Longwave atmospheric cloud-radiative effect (LW ACRE) is important in supporting a variety of spatially organized moist systems in tropical weather and climate, from tropical cyclones, the Madden-Julian oscillation (MJO), to Walker circulation. Our works aim to understand the fundamental properties and the processes that generate the feedback between LW ACRE and tropical precipitating convection. I will discuss two major findings. (1) We explain why LW ACRE only supports larger-scale circulations, as in the MJO but not in mesoscale convective systems (MCSs), by studying a field campaign MCS case. (2) Why does such scale dependency exist? We propose a minimal conceptual recipe to quantitatively describe the scale dependency of LW-convection feedback by simply assuming that anvil clouds, which primarily generate LW ACRE, act as long-term memory of precipitation. The scale dependency of LW-convection feedback is, thus, a natural emerging property of longer memory of anvil clouds than precipitation. Finally, I will discuss how this conceptual model may provide information of how anvil LW ACRE is generated in tropical convective systems.