Identification tools identify which virus has infected a particular executable. Defining failure in this process turns out to be easier than success. The identification tool has failed if it cannot assign a name to the virus or assigns the wrong name to the virus.
Determining if a tool has correctly named a virus should be a simple task, but in fact it is not. There is disagreement even within the anti-virus research community as to what constitutes ``different'' viruses. As a result, the community has been unable to agree on the number of existing viruses, and the names attached to them have only vague significance. This leads to a question of precision.
As an example, consider two PC virus identification tools. The first tool considers the set of PC viruses as 350 distinct viruses. The second considers the same set to have 900 members. This occurs because the first tool groups a large number of variants under a single name. The second tool will name viruses with greater precision (i.e., viruses grouped together by the first tool are uniquely named by the second).
Such precision problems can occur even if the vendor attempts to name with high precision. A tool may misidentify a virus as another variant of that virus for a variety of reasons. The variant may be new, or analysis of samples may have been incomplete. The loss of precision occurs for different reasons, but the results are no different from the previous example. Any ``successful'' naming of a virus must be considered along with the degree of precision.