by Jonathan Haehn (from a CC talk, October 2020)
Open data & material
…Fully reproducible science
Hello, Sci-hub. RIP Aaron Swartz.
Ex: Paper on your department site and code on Gitlab. Version control, permalinks to specific code versions. Makes experimenting easier.
Ex: Paper on your site, code on Gitlab, data + files on Zenodo or Dataverse.
Version control + uniqwue identifiers (HDL, DOI). Zenodo supports DOIs for code too.
Makes running similar experiments on the same data easier.
Use permanent IDs for models, datasets, protocols. Zenodo does some of this; other specialized groups like Datacite and Protocols.io do as well.
This is reproducible science! Makes replication possible. For full replication cleanliness, aim for something comporehensive like the Whole Tale.
(Is this for bio specifically? like Addgene, reagent IDs, making sure copies of any custom materials / environments are copied + stored + accessible?)
For instance: stored on OSF.io
Use a free-content license to support access, reuse, and redistribution forever.
via aspredicted.org or others
avoiding the forking-paths and other fallacies in understanding the implications of research, possible parallels, and the importance or impact of carrying out further successful replications.
Use collaborative tools and interfaces (and annotatable documents)
Implement collaboration policies in classes
Encourage sharing at public conferences, fixing issues in open source packages, and using free licenses to share work.
by Jonathan Tennant (from a TG chat, March 2020)
Inspect the research process in its entirety (materials, data, and code are examples of prerequisites for this)
Redistribute copies of research outputs so you can help your neighbor.
Reuse any output of published research as you wish for any intended purpose and without restriction
Adapt research and share adapted processes and outputs, so that whole communities can benefit from changes
Collaborate through revision and forking, so that research becomes a truly public endeavor
Capture and preserve provenance: acknowledge all manner of contribution to research
Make evaluations of merit or significance that are explicit, transparent, and selectable.