marc bohlen

Present Tense: The other Ready-Made

Every crawler visit to the web is an exercise in extracting manifestations of the present tense. The idea of the purposeful search is, for me, only a pretext. Not that the methodologies of transmission protocols and decentralized packet switching mediate the accidental. While the kindly connected grid guarantees that all linked entries remain findable, the http protocol is designed for data representation and transport independent of data content. My attraction to the accidental is more a result of the linguistic oddities and quirky hyperlinks heaped together through the topology of world wide wired servers. Wandering these links, crawlers generate curious finds, some of which deserve entry into the equivalent of the 16th century German Wunderkabinet, or collection of curious man made and natural objects. More importantly however, is the understanding that every launch of a crawler is a call to create a document. These crawler results are always historical documents, transcriptions of a current state of western indexed knowledge represented on the web. Epistemologically restricted to the digitally conform, crawler search results show us precisely how we have mapped the world. They are cultural artifacts in their own right. Precise time stamps authenticate the capture of a singular moment in time. But the same time stamps that stake the claim to present tense also guarantee that the results are obsolete upon creation.

What we involuntarily uncover in the process of spidering through the web may be of more interest than what we sought out to find. The forms by which we choose to encode the results are, in my view, secondary to the act of collecting them. It is pleasurable to imagine crawler activity while one waits for a result. When our crawlers seep through linked data they perform motion similar to that of a random walk. In Physics an n-dimensional random walk is a path followed by a particle as a function of only its starting point and the probability of transitioning to one of n next locations. Brownian motion of a solid particle suspended in a liquid medium is the classic example of a 3 dimensional random walk. In the crawlerÕs case the transitioning probability itself is multi-varied. The transitions are of variable length. Crawler initializations mixed with imperfections of the search engine results make a closed formulation of the probabilities of transitions opaque. More than anything it is the linked nodes of this opaque path, collected in time, that build the signification system of a web crawler. Semantics assemble themselves ad hoc to ready-made manifestations of the immediate present, expressed as action through index. In a curious way, crawlers automate the extraction of readymade knowledge links. This notion is orthogonal to the utilitarian pretext of data mining. There is less gold then there are fossils in the sediments of layered data. Maybe it is too early to understand what these strange findings are good for. The data forests and crawler paths through them may become fossil fuel for engines we have yet to conceive.

[ .html]

[ .html]