A theoretical analysis bridging the gap between the Memetics of the 1990s and the Transformer architectures of the 2020s. Over thirty years ago, the discipline of Memetics attempted to quantify cultural evolution using the tools of information theory, epidemiology, and systems dynamics. While the sociological application of memetics stalled in the early 2000s, its core mechanisms have been inadvertently validated and industrialized by the rise of Large Language Models (LLMs) and Generative AI. This paper revisits my foundational work from 1990–1995, specifically the concept of the L-meme (Linguistic meme) and the prediction of Postorganic Immortal Personas (PIPs). I demonstrate that the "George Washington Persona" I hypothesized in 1995—a predictive model reconstructed from an individual’s textual corpus—is an exact description of modern Generative Agents. By updating my original Volterra-Lotka and Zipfian formulations to account for Transformer-based architectures, I argue that while we have achieved the mechanics of Cybernetic Immortality, we have lost the humanity of selection, surrendering the evolutionary pressure of ideas to algorithmic optimization functions
Elan Moritz (Mon,) studied this question.