potentials, which limits their spacing to a few 10s per second. It is probably no accident that the “frame rate,” at which we can distinguish that movies are actually a sequence of stills, is about 40 per second. Thus, electronic processing is close to a billion times faster. e Size: The linear dimension of a typical neuron is about 10 microns. Molecular dimensions, which set a practical limit, are about 10,000 times smaller, and artificial processing units are approaching that scale. Smallness makes communication more efficient. e Stability: Whereas human memory 1s essentially continuous (analog), artificial memory can incorporate discrete (digital) features. Whereas analog quantities can erode, digital quantities can be stored, refreshed, and maintained with complete accuracy. e Duty Cycle: Human brains grow tired with effort. They need time off to take nourishment and to sleep. They carry the burden of aging. Most profoundly: They die. e Modularity (open architecture): Because artificial information processors can support precisely defined digital interfaces, they can readily assimilate new modules. Thus, if we want a computer to “see” ultraviolet or infrared or “hear” ultrasound, we can feed the output from an appropriate sensor directly into its “nervous system.” The architecture of brains is much more closed and opaque, and the human immune system actively resists implants. e Quantum readiness: One case of modularity deserves special mention, because of its long-term potential. Recently physicists and information scientists have come to appreciate that the principles of quantum mechanics support new computing principles, which can empower qualitatively new forms of information processing and (plausibly) new levels of intelligence. But these possibilities rely on aspects of quantum behavior which are quite delicate and seem especially unsuitable for interfacing with the warm, wet, messy enviroment of human brains. Evidently, as platforms for intelligence,