I was asked by @vianeylb who is trying to kickstart Stan community for ecology how to best support such communities here on Discourse. I am convinced we need to support field-specific communities here, the question is how. This turned out to not have a clear best answer and potentially invites for some restructuring of the forums and thus warrants further discussion with the broader community. As you may have noted in my status update from January I didn’t want to invest energy into thinking about Discourse structure at this time, but I think @vianeylb’s effort is a good reason to consider changing priorities.
I see only two sensible options (except for “don’t do anything”). In the following I will use Ecology as the main example, but everywhere “Ecology” is mentioned, I mean “Ecology” or other field with a Stan community.
1) Field Categories
A category is possibly the most prominent place to gather a discussion around a topic. When a new user posts, it is hard to miss that a relevant category exists. Filtering by categories is also the easiest to do.
The main problem I see with Ecology (or any other field) as a category is that currently the Discourse categories are organized in a way that is not really compatible with having categories around fields. “Ecology” cuts across Modelling, Interfaces, Publicity… It could also be confusing to users - if I fit my ecology model in brms, where should I post? And sometimes I might benefit from people outside of ecology looking at my model. Maybe having field-specific categories under “General” or a new “Field-specific” top-level category might work to an extent. This would assume that modelling-specific and interface-specific questions still go to the original categories, which might actually be desirable.
But maybe a bigger change of structure is warranted: the current forum categorization is IMHO not very functional, as - except for the Developers category - very few people can actually use categories to see only stuff they are interested in (the Jobs and Announcements categories also IMHO work well but have very low traffic). The exception are people maintaining interfaces who I presume follow the relevant categories. Rearranging the structure across fields/application areas (and use tags to mark interfaces, modelling, …) might be an improvement in this respect. I imagine there is a bunch of people who would only read a “Biology” or “Marketing” subcategory. I’d be happy to hear how people use the currently existing categories - maybe I am mistaken. This would however be a major change and if an agreement is reached that this is the better option, some more discussion on how to actually go about the restructuring will be held, so unless you believe it is important for this topic I don’t want to go into many details of how that would look like.
2) Field Tags
Another option would be to group Ecology topics by a tag. This nicely fits with the current structure as tags can cut across categories. You can “watch” a tag (so you get notified of new posts) etc. The biggest downside is that tags could be missed by novice users (and sometimes also advanced users). To some extent we could mitigate this by having better post templates and introductory posts, but still many users could totally miss that tags are a thing. We also currently don’t use tags much, but this could change if we want to. In that sense the effort involved in making field-specific tags useful and widely used could be of similar magnitude as the one required to restructure the categories system.
Meta: Which fields should be represented?
With both tags and categories, the question is how fine a structure should we have - is “Ecology” a category? Or should it be subcategory of “Life sciences”? Should “Microbial ecology” be a separate category/tag? I believe the best answer is to provide a few high-level categories/tags by default and create finer subdivisions/new branches whenever someone steps up as a moderator/maintainer/leader of the respective community (as Vianey did for ecology).
Looking forward to you thoughts on this.