ABSTRACT: The indexing systems used to systematise our knowledge about a domain tend to have an evaluative character: they represent some things as more important, general, complex, or central than others. They are also imperfect and can misrepresent something as more or less important, etc., than it really is. Such distortions mostly result from mistakes made due to lack of time or resources. In some cases they follow systematic patterns which can reveal the implicit judgements and values shared within a community who maintains and uses an index. I focus on the example of PhilPapers, the largest database of philosophy texts available, to show how the arrangement of categories and the way items are assigned to them, can effectively marginalise certain topics, authors, and entire traditions. I draw attention to such issues as: ordering, size and depth of categories, the use of miscellaneous categories, localising indexes in category names, and assigning items to some but not other categories. I suggest that such structuring of the index can have an impact on users, normalising marginalisation and contributing to the perpetuation of inequalities. I conclude by offering some suggestions for improvement which might help our databases flourish and become even more useful.