I was looking for a brief introduction on the subject for someone. Since I cannot find one, I’m going to give it my best shot. It’s unfortunate, if you search for Link Analysis today, you’ll mostly find references on search engines and the internet. This is one application of link analysis, but really doesn’t do justice to how broadly applicable the technique is. For example, searching Google for “Link Analysis -hyperlink” reveals that the technique is applicable to police investigations. Actually, I was aware of this application, as I used to work for a company that did training in investigative methods for various police and police-like organizations. I’ll provide a brief explanation as it was first explained to me (and I cannot vouch for the historical accuracy of this):

Link Analysis was largely invented by Henry Ford. He wanted to improve the efficiency in his assembly lines. He would write down the worker’s position on a piece of paper with pencil. Every time the worker would first reach for a tool, Henry would write the name of the tool down and draw a line from the worker to the tool. Each additional time the tool was grabbed, Henry would draw over the line (making it darker). The workstation would then be re-organized so that the tools that had the darkest lines were closest to the worker. When applied to the whole factory floor, tools with the most/darkest lines would be the first to be duplicated. Imagine if everyone needed a 3/8” box wrench, but only one was available.

For police investigation, the technique is generally used as a sorting heuristic. You have a known bad-guy. Look through his cell phone history and draw lines to his contacts. The darker lines are a prioritization scheme to talk to new people and gather information. The darker lines are not and indication of probability of guilt, but they indicate people that might provide additional information about the goings on of the bad guy. You can do the same analysis with bank transfers, emails, etc. For those who are paranoid about such things, I hope I’ve provided some justification for your fears. It is simple fact that police often have to deal with sub-optimal information. Such heuristics have been empirically determined to be valuable. Obviously this can be abused, but I really only intended to highlight the applicability of the technique.

For search engines, the applicability of the technique should be obvious. Even if you have no ability to process the content of the hyperlinked resource, the link itself is information. Baseline recommendations can be made just from the link structure. Of course, if most hyperlinks are of the “I absolutely hate link” type, the technique would probably not be empirically validated. Fortunately, this is not the case (though you can see that this is easily abused by SEO companies).

Anyway, I’m not going to tie this to economics. I just needed the virtual scratch paper. Sorry.