People who work collaboratively often review each other' s work. They may do so by annotating work artifacts, such as documents. People seeking to understand annotated artifacts confront two problems. Clutter occurs when people see an overwhelming number of annotations. Clutter is a well-studied problem, whose solution involves narrow-filtering annotations: showing a viewer only a subset of annotations on the viewed artifact. Another, less widely studied problem is delocalization. Delocalization occurs when people must understand multiple parts of an artifact, and annotations on those parts, to understand the part they are viewing. For instance, a person viewing a system design document might need to understand the system requirements to understand the design. We propose that delocalization of annotations can be addressed by broad-filtering annotations: showing viewers of an artifact some annotations that were made to other parts of the artifact, in the context of that portion of the artifact they are viewing.We are particularly interested in annotation of software artifacts, which is a common task in software development. We introduce a three-dimensional taxonomy for delocalization within software systems: lateral delocalization (between items of the artifact within the same development phase), longitudinal delocalization (between items in different development phases), and historical delocalization (between successive versions of the same item). We discuss methods for addressing clutter and delocalization concurrently. We develop a graph-theoretic model of annotated artifacts incorporating clutter and delocalization, prove the mathematical consistency of this model, and apply this model to various existing artifacts and systems to demonstrate its breadth of applicability.Annotation is a vital part of software inspection, which is a well-known means of software quality improvement. We discuss how to address delocalization within software inspection. We describe the development and use of AnnoSpec, a software inspection tool supporting clutter reduction and visibility of delocalized annotations in accordance with our model of annotation. We conducted a pilot study in which software professionals performed inspections with AnnoSpec. Results of this study suggest that addressing delocalization has some benefits in the helping people perform software inspection effectively, providing a "proof-of-concept" for the use of delocalized annotations in software inspection.