### Managing Research Software Projects # Sharing
Always be wary of any helpful item that weighs less than its operating manual.
--- Terry Pratchett --- # What problems are we trying to solve? - Re-use - "Paywalls are the raised middle fingers of academia" - Credit - "A hundred thousand users doesn't get you tenure" - Privacy - "What can be mis-used, will be mis-used" --- # Scope - Discussion of publication applies mainly to academics - Commercial reports are often confidential --- # Software licensing - Standard file #4 is `LICENSE(.md)` - Who can do what with the project's software - Use a permissive license for software - The GPL's virality has proven to be unnecessary - MIT License: anyone can do anything, you are not liable, they have to give you credit - Hippocratic License: all of the above plus do no harm - *Do not write your own license* --- # Everything else - Use Creative Commons (CC) for data and publications - CC-0: anyone can use it any way they like - Best for raw data - CC-BY: must cite you as source - Make sure `CITATION` is up to date - CC-BY-NC: need your permission to use commercially - Makes institutions happy - *Do not write your own license* ---
As project lead
- Choose licenses - Check with your institution - Update `CITATION` after each publication --- # Data manifests - One per directory - Describe the software used under Sources ``` What is this? Source(s) Last modified Contact Spatial/temporal applicability Variables - Header - Datatype - Description ``` ---
As project lead
- Check manifests against actual datasets - A script that checks Markdown headings against filenames is a useful addition to your CI - Decide which results need to be in version control and which can be regenerated - Add `.gitignore` to `results` directory to ignore certain files --- # Publishing - Put open access versions on the web - Preprint servers like arxiv.org - Your own website - Give every report a DOI - Zenodo does this for free - Give datasets and software releases DOIs as well - GitHub-Zenodo integration makes this easy --- # Make yourself citable - People change names, institutions, email addresses, genders, titles… - Get an ORCID and include it in all publications ---
As project lead
- Get DOIs for all publications, datasets, and packages Make sure they're included in `CITATION` - Ask all new contributors for ORCIDs - Submit reports to preprint servers - Better yet, publish in open access journals --- # Privacy and security - Privacy: don't collect information - Security: safeguard what you do collect - Laws are complex, inconsistent, and constantly changing - *Ask an expert* - And get the answer in writing - *Always ask those affected for permission* - "Nothing for us, without us" ---
How are you and your work identified?
1. Search for your reports online: how accurate and complete are the results? 1. Do it again in an anonymous browser tab: how much is now inaccessible? 1. Search for yourself online: how accurate and complete are the results? ---
Check your own citations
1. Are your datasets and packages citable? 1. Do your own publications cite them? 1. Are your contributors citable and cited?