How much work should the RSS aggregator do for me? Should it try to group the various feeds by some sort of context such as authors’ topics or tags, or by day, or by geographical location? Despite of some tempting ideas discussed in a cluster of posts around J. Vinson’s great two pieces about Shifting interests of topical blogs and about shared contexts, I don’t want my aggregator to do much more than just counting the posts and perhaps caching them.
A very intriguing concept of the aggregator as an agent with extensive empowerment, is to follow the metaphor of a newspaper which organizes the various articles for me in some deliberate way. Not only that I can carry it with me, but also that it constitutes some self-contained wholeness (e. g. Flock), that it ideally even substantiates the shared context of a set of authors (e. g. Memeorandum), and that it is chronologically accurate. Considering the contrast of scattered blogs, one might even want old email lists back which were more similar to newspapers?
No. In old email lists, the shared context was only to be obtained at a given point in time by having read all postings up to that time. Like a conversation taking place in a large room, it made you feel uncomfortable to comment without having listened to everything that the others had listened to. The chronological strictness enforced a linear processing, and aggregators mimicking newspapers end up with such linearity because they still cannot mimic the real advantage of a pair of newspaper pages: to be a large, two-dimensional landscape. Thus, the linear order retains its characteristic of a knowlegde workers production line that relentlessly marks ever new items “unread”.
In blogs, the shared context is much more subtle, and has different focus and emphasis among different subsets of people. Also consider how difficult it is to identify topics of blogs whose labels could be used to arrange them in a newspaper-like aggregate.
- If the labels are categories, the phenomenon of fuzzy and shifting interest of bloggers devaluates them;
- if the labels are tags, they tend to become an amorphous mass of atomic words without any coherence except questionable machine-made “relatedness”.
I think it is impossible to put topical labels on blogs that interest me. And the commonalities and connections between them are too complex to be delegated to a machine agent.
Similarly, I have my problems with searchwords when I want to retrieve a new meme from the blogs I’ve read, since I often don’t know labels yet for the nascent ideas. I prefer to revisit some blogs instead, and I don’t mind to check several bloggers in vain before I find it because, like browsing folders in vain, the repeated traversal of the network just helps to strengthen the connections in my memory.
For retrieving, I also need a “real tangible location” that Ed Vielmetti mentioned in his comment on Jacks posts. For me, this is the respective blog’s folder in my reader, containing the original pointers, and I don’t like a secondary location such as a newspaper-like digest. And it’s OK for me that the blogs I search, often do not have topical labels in their names but that the names are just imaginative, much like the medieval dynasties often called themselves after imaginative castle names that were more stable than the geographical labels of their (shifting!) territories.