Patents often contain technical information intertwined with legal meaning, and inventions are often difficult to describe in words. Despite complex interpretive rules, patent law has failed in one of its essential missions - giving those who need to read patents the ability to understand the scope of a patent's claims in a consistent and predictable manner. As a result, those who rely on patents - patentees, potential and actual licensees, potential and actual defendants, future patent applicants, courts, and even the Patent and Trademark Office - may find it difficult or impossible to discern the metes and bounds of any particular patent at any particular time. This article examines a root cause of this failing: patents are interpreted in different ways during the application and enforcement processes. When someone applies for a patent, the patent examiner considers the 'broadest reasonable construction' of the claims regardless of how one skilled in the art might construe them. During litigation, the court considers how one skilled in the art would interpret the claims. The article considers why different interpretive rules are applied, whether the rules are applied as intended, and the problems associated with using two different rules. Finally, the article suggests that patent claims should have the same meaning at all times, namely the standard used in litigation: patents should be interpreted as one skilled in the art might interpret them. The article then suggests additional policies designed to help achieve the goals ascribed to the 'broadest reasonable construction' standard: rejection of poorly drafted patent specifications, the use of disclaimers and definitions to clarify vague claims, and a relaxing of the evidence required to find a patent claim obvious.