With just 100 participants in the public trial, each using the system for no more than 2 hours each, the UT prototype map quickly became covered with 'pockets' and 'threads'.
How can we imagine and design filtering systems that reflect the complexity of our moods, of our social and cultural interactions, of our emotions and feelings about place and identity?
Filtering is often structured around inclusion and exclusion lists, yet how do we retain an essential spark of serendipity: one of the signal properties of city life? Preferences, too, focus on the individual and notions of choice. Can this be too limiting – only showing what we think we want, rather than what we might be interested in if only we knew of its existence?
What kind of trust or reputation systems can be created, adopted or combined which are not self-serving, but reflect the dynamic process of how individuals and communities
Posted by Giles Lane at February 13, 2004 05:30 PMI was just thinking -- and blogging (http://gumption.typepad.com/gumption/2004/02/hypertransient_.html) -- about the filtering issue with respect to the recent announcement of WaveMarket, when I discovered these new threads of yours (via Anne Galloway's blog). Basically, I think that to balance the desire to reduce clutter while still remaining open to serendipity requires a holistic solution wherein more complete digital representations of selves -- beyond the [rather ephemeral] posts themselves -- can be used to enable collaborative filtering techniques to identify posts that are likely to be of some interest. Although, as Rachel points out under the Collaborative Cartography thread, different people will have different levels of willingness to share different parts of themselves (and digital representations thereof).
Posted by: Joe at February 20, 2004 07:07 AMOops, wrong blog URL, make that: http://gumption.typepad.com/blog/2004/02/hypertransient_.html.
Posted by: Joe at February 20, 2004 07:32 AMThis filtering problem goes to the root of electronic knowledge management. Both our human-machine interfaces and our categorization methods are too primitive, labour intensive and narrow even for the blandest monomedia.
Some examples:
I want a picture of a white tree.
I want to find choral music by Liszt.
I need my diary to tell me where I might have left my coat.
Until interfaces can ask pertinent questions that we can answer quickly at upload time, or, like pets, intuit our intentions, the context and content of electronic media will be woefully underdescribed.
Posted by: Ant at February 22, 2004 04:03 PM