Search engines like Google.com use the the link structure of the Web to determine whether web pages are authoritative sources of information. However, the linking mechanism provided by HTML does not allow the web author to express different types of links, such as positive or negative endorsements of page content. As a consequence, search engine algorithms cannot discriminate between sites that are highly linked and sites that are highly trusted. We demonstrate our claim by running PageRank on a real world data set containing positive and negative links. We conclude that simple semantic extensions to the link mechanism would provide a richer semantic network from which to mine more precise Web Intelligence.