Consider now readers of a hypertext document. Every time they go on to read a new section of a document, they send an http request to a machine in order to obtain a copy of the new section. The machine can be configured to record these requests in a log. Quantitative analysis of this log can be then used to empirically measure the depth of curiosity of a document's reader. The ALife Onlife service is so configured, permitting the analysis undertaken here for the documents AL-SIM and CA-FAQ. The technique is similar to the "preferential-looking" test used in psycho-physiology of infants. In each case, one attempts to obtain information about a psychological state on the basis of external, physical, action.
The data are shown in figures 2,3, and 4, where the distribution of curiosity is plotted in linear-linear, linear-log, and log-log scale respectively. It appears that curiosity about CA-SIM is best fit by an exponential law, while curiosity about CA-FAQ is best fit by a power-law.
One of the signature features of critical phenomena is a loss of scale at the critical point. In the case of documents, scale may be associated with the type of interconnectivity represented in the information in the document. In CA-FAQ there are several levels of interconnectivity, correlated with the varying degrees of authorship and editorial pressure on sections of the document. While CA-FAQ has only four such levels, one can imagine that very large and complex documents could have many. Thus the power-law scaling exhibited by CA-FAQ could be generic for network-generated hypertext.
