Per-capita light use does increase. In order for Jevon's paradox to apply, per capita energy use would have to increase more than the decrease from improved eff

Per-capita light use does increase. In order for Jevon's paradox to apply, per capita energy use would have to increase more than the decrease from improved efficiency. So if LEDs are ten times more efficient, per capita light use would have to increase by 10+ times for Jevon's to apply. Putting it another way, Jevon's is just an example of the rebound effect where more energy is used after the efficiency improvement than before. It can happen, but it's not very common. If people were to switch to LEDs and their energy consumption dropped to a tenth of what it was, they could triple the number of lights in their house, leave those lights on 24/7/365 (as opposed to for 8 hours a day), and still be using less energy than with incandescents. That's the rebound effect. If they quadruple the number of lights and triple the time those lights are on, they would use more energy than they did before the switch to LEDs. That's Jevon's paradox.