Given that I'm currently beginning to tackle site search (currently still in the background) it made sense to also look at the state of content taxonomies on this site. Although I now have a few different content types (including some still entirely invisible to everyone but me) I'm effectively limping along with the categorisation system that just sort of happened on the first version of theAdhocracy. Not only is that a system designed around the particular quirks of WordPress, but it's also one which never had any real thought put into it. I created categories and tags as I went, because that was simpler.
That's why my categories include now utterly redundant legacy groups such as 100 Words. Whilst I'd still like to somehow relate all the posts from that particular challenge, a category was always a bad way to go about it; a content tag would have been more appropriate.
So I know that I have a problem, but I've been putting off tackling it because, well, I don't really know how. Taxonomies are scary. Once you have one in place, it becomes integral to the way a site works. To some extent, I've just ignored the issue since launching version three by making my categories and tags little more than window dressing. Sure, you can see what they are, but they don't do anything; they don't interlink or let you find related content at all. But as I start to put search functionality in place, that's no longer going to be the case. Damn.
What Does the Internet Think?
Like most problems, then, the first step is to start searching around for advice and potential solutions. Weirdly, though, I've found the results pretty lacklustre. Most blogs on the subject are coming at it from an SEO or marketing perspective, but whilst these are both niceties I don't want to completely ignore, they're also not my priority. This site's taxonomy should, first and foremost, serve me. It should enable me to locate things as quickly as possible. The second goal is to serve other people, letting them find relevant content. Only after those two elements are complete do I care about things like SEO.
Searches did pull up a few useful pieces of information, most of which feel quite like common sense but are still worth keeping in mind:
- Categories should be broad; tags should be targeted and are for the detail;
- People don't have a particularly high capacity for remembering lots of keywords, so minimise the number of categories (less than 10 seems to be a good rule of thumb);
- If you find a need for more than 10 categories, consider subcategories instead; again, these should be minimised and you really, really want to avoid nesting more than 2-3 levels deep;
- A broader taxonomic tree is always easier to use than a narrow-but-deep one.
Useful, but ultimately not the silver bullet I was looking for. Time for my next tried-and-tested method of solving a problem: looking at what other people do. Except, most of the places I normally go to are getting around the issue of how best to use taxonomies by... not having them. Adactio/Jeremy Keith (normally my gold standard for answering "how do other people do this?") just uses tags and seemingly adds them with wanton abandon. There's no issue with doing that (it's almost certainly how I'm going to use tags too) but it didn't give me an aha! moment like I'd hoped. I found the same thing with Andy Clarke and Graham; others simply don't bother with categorisation at all, such as Ethan, Andy Bell, and Amelia; and a few go for a simple date based hierarchy, such as John, Mark, and Khoi. Incidentally, I found myself really liking that timeline-style, date-based view... but I know it won't help me in the long run, so I'm ignoring it for now.
OK, but what about big, content-based organisations? Bizarrely, this was initially even less help. Sites like CSS Tricks and Source simply don't bother. They might include some filtering options, but categories and tags are largely absent or hidden from users. Others, like MDN, do have taxonomies but they're opaque at best. A List Apart comes the closest, with a simple and intuitive hierarchy that I really, really like. It has six top-level categories, each with between four and six sub-categories, and all of that is immediately visible from the topics page, along with the total number of related articles and a quick description. It's clear, easy to navigate, and allowed me to find information fairly quickly. Intuitively, it feels like a combination of this and the hyper-detailed tagging that goes on at Adactio would be a good fit.
Digital Gardening
It was around this point in my research rabbit hole (oh, I had no idea what was to come next) that I oversaw a conversation Eric was having on Twitter. They were talking about the concept of Digital Gardens, an idea that I began to dig into[1]. I ended up reading Joel's thoughts, and then some more of them, which very neatly outlined some of my unvoiced feelings around why latest-post-first blogs aren't a good content pattern:
Using this newest-first approach makes the site impenetrable and less useful for me and especially for anybody that might be visiting to learn or research.
Interestingly, he has landed on a taxonomic-free site structure, instead producing "guide" posts which link his articles together under themes and then relies on Algolia (hello!) for full-text search of the content of blogs to return tags etc. That was something I was actively trying to avoid as I feared it would result in gigantic indices in Algolia, but Joel's clearly running a similar setup so maybe it wouldn't be too bad? Still, this strikes me as mimicking categories and tags without the effort of curation, which is fine but doesn't provide the editorial control I know I prefer. And, as he puts it so well:
Being useful for me is the primary use case for this space on the internet. It's not that I don't care about you, but this is for me. It's here so I can record what I think and know and preserve it in time and space.
It's my garden, but I'm happy for you to hang around and eat tomatos with me.
Joel's posts led me to Tom's, which may be the original source of the "digital garden" concept (or possibly not). Tom had created a wiki for himself to better manage information flows on his site and in general. You can see the final output here but it was the structure and root reasoning that struck me as particularly relevant. Tom was looking for a way to create an "information garden", a tool that would allow him to both curate and connect information.
This is where there’s a gap for me personally. No place to store and evolve deeper longer-term thinking.
He specifically mentions tools like Pinboard and Evernote[2], which is precisely what I had tried to do years ago. The problem was always the editorial experience; there was too much friction involved in cataloguing a datapoint into those systems. But what I've found more recently is that my Notes section largely removes that friction. Not completely – I still don't have a good way of moving information from my phone or certain social media sites – but the potential is there to tackle those problems in the future.
Interestingly, his aha! moment came from realising that folder structures are a "best in class" tool for personal information management. Wait, really? Apparently so:
Yet Bergman and Whittaker have found in their pioneering personal information management research that these other methods that work best for public information management don’t work as well for personal information management.
Okay, but there's a paradox in that quote. I'm creating an information store for public and personal access... so great, I need to do both and neither simultaneously? Well, having spent a bit of time browsing around both Tom's wiki and a few others he links to I concluded that this wasn't the format for me. However... holy rabbit hole, Batman!
A Diversification of Ideas
By this point, I was bouncing around the indie web like a jackrabbit in March. I found a whole bunch of other digital gardens to dig around in (thanks Maggie), some in a wiki format, others in traditional blog formats, and others in the in-between. I also began rapidly uncovering a whole list of other really interesting points of view. There were the highly technical but awesome ideas like the open transclude UX pattern, to the simpler yet equally as novel concept of a /now page[3].
Then there were the concepts that made me sit up and think aha! (it just wasn't the aha! that I was hoping for today 😅). I love, love, love the idea of Mandy's Working Library. It feels a lot like my own concepts around both Notes and Reviews (reviews are one of them invisible things) and there's definitely some parts which I'll keep in mind as I flesh those sections out. Ditto for Gordon's bookshelf, which just feels like a nice information layout for that kind of concept.
In a similar vein, I really liked the interconnectedness and hierarchy of information that Buster achieves through his use of Piles and Beliefs. It's definitely got elements of the digital garden concept to it, but the way it naturally reinforces ideas until they solidify into something with greater meaning is a neat trick. Brendan has put together a template for a similar idea, which he refers to as Canons. These have a clear added benefit of breaking out of the date-based hierarchy that I'd definitely like to move away from.
Finally, there was Derek's extremely well-reasoned take on the benefits of daily journaling. Whilst I know from experience that this kind of structured log doesn't work for me, his "thoughts on" journals sound really interesting:
For each subject that you might have ongoing thoughts about, start a separate “Thoughts On” journal. Whenever you have some thoughts on this subject, open up that file, write today’s date, then start writing.
The idea is a simple one which lets you keep all your thoughts fairly well contained and allows you to see them evolve over time. That would definitely be beneficial in the long run, particularly if you pair it with the piles --> beliefs info concept I mentioned above. I'm just not sure that the web is the best place for doing that. Personally, I've come to love working in Workflowy for this type of rapid-fire ideation and thought-capturing. That said, categorisation is another way to achieve this, presenting all your related ideas together in a single stream.
The End of the Tunnel
Is it quite an emergence into a pure, white light full of serenity and clarity? No. I've spent nearly four hours hopping around, absorbing new concepts and reading about cool new ideas, and I've come away with a new way of looking at the problem, but not necessarily a solution. That's why I'm writing up this blog post, to try and help it coalesce in my mind into something actionable. So, here are the key things that felt somehow right from each of the models I looked at:
- There should be a clear sense of hierarchy that you can drill down into;
- The top-most level needs to be concise and broad;
- The bottom-most level needs to be infinitely scalable to allow interconnected ideas to build up into a web;
- Connections should be able to be visualised as single information streams;
- I really like the model of clustering information under pillars that can evolve into beliefs;
To me, then, something akin to A List Apart's category -> subcategory but with tags on top still feels right. Categories give me the high level "Canon"-style topics, subcategories let me build up my own personal beliefs, and tags allow for a web of interconnected ideas, a bit like a wiki. As those connections grow, it gives space for a tag to evolve into a subcategory.
What this has really taught me, though, is that I'm not happy with how information is structured on this site. My current breakdown of Articles, Journals, and Notes is useful, but I'd rather group these individual concepts into categories. I'm not sure how that would work and I dread to think about the URL changes it might need, but perhaps I can provide both architectures simultaneously and get away with it. I'm also stuck in a weird in-between of liking date hierarchy in terms of seeing how ideas have evolved, but no longer wanting that to be the dominant/default sorting method. Journals are inextricably linked to dates, which is fine, but articles and notes could easily be broken away from that model[4]. There's an actionable path here, but also a lot more thinking to be done.
I knew this was going to be a can of worms 🐛
