The usefulness of web spam

Spam comprises at least 60% of the public web, and search engine companies invest considerable effort in rejecting these apparently useless pages. But how bad are spam pages in search results? Can spam be dealt with as a side-effect of dealing with page utility, or is the relationship more complex? Thirty-four volunteer judges rated selected individual documents first on usefulness to a specified task and then on degree of "spamminess". Our results show that the relationship between spamminess and utility is far from clear cut; judges found that an important proportion of spam documents were useful. We conclude that evaluation should consider both utility and spamminess, as separate factors; and that search engines should not summarily discard spam pages but should take their utility into account as well.