One part of pagerank which I've always been confused about is how it factors Google itself into the model. When computing pagerank would Google add a node for its own domain (i.e., google.com and all its search result pages) that points to every page on the web? They must realize that they themselves are a major driver of traffic on the web and that would affect the assumptions of the random surfer model (i.e., a random surfer would go back to google to do new searches occasionally). Doesn't this become a chicken and egg problem?
Comments
One part of pagerank which I've always been confused about is how it factors Google itself into the model. When computing pagerank would Google add a node for its own domain (i.e., google.com and all its search result pages) that points to every page on the web? They must realize that they themselves are a major driver of traffic on the web and that would affect the assumptions of the random surfer model (i.e., a random surfer would go back to google to do new searches occasionally). Doesn't this become a chicken and egg problem?