Next steps for NRP Development

Short term foundational tasks

These tasks are prerequisites for, or may facilitate, more specific tasks below.

1. Implement "automated builds" with CircleCI, like the UK platform's Travis builds:
   • This will ultimately be necessary to support a few tasks, including 2.3, 4.1, 4.2, and 5.2 below.
   • Allows for the use of Jekyll plugins that are not allowed by Github Pages.
2. Establish naming convention for CSV columns, including disaggregation rules
   • This will be directly necessary for 4.1, 4.2, and 5.2 below.
3. Decide on minimum requirements of data management:
   • This will inform the tasks under #2 below, for improving the data management experience.

General initiatives and tasks

These items will improve the user experience across the various user groups (identified below) of the US National Reporting Platform through increased automation, interoperability, user-friendliness, and functionality.

1. Developers: Those tasked with the technical steps necessary to build and deploy the NRP.
   1. Easier forking process:
      • Run through the forking process again to identify pain-points or missing documentation
   2. More opportunities to customize:
      • More ability to customize the site without editing the most complex files - ie, more options set in _config.yml.
   3. Documentation roundup
      • Provide any missing documentation, and maybe look into a service like readthedocs.org
   4. Modularity of systems (may be a long-term goal):
      • Split certain sections of the platform into swappable object-oriented modules, which can be maintained in their own files (or repositories?). First example might be modularizing each "indicator data source" and each of the data visualizations.
      • Implement the platform as a “core” product which can be used without forking it.
      • Move the data and metadata into separate repositories.
2. Data managers: Those tasked with updating the data and metadata for each indicator in the NRP.
   1. Improvements to prose.io
      • Fork prose.io to make improvements - could be a joint UK/US project
   2. Investigate alternatives to prose.io, such as jkan.io and octopub.io
      • Also possibly joint UK/US endeavour
   3. Folder-based disaggregation
      • Mostly for subnational data management - keep each state's data in a separate folder for easier maintenance
3. Administrators: Those tasked with accepting/rejecting the changes that are proposed by the data managers.
   1. Automated updated dates
      • Fully automate the goal pages to depend on dates in metadata, rather than manual changes
   2. Folder permissions based on Github usernames
      • Indicate failed validation of pull-requests from unknown Github usernames
   3. Add new data and metadata validation rules:
      • Help administrators by not allowing data managers to commit mistakes to the repository.
4. Third-party: Those implementing completely separate systems which need to consume the data and metadata of the NRP (such as the UN).
   1. Tabular data package output:
      • Possible first step towards SDMX.
   2. SDMX output (may be a long-term goal):
      • Eventually implement indicator data and metadata as SDMX output.
5. End-users: Those who are browsing the NRP to see the status of the indicators and view the visualizations.
   1. Subnational data
      • Gather the data and implement the user interface for subnational
   2. Filtering of disaggregated data
      • Implement facet-style disaggregation filters, like UK platform
   3. Translation audit
      • Make sure that all hard-coded language can be (and is) translated