Recently, I wrote about the challenge of managing tasks in the unstructured world of social computing, where I discussed how enterprise social networking had departed from the filing cabinet metaphor of 1990s desktop computing — where we couldn’t find enough information due to immature search capabilities and the inefficient method of placing documents into tidy folders.
So as the Web evolved in the early 2000s, we saw the rise of tagging, folksonomies and search. The theory: Users do their part by lightly tagging bits of content, let it be unstructured, and search will do its job.
The problem is, social computing has become too unstructured for its own good. The Web 2.0-era notion that we should trust everything to the crowd’s folksonomy and new search technologies hasn’t come to fruition in the way we’d hoped, both in the consumer world and in the enterprise. On Facebook, I can’t find a thing when I swim upstream to find posts more than a few days old (Facebook search still doesn’t do much for me). On Twitter, which hashtag should I use or search for? Google + may address this problem with Cirlces, but that would only solve the people element — not data and topics.
Blogs have been a little better, with WordPress’s categories hierarchy as an example. But I’m generally disappointed, if not annoyed, with how long it takes me to find content both long and short on these platforms.
The enterprise has been moderately better than the consumer social world at this problem. With auto-fill, natural language tagging capabilities, and recommendation engines, we have made it easier to suggest to end-users how they should sort and search for information. But in my experience, once we hit a critical mass of content, it generally becomes uncontrollable and some poor community manager goes in and tries to pick up the pieces. Some people tagged something “sales presentation,” while others tagged it “sales preso” or “sales powerpoint.” Some tag it with one; others will all three. Either way, it’s one of the reasons I haven’t found a tag cloud to be really helpful in years.
Imposing a Little Structure
While we don’t want folders like on a PC or document management system, the time has come to add some hierarchy and structure to social computing platforms. This will make certain topics, projects and groups more visible to the end-user, while ensuring that people access the most relevant filters and streams within the company.
Now, this doesn’t entail turning off end-users’ ability to create and contribute to taxonomies. However, the way the hierarchy of existing classifications are presented should encourage them to assign their content to the right bucket — rather than just construct one wlly-nilly with no additional thought. At tibbr, we approach this through “Subjects.” We place subjects into a hierarchy. There, users can visualize a high level subject (i.e. sales), but then go a level deeper into that subject, and create a sub-subject (i.e. “sales presentations”). As they peruse the subject hierarchy, they can choose which subjects it makes the most sense to follow. As users create a subject, the system scans for similar ones so the end-user has confidence that a similar one already hasn’t been created.
This model allows much better admin control, which, like it or not, is important with taxonomies inside companies. Because you can put all the recommendation and auto-suggest features you want in place; users are still going to create useless topics and tags that will muddy search results and discourage access to the most relevant content.
So with any kind of subject-based hierarchy, the admin needs to be able to move around subjects easily (drag-and-drop), rename them and seamlessly change them on the fly with as little friction as possible.
Cleaner, More Relevant Taxonomy and Architecture Means Cleaner Activity Streams
Having a cleaner social taxonomy will improve the quality and relevance of activity streams. Right now, too many streams inside companies are either a firehose of “everything” — or they filter by an individual group.
But businesses — especially large ones — are way more dynamic than that, and need more fine-grained controls that make it easy to create an activity stream based not only on people or system data, but concepts, ideas and projects related to each. So if you’re able to pull together relevant subjects — or sub-subjects — from your cleaner taxonomy, you’re going to have better, more relevant streams that map to specific business processes.
Conclusion: Finding the Middle Ground
Again, I’m not suggesting we revert back to the days of heavyweight information management that turns end-users and their administrators into managers of digital filing cabinets. At the same time, the sheer volume of information shared by both humans and machines today means that leaving this to the wisdom of the crowd might not work as well as many of us had initially hoped in the early waves of social computing.
