This weblog had moved: http://www.coolweblog.com/bilodeau/

# Control mechanisms essential to collaborative categorization : In order to be useful to more then one person, a classification/categorization scheme must be controlled. The two or more people involved need to have some sort of shared understanding of the terms being used to classify items, and the process by which classifications are assigned to terms. This process can be formal or informal, strict or loose, static or dynamic, etc. But the control mechanism has to be there if the metdata produced from the effort is going to be useful.

Traditionally, control has meant the development of formal taxonomies and vocabularies, and the professionalization of the role of classifier (i.e., librarian/information specialist, but let's just call them librarians for the sake of simplicity). Librarians, charged with helping their clients to locate the information relevant to their information needs, have devised these schemes, along with the methods, tools, and training needed for users to be able to locate items using these schemes. It is hard work for everyone, and far from problematic. But it works fairly well for what it was designed to do.

Adhoc tagging (sorry, terms like folkonomies and tagsonomies still sound ridiculous) is aiming at a slightly different problem space, although, done properly, the solutions should scale to the more formal case outlined above. The main idea is to build a system that allows a group of people to collaborate on the categorization/classification of digital items of some sort.

The current tools, to a large extent, define this group of people as "everyone using the systems." There are,mind you, ways to define subsets of users. Delicious gives you an inbox. Flickr allows you to create a group of contacts, and further refine that list into friends, family, and other contacts. A good start, but still much too rudimentary to be useful. (For the record, I think these sites should focus on aggregating the basic data, exposing that in a useful and reliable way, and then letting others develop software that uses the raw data streams to provide a useful tool. More on that some other time.)

A lot has been written about the problems faced by the current systems: spammers, inappropriate material, inconsistent application of tags, etc. The solutions that I have seen are mostly variations on the traditional, centralized control. Top-down rules imposed by the aggregators, to be adopted by all users.

The strength of collaborative classification tools is the people using them. The control mechanisms must be there, but they cannot be imposed. They have to be socially negotiated by the members of the group. There are limits to the size of group that can accomplish this negotiation in a meaningful way. These groups of interest might be a group of friends, a group of students in a class, people working together on a project, people in a department, members of a professional organization, a group of subject librarians, a group of specialists spread across the world. These groups need tools that allow them to control who participates in the process.

(You didn't really think this was going to be a big, warm, fuzzy group-hug-free-for-all, did you?)