We consider nine simple cardinality estimators, seven from previous work, and two new ones, and evaluate their performance with respect to plan quality using the Join Order Benchmark (JOB). Three estimators from the literature rely on the frequently used independence assumption and are prone to underestimation. In contrast, five estimators always overestimate and promote pessimistic plans. Our major finding is that all cardinality estimators produce relatively good plans. Surprisingly, the best contender uses the independence assumption under rule L.