The disc instability model successfully reproduces many of the observed properties of cataclysmic variables. However, additional ingredients such as mass-transfer variations, disc irradiation, stream-disc overflow, or inner-disc truncation must be included to explain certain systems. The physics underlying these processes is often poorly constrained, and our lack of knowledge is typically absorbed into extra free parameters, much like the $\alpha$-prescription for viscosity. In this paper, I examine how each of these ingredients affects the predicted light curves and discuss the limitations that arise from the growing number of unconstrained parameters on the model's predictive power.