The empirical literature on the economics of patents employs a large array of patentbased measures to proxy economic characteristics of patents such as the "sciencebase", "importance" or "value" of patented inventions. Although scholars have attempted to assess the validity of these proxies by relating them to e.g. inventor surveys, the appropriateness of these indicators is currently heavily debated. We contribute to this discussion by relating a group of obviously “wacky” patents to two control groups. If the patent-based indicators are appropriate, they should unambiguously identify the “wacky” patents. We present descriptive statistics and run probit regressions to evaluate the performance of commonly used patent-based measures. Our findings show that forward citations are good predictors of importance. However, the "wacky" patents have higher originality, generality and average citation lags than the controls, which suggests that these indicators should be interpreted carefully. At best, scholars should provide an external validation for their interpretation of patent-based measures.