Update community roadmap

[Peter-Metz]

The current roadmap is due for an update (because we have accomplished most of the stated goals in the initial version!). Here are some possible items for a new roadmap:

• Consider and approve proposals for PSL initiatives
• Transition the PSL Meeting series to a PSL Webinar series
• Build a subsection of pslmodels.org with PSL model demos
• Expand the PSL Catalog to include more models across more policy disciplines
• Assemble a Board for the PSL Foundation

Do others have changes/additions to this list?

cc @MattHJensen @hdoupe @jdebacker @rickecon @towashington @MaxGhenis @andersonfrailey @chusloj

[chusloj]

@MattHJensen Would the list of citations for PSL models fit in here?

[jdebacker]

Have a PSL model presentation at PyData Global 2020? could be any model(s), but just aim for some representation.

[jdebacker]

Another suggestion: build out an integrator model for the tax models to produce a broad set of analysis for tax proposals.

[rickecon]

I think PSL-Infrastructure should aggregate project citations that are collected in each project. I think the best way for each project to record citations is to have a citations.md Jupyter Book markdown chapter of the project documentation. These markdown .md files have references that look like the following,

{cite}`ReferenceTag`

which ReferenceTag refers to the tag on a BibTeX entry in a file that is usually located in the Jupyter Book documentation in the following path:

jb_book_repo_main_dir/jb_book_sub_dir/_bibliography/references.bib

It would be nice to have a GitHub Action in each project repository that collects the appropriate BibTeX entries and adds or updates them to the PSL-Infrastructure main aggregate BibTeX file upon commits to the master branch citations.md file.

We had a nice discussion about this in today's PSL Community call.

[jdebacker]

I like @rickecon's idea above. It's probably worthy of it's own issue to flesh out implementation.

A couple thoughts: 1) It may be least duplicative if the PSL-Infrastructure repo were the only repo with the GitHub action to look for new references. It could crawl the repos for all cataloged models and update the citations page on PSLmodel.org. This would mean one GH action instead of one for each PSL repo. It's probably fine that this GH action is triggered by a clock - e.g., running once or twice each day - rather than a specific action like a merge to master. 2) To make (1) more feasible, it would probably be helpful if all references.bib were placed in the same relative path. Since PSL repos have different directory structures for their docs, perhaps the references.bib files goes in the top directory (similar to PSL_catalog.json, which @Peter-Metz already has a script reading each day). Any Jupyter-Book documentation can just point to that references.bib file to create it's own html references page if it wishes to.

[rickecon]

@jdebacker It is important to note that PSL-Infrastructure does not want the entire references.bib file of a project. It only wants the subset that contains the relevant citations. In other words, a project's references.bib will contain many references to other relevant work. The subset of those references that are citations that use the particular project will be those referenced in the citations.md page. We should have PSL-Infrastructure updated upon commit changes to citations.md in the master branch of the particular project.

[jdebacker]

@rickecon writes:

It is important to note that PSL-Infrastructure does not want the entire references.bib

Correct. I should have noted that a project could have a references.bib for work they cite and a citations.bib for references to places where the model is cited. All references in citations.bib would be cited by PSLmodels.org.

[Peter-Metz]

@rickecon @jdebacker @chusloj thanks a lot for your suggestions here. I've opened #185 to replace roadmap.md. Also, I've opened #186 to continue the conversation about aggregating project citations.

[Peter-Metz]

Closed with #185