I’m setting up a new Discourse community and wondering about how to create categories. In particular, I’m nervous about making categories that are narrow enough to be meaningful, without making them so narrow that there are too many categories, topics could easily belong to more than one category, and people won’t know where to look.
Why aren’t categories implemented like tags in stackoverflow, where you can simply have more than one? I’m sure this has been discussed elsewhere, but I can’t seem to find it.
My two cents: start broad, and add categories as needed, especially is the community is brand new. Iterate. (Iteration is almost always a good thing.)
The Discourse board that I manage and spend the most time on is centered on video games. There is an “Other games” category, and I sometimes add new categories if the discussions concerning such a game reached a critical mass that warrants its own category. Ask your users if they feel like it’s needed.
But how do you know what category to post a topic in, if it could belong to more than one? How do you design categories so they’re comprehensive without being big enough that people aren’t sure which category is the best fit?
From what I know, the idea behind this is to simplify posting.
Having tags instead of categories may complicate posting - instead of choosing just one category, users will have to decide whether a tag is good or not for their post - for every tag separately, which is a real rigmarole.
Instead of tagging, the focus is on two things:
What is hot / new right now
The search function.
Actually, if you have an excellent search function - and it just works - then you don’t need tags. And now this makes readers’ lives easier, not only writiers’ ones - because now readers do not have to find out which tags will possible contain a post which they want to read.
To better understand the issue with heavy categorization or tagging, consider movies discussion forum.
One cannot disagree that there are many movies which belong to many different categories.
What is more, while for one a particular movie can be hilarious, for someone else it can be not. So putting such a movie under hilarious tag will confuse those for whom it is not such a move.
Now consider you don’t have that hilarious tag / category.
So a curious person would just use the Search button and will see only those movies where hilarious is mentioned most of times, with most likes etc etc, which means more than one people considered the movie to be hilarious.
As you can see, getting rid of categories/tags and focusing on better search and what's hot right now makes the quality of your content higher in the eyes of your users, and confuses people less, and as a bonus you simplify posting, so no one has to break their head thinking about how to categorize/tag their new topic.
It’s a long story @jfrank19701407, but traditionally you want categories to wall people off from each other in social discussions. In other words the [politics] people and the [android] people don’t have much in common. Or the [democrats] and [republicans]. This is typical and expected in discussion software. It’s a strong, hard wall.
Tags are for looser, less important distinctions. And after the 2nd tag, average people cannot tag with any semblance of sanity, so you rapidly reach a point of hugely diminishing returns with tags. For most uses, search is a better solution anyhow; forget magical tags, just make sure your topic or post has the correct search terms in it and you’ll be fine.
Over specialising categories is a common problem that manifests itself differently in communities old and new. It most commonly manifests as a gigantic front page, filled with every possible permutation of the community’s general topic. A community about games might have separate categories for firstperson shooters, thirdperson shooters, side scrollers etc, where they would better be served with either a category marked “Shooters” or simply an overarching category marked “Games”.
In new communities, this stems from a desire from the founder to make sure that any topic the community could want to talk about is covered. While this is understandable, it will lead to users simply being intimidated by the enormous number of categories at their disposal. In most communities with vast numbers of categories, the majority of them will lie fallow. If a new user posts in one and doesn’t get a response, they’re unlikely to persist much further.
This is what we did. Or rather the owner did - he copied most of the ‘categories’ from the old forums that were the busy ones, omitted the less frequently used.
As time went on, users started requesting sub-cat’s and they (well some of them) got added as we went along.
But the examples you stated are unlikely to occur in the same forum. The forum itself is a walled area for Democrats, for example (Or politics and android? Would you ever see this in the same forum?) It’s more likely that there is a forum for Democrats and the tagging would be useful, for example for an Obama tag. You don’t want an Obama category, because nobody just wants to chat all day long about Obama. But being able to tag a post with Obama or Hillary or whatever would enhance navigation.
StackOverflow works extremely well based on the tag system. I don’t think search itself is a replacement for that. I have never gone onto StackOverflow and did a search for ‘django’ but I regularly go in and navigate using that tag.
I guess it is possible to just create fake tags in Discourse that just return search results, but the search implementation I’ve observed on this site only returns search results within the search box widget. Would it be hard to create an actual page of search results?
But the examples you stated are unlikely to occur in the same forum. The forum itself is a walled area
This is what I was thinking too. I’m not so sure I want rigid walls like that, because my community is going to be dealing with people talking about related topics. I trust the brains behind this, though, so I’ll go with it.
But also, if tags are this unhelpful, why does StackOverflow use them in such a fundamental way?
Tagging is absolutely not essential for every forum. Which is why it was omitted from core… I believe the team are looking to bless this plugin as “official” in due course after Discourse v1 is released.
There is lots of previous discussion on tagging vs categories… you just need to do a search for “tagging”
In my experience the majority of our forum users have a hard time writing coherent/relevant titles. I must rewrite around 80% of them. The thought of letting them loose with tags is a daunting prospect! I know what will happen we’ll just get people using every tag they can find, it would probably end up creating more work for the admins, de-tagging everything.
We really don’t need that level of granularity… categories and sub-categories work fine. I’m sure it works great for technical communities like StackOverflow (i.e. long tail QA). When you’re dealing with users who struggle with the fundamentals of using the web it’s not going to add any value and will just introduce clutter.
It didn’t even work that great on Stack Overflow. It takes a LOT of discipline to not have a tagging system descend into madness. Even a technical user is in crazytown by the time they reach Tag #4… and Tag #3 is usually sketchy as well.
(don’t talk to me about tag hierarchies, relations, and singular vs plural tags either. Many many subtle gotchas with “wheee its easy just type some tags everyone” systems. A panacea, they ain’t)
The reasons for omission are fair but it seems that it’s too much of an assumption to make about the types of forums that will run on Discourse. Tags are very useful if not for nothing else but to give a URL to a page with a list of tagged items. A link to a search results page is also very useful and I actually like sharing search results pages (an often overlooked resource type). But this does not lessen the value of using tags. A post may not contain that tag in the body or title but still be associated with the tag. It’s not that uncommon. Tagging is just a way to informally group content into secondary collections. I’m surprised that they are not deemed important amongst the Discourse team.
In reply to the orginal post and the advice to stay broad intially with only a few categories… Taking that approach is smart and also highlights a good use case for tagging which would help determine what new categories to add, based on the tagging data.
I’m all for KISS and minimalism wherever possible. Tags, however, belong in discussion systems imo.
For Discourse, I think it would be a good option to give to staff+ since discipline is implied and would be used for special case organization purposes and not for the community-at-large to use.
I could agree with the reasoning for a feature omission based on usage pattern design, yet still may need or want that feature and, most importantly, can handle that feature as the forum operator.
However, with a plugin system, it becomes mostly moot. And tagging is a good candidate for a plugin. So that makes it an acceptable omission in this case
