacements:

* Apple Keynote and Microsoft Powerpoint
* YouTube/Vimeo
* Flickr

What about field-specific tools? Need list of good examples!

  • Matlab, SPSS, Stata, etc. replacements: Python, R

Reproducible research

  • Code
  • Non-code methods
  • Anything else? Documentation?

New ways of disseminating research

  • Not everything has to be a paper
  • exectuable papers which link software, data, and text

Case studies

  • Success stories

  • Projects that went from closed to open

    • saw changes in the number of papers published

    • more successful junior scientists

    • more diversity in institutions that participate

  • Arxiv.org is a great example of open access publishing, and it works well with publishers, too.
  • SAO/NASA Astrophysics Data System adsabs.harvard.edu for astronomy and physics papers
  • Genbank
  • Astrophysics Source Code Library/Zenodo

  • How did the Human Genome Project make science more accessible?
    http://www.yourgenome.org/stories/how-did-the-human-genome-project-make-science-more-accessible

  • Public vs Private Efforts to Map the Human Genome http://www.knowledgene.com/part2.html

  • https://github.com/ehec-outbreak-crowdsourced/BGI-data-analysis/wiki

  • https://royalsociety.org/topics-policy/projects/science-public-enterprise/report/

  • Lifting proprietary period for the K2 mission compared to Kepler, effects on author stats, accessibility (no write up on it yet, just a few figures: https://github.com/barentsen/exoplanet-charts/tree/master/publication-stats)

Checklist for doing open science (easy to hard)

  • Publish papers open access [ ]
  • Publish underlying data [ ]
  • Publish code/analysis scripts [ ]

  • ???

  • Software use in astronomy (survey by Tollerud&Momcheva, 2015)

  • How python became the No1 language in astronomy: 1 and 2

  • addition Bianca: also check out Jon Tennant @protohedgehog initiative for an collaboratively put-together Open Science MOOC - might be worthwhile to collaborate / benefit from each other’s ideas/materials)
  • also added as asked by Pen: Utrecht Summer School Open Science & Scholarship (this also has Jeroen & Bianca’s contact details)