Proposal: Restructuring the Jupyter Community Forum categories by subproject

note: this is a non-urgent issue, so if it seems more complex than “yeah just go for it”, I’ll likely drop this and wait until there’s more time to dedicate to doing this more thoughtfully!

I propose that we re-factor this community discourse structure so that top-level sections all map onto sub-projects and major Jupyter-org structure. This would make it easier to know “what belongs as a top-level section” while roughly following the practices we’ve already been following. The steering council of each subproject would then be responsible for structuring and managing the conversation within their category (or, opting out and not having a category if they wish).

This would allow us to consolidate and streamline the forum’s structure, define some principles that make it less arbitrary to say what gets a top-level structure or not, and make it easier for users to navigate by topic. (edit: as @manics notes below, this is more about making it easier to decide what gets a top-level category, it may not actually help users navigate the forum)

Next steps

I’d love comments from folks about whether this would be a helpful way to make the forum more manageable and easier to navigate, and whether it would reduce the cognitive burden of deciding who has the authority / responsibility to add categories, and where.

In particular I’d love to hear from the Media Strategy Working Group since they’re responsible for the forum! (cc @SylvainCorlay , @andrii-i and @jake-stack chris couldn’t find a forum user for this person)

There’s an issue tracking this proposal and next steps below, I’ll focus this forum post on the content and iteration on it.

Context

Over the years, the categories of this forum have grown and evolved semi-organically as new initiatives have arisen. After a few years of this, we have a lot of top-level categories, which can be hard to navigate and hard to understand who is responsible for overseeing the content in each. For example, there’s both a JupyterHub and a Binder top-level category, even though both are subsets of the JupyterHub subproject.

I don’t believe we have clear guidelines for “what becomes a top-level category”, and I think having something like this would make it easier to make sensible decisions that don’t feel semi-arbitrary. This proposal is that we adopt a guideline like "top-level categories map onto organizational structure such as subprojects, governing bodies, and working groups.

Implementation

Here’s a rough idea of the suggestion:

  • Any subproject of Jupyter has the option of getting a top-level Discourse section (aka any of List of Official Jupyter Subprojects — Project Jupyter Governance)
  • A top-level “Governance” channel will be a place for channels for the Executive Council, the Software Steering Council, and the Foundation Board (if they choose to utilize this)
  • A top-level Working groups and special topics section can contain a sub-category for each working group, if they wish.
  • A single “General” section will have cross-project conversation (announcements, meta-conversation about the forum, etc)

Example of what our forum structure would look like

Here’s an example of what the forum could look like after implementing this proposal:

  • Jupyter Frontends
    • JupyterLab
    • Notebook
  • Jupyter Hub
    • Binder
    • Z2JH
  • … other subprojects
  • Working groups
    • Any working group…
  • Governance
    • Executive Council
    • Software Steering Council (if they wish)
    • Jupyter Foundation (if they wish)
  • General
1 Like

I think the main challenge with categories is you need to have existing knowledge of the Jupyter ecosystem to know which category to choose. A typical example is mixing JupyterHub, Jupyter server, Jupyter lab, Jupyter notebook.

My general impression (though I have no solid evidence, might be an interesting natural language processing analysis for someone!) is that the majority of new topics are from people asking for help or advice, rather than members of subprojects starting a discussion.

I don’t object to changing the categories, but I also don’t think it’ll make much difference in practice.

3 Likes

That’s a good point @manics - I think that you’re right, and also implementing this proposal won’t really change the forum structure that much, it might just mean that there are a few categories that are consolidated as sub-categories etc. To me, this proposal is about making it clearer to decide what our top-level categories are. I don’t think I have the right expertise to make a more complex proposal.

I’d be interested in seeing proposals from folks that wanted to do a deeper re-work of the forum structure. For example, Rust uses topic-based categories rather than project-based ones, which results in much fewer categories.

In an ideal world, ignoring limitations of the forum software, I’d get rid of categories and only have a curated set of tags, so you can have multiple tags on one topic.

1 Like