# Understandable modularity > ToDo Properties such as recursive decomposition and composition, or modularity (hereafter, modularity), are generally considered good. But what does "good" mean here? What is good modularity when human comprehensibility is left out of consideration? ## What does good mean? "Good" mixes meanings such as moral goodness, fitness for a purpose, and beauty (see the Archaic–Classical period in ). Goodness in engineering design also mixes several meanings, and one of its most important properties is comprehensibility. Recursive decomposition and composition allow humans to understand complex systems reductively, while modularization lets us hide complexity. --- The moment we say that "goodness" is independent of humans (or other sentient beings), it must be defined in another way that contains no value judgment. For now, let us define "goodness" as "a property that increases the likelihood of existence." The claim then means that "things with properties such as recursive decomposability or modularity are more likely to exist in the world than things without them." Connecting this to Dawkins's "survival of the stable" lets us interpret the claim in an evolutionary context. Things that can produce relatively accurate copies quickly are more likely to exist, and evolution occurs because they compete for limited resources. We can think about evolution in two dimensions: increasing fitness in a given ecology, and increasing the ability to maintain high fitness (evolvability). By analogy, these roughly correspond to parameters and hyperparameters. Because recursive decomposability and modularity enhance evolvability, products of evolutionary design often have these properties. A better test would examine whether these properties repeatedly appear in evolved things (mainly organisms), rather than whether systems with radically different cognitive structures from humans discover the same structures. The recursive decomposition or modularity discussed here need not fall within the scope of human understanding. Tangled recursive hierarchies or strange module boundaries that humans cannot understand are probably more likely to be "good." The recursive hierarchies and modularity that humans understand may be mere approximations of those actually found in nature, much as naive physics or folk physics approximates actual physical laws.