Life in the Network

This photo was taken by my friend Lisa Skaff when we worked at a snowboard camp at Mt. Hood in Oregon. My job at the camp was more or less "IT Guy". I fixed computers, adminstrated the network, and wrote software to manage their inventory. This image shows me doing my favorite thing: enjoying the mountains on a snowboard.

As previous posts have mentioned, oftentimes when searching for content by tag one expects that a particular topic would be tagged with particular keywords. For example, I would expect, when searching for images of, say, a particular snowboarding trick (the above image depicts an "indy"), that a search for the keyword "indy" would yield relevant results. Of course, many other things would also be returned for such a search, for instance images related independent films, or the skateboard company "Independent". This is not much of a problem, as I can simply further winnow the result set down with more constraints. However, I have no way of knowing whether someone posted an image that I would be interested in if he or she instead only tagged it with "snowboarding", or perhaps "half pipe". Certainly I can search simply "snowboarding", and browse around until I find an interesting picture. Or I can peruse tags on images also tagged "snowboarding", and find tags related to other tricks that I might be interested in. This problem seems related to, or another face of, the issue a previous poster brought up of different cultures or groups using different vocabulary to describe data.

Therefore, tagging seems to be useful more as a method of browsing a data set, and less as a method of searching a data set. Specifically, because tagging relies on the tagger's understanding of a piece of data in order to properly organinze it, a search over a set of tags will often yield incomplete results, with no way of determining whether the results are in fact incomplete. Meaningful, useful data can become "lost" to searching if tags are the only means of categorizing it.

Another issue associated with tagging is that tags seem to have a tendency toward either generality or over-specificity. By this I mean that one might find an image taked with "restaurant", which yields almost no information; it is too general. On the other hand one might come across an image tagged "Mike"; it is far too specific. Finding the "sweetspot" of useful information is difficult, and it seems that people often miss the mark. I know that my tags, for example "tree", might be a bit too general.