How many patents are truly valid? Extent, causes, and remedies for latent patent invalidity

Abstract

A substantial number of patents tested in court for validity are invalidated. If a similar portion of all patents was deemed invalid in hypothetical validity tests, then this would indicate a seriously flawed patent system due to restrictions unduly imposed by these erroneously granted patents on users and follow-on innovators. Thus, we ask, if a randomly picked patent underwent revocation proceedings, what are the odds of its invalidation? We address this question by analyzing the various selection effects through which patents become subject to validity decisions. Empirically, we focus on Germany, where revocation proceedings are separate from infringement suits and where, in court decisions during the period of 2010 to 2012, 45% of patents were determined to be fully invalid and 33% to be partially invalid. Based on data gleaned through expert interviews, a survey among lawyers, and an econometric analysis of court judgments, we find the likelihood of (hypothetical) invalidation of a randomly picked patent to be in the same range as that for actually adjudicated patents. As the main cause of patent invalidity we identify incomplete searches for prior art during examination. Our arguments carry over to other legislations. To remedy this situation, we suggest a significant increase of the inventive step required for patent grant combined with a smaller increase of the inventive-step standard in litigation.