In this paper, I attempt to ascertain why Relevance Feedback techniques are currently absent from the Web. I survey four feedback techniques, spanning both probabilistic and vector models as well as emphases on term weighting and term selection. I further detail some of the results of combining these techniques and the gains that can be obtained. Finally, having laid the groundwork of how these systems work and what they require, I speculate on the possible reasons as to why Relevance Feedback is not available on the Web in any useful form. In doing so, I detail several open questions regarding Relevance Feedback. It is my belief that in investigating these questions both existing and new, innovative Relevance Feedback techniques will be implemented on the Web and thus improve everyone's ability to find good, useful, and relevant information.