Tagging conventions and hive organization

March 9th, 2009 •
Tags: organization, tag, tagging, tags

When the excellent social bookmarking site Delicious launched in 2003, it re-branded an old idea about categorizing and organizing data—namely tagging. Tagging is basically the same concept as keywords, but somehow the rebranding makes it seem more friendly and accessible. Delicious provided users with a system that made tagging easy and gave users a way to exploit this organization allowing them to browse tags to find sites and pages of interest.

With tagging comes debate about how best to utilize the tags to the benefit of the community. In the summer of 2007, Tagamac—a blog about utilizing tagging on the Macintosh platform—wrote a post outlining a set of general tagging conventions that can be applied to any tagging system. The post excited me, and I began religiously putting these tenets into practice. The conventions as laid out in the post are simple:

1. Be succinct
2. Use lowercase tags
3. Use singular words

A couple of months back, I realized this system wasn’t really working for me. The realization came from my own habits in searching Delicious for content. First, let me say that Delicious is a fantastic search tool. Since Google uses an algorithm to return search results, it can be gamed and isn’t always accurate. Delicious having user-generated tags is much more accurate. If I browse through a tag on Delicious, I know some user has looked at each of these sites and decided to give it that tag. It’s much more reliable than an algorithm no matter how sophisticated. Needless to say, I often use Delicious to find resources. I noticed that when I searched Delicious, I didn’t assume that others followed the conventions laid out in the Tagamac post which I myself followed. In fact, I couldn’t assume that… at least not with any success.

The author of these guidelines has an admirable goal. By setting out rigid guidelines for tag use, it would make tagging more useful to those who use the organization to find things while also making it easier for taggers to tag a piece of content. The problem may be that tagging is more of an organic activity. People tend not to dwell much on composing tags. It’s almost word association in which the tagger asks, “What does this make me think of?”

I want to address each of the three guidelines in turn. Succinctness is without a doubt important and is encouraged by most tagging systems. Tags are often limited to a single word by using the space as a delimiter. For example, if I bookmark a site in Delicious and want to tag it hardware, tutorial, and diy, I simply type each of those words in the tag box with spaces between like this: “hardware tutorial diy”. If I decide I also want to tag it “how to,” I can’t just type that into the tag box. Rather than adding a single “how to” tag, the item would be tagged “how” and “to”. Instead, I have to either butt the words together or use a hyphen to avoid their being split into two distinct tags. This automatically discourages tags with multiple words and, although there are times it is appropriate (such as the aforementioned scenario), multi-word tagging in general is not desirable or very useful. Using lowercase tags is a reasonable guideline, but most systems are case-insensitive. Why even worry if people searching tags will not see a difference? The last guideline is the most problematic: singularity. It is sometimes difficult to decide which form of a word to use when tagging. A simple rule like “Never use plural words,” would be great if everyone followed it all the time. The problem comes when only 10% of taggers follow the guideline all the time while the rest use whichever form of the word pops into their heads at the given time or whichever form seems more appropriate given the circumstance.

I propose an alternate method: the shotgun approach. Use every form of a relevant word that comes to mind when tagging an item. It may sound a bit like overkill, but consider the stakes: as a Delicious user, I want to find sites about games. I come to Delicious and view the bookmarks tagged “games” to the exclusion of anything that is tagged only “game” or “gaming“. Because of the simple fact that, when I ask myself, “What am I looking for?” my brain replies “games” rather than “game”, I will miss out on the benefits of any resources you have tagged “game”. What is lost by tagging an item with both plural and singular forms of a word? A few seconds at most. If I have a new bookmark that I want to benefit the most people, I will want to tag it with as many synonyms as possible to reach the widest audience. My tag shotgun will sacrifice accuracy to hit a wide range of users with my chosen tags. I can apply this tenet to more than just plurality. In terms of the example given in the previous paragraph, tagging a tutorial with “tutorial tutorials howto how2 how-to diy” will be much more effective to users than giving it any one of those. Even though my tag cloud will look a bit messier, I have provided more value with my post.

This method actually better exploits the advantages of the tag model of organization. For so many years, computers have been dependent upon the hierarchical folder structure for organization imposing a strict one-to-one ratio. We needed to decide the one location that would be the easiest place to find an item and hope we could reason that out the same way when we were later trying to find it. Tags grant us the flexibility or giving an unlimited number of descriptive words to an item to make it easy to find later regardless of how we decide to describe it.