To understand whether the population response expresses information better than the aggregate of the individual responses, the sum of the individual contributions is frequently used as a baseline against which to assess the population's coding capabilities. Using information processing theory, we show that this baseline is illusory: the independent baseline case is theoretically impossible to apply consistently to any population. Instead, we use as a baseline the noncooperative population, in which each neuron processes a common input independently of the others. Using the information transfer ratio, the ratio of Kullback-Leibler distances evaluated at a population's input and output to measure a population's coding ability, we show that cooperative populations can perform either better or worse than this baseline. Furthermore, we show that population coding is effective only when each neuron poorly codes information when considered out of context of the population.