Memory access time in computer science should be modeled as O(N^α) instead of O(1), as it is affected by the distance to memory and the physical limits of signal transmission. This has practical implications, particularly in optimizing algorithms, such as those used in cryptography, where the size of precomputed tables can significantly impact performance. Understanding these nuances is crucial as computing approaches the limits of general-purpose CPUs and explores specialized hardware like ASICs.