Further thoughts on topics, categories, tags and ontologies
I had a long chat with @Nadia about this, and she helped me clear my thinking. Let’s see.
- You can never neatly map the world with ontologies. All ontologies are unstable and incomplete, and run into the platypus problem. I highly recommend reading Shirky's forceful argument – it certainly changed my way of thinking. In terms of our own experience, what was the unMonastery group about? Saint Benedict? Work? Matera 2019? Rituals? Individual threads were about all these things, and others. But the unMonastery group was really about unMonastery. A lot of what "unMonastery" was encoded a social relationship linking unMonasterians, not anything related to knowledge mapping. This is fundamental. Every new project will be about itself. We will keep inventing words (The Reef, OpenVillage...), and having to go back and change our ontology around.
- You should not point the user to dead zones. Suppose the following user story: Alice arrives on Edgeryders and is interested in this OpenVillage thing. The site seems to be about Agriculture and Food, Accommodation and protection, Environment and nature etc. Alice thinks this Accommodation stuff seems interesting so she clicks. She promptly finds out that the last update around accommodation was two years ago. She replies to Bob's two-year-old topic. Bob, meanwhile, has drifted away from Edgeryders and does not reply. Alice is having a terrible user experience! We should try to steer her to the inhabited part of Edgeryders: the flagship project groups.
So (asks Nadia) what, in my proposal, is different from what we have now? This:
- We use tags (not categories) as a best effort practice, knowing they are patchy and imperfect but still improve searchability.
- We use project names as tags, (as suggested by Matthias above). Any topic created within a project is tagged with the project's name.
So, how would this work in practice? For example, now OpenCare is a flagship project. You go to discourse.edgeryders, you see the category right there. You are encouraged to join. People will be waiting to engage. All good. Content is tagged, of course. In 2018, we end the project, remove the OpenCare category (or demote it, so that it does not show up, or clearly flag it as an archived project). But all the topics are still there, still tagged. For example, this topic about open insulin will be tagged open source, insulin, diabetes, Belgium. But also OpenInsulin, and possibly OpenCare.
Suppose also in 2018 we get a contract to do work around the smart open ecovillages in Europe (not impossible). We create a new category called “Smart Open Ecovillages”; we create a list of tags that are related to the new project and we ask Discourse to bring in all the topics that are tagged with one or more of the tags. This has an added advantage: we instantly populate our new project with content.
Result:
- categories are unstable and follow activities.
- tags are stable
- projects are also encoded into tags.
Technical question: can the DB record tag authors?