Evolutionary algorithms (EAs) have been widely used in evolvable hardware. The very term, evolvable hardware, reflects the importance and omnitude of EAs in this field. However, EAs have primarily been used as an optimisation or search tool, which can explore a large and complex space. While success has been demonstrated by EAs in exploring unconventional designs that are hard to reach by human experts, it is interesting to ask the question whether we have fully used all the potentialities of EAs. We argue in this paper that there is rich information in a population which can and should be exploited. The classical approach of evolving the best individual in a population may not be the best one. A truly population-based approach that emphasizes population rather than the best individual can often bring in several important benefits to evolvable hardware, including efficiency, accuracy, adaptiveness, and fault-tolerance.