OK I'm genuinely curious now.. wouldn't the fact that a computation has taken place mean that energy has been used to do work (non-heat)? My intuition tells me that it would not be possible to simultaneously use energy 100% to generate heat and do something else at the same time. For example if you used an electric motor to move a weight against massive friction, it could be that 99% of energy is lost to heat due to resistance in wires and friction, but if the weight moved, it would mean that not all energy was lost to heat.
Yeah, but in the computer example you are doing the equivalent of then lowering that weight again (possibly getting 1% of that gravitational energy back) : we're assuming that the energetic potential of the bits across the computer stays on average the same across enough time.
