Martin Rinard’s group at MIT has achieved significant peformance gains in a huge variety of applications by a very simple optimization: simply skip every other iteration of certain loops. This is a brilliant idea, but it’s also a symptom of huge problems in computer science. We should always have the option be more flexible about how we process information in artificial systems. Computer science is culturally opposed to this; certain philosophers as well as much of the lay public presume this to be a technical limitation (that is to say, a fundamental limitation of technology as opposed to humans), but it’s not.