Why Good Arguments Lose: Semantic Speciation and Memetic Fitness in Normative Concepts

Argumentative correctness is a necessary but insufficient condition for normative concept survival. This paper introduces a two-layer framework for understanding why: the Extended Phenotype Theory of Law (EPT) explains semantic speciation across institutional environments at the level of meaning; computational analysis of corpus data reveals a deeper, prior layer of lexical exit, in which high-rigor environments abandon a term entirely before low-rigor environments colonize it. Applied to the concept of equity across five institutional environments, the analysis produces five empirical findings. First, and most importantly, only 0.22 percent of abstracts in seven major political philosophy journals use the word equity (6 out of 2,732); the philosophical environment exited the equity lexical space decades before the DEI meme entered it. Second, the social doctrine environment (Vatican encyclicals, 13 documents 1891-2023) likewise uses equity zero times after 1965. Third, the public policy environment (World Bank documents, n=500) shows precision collapse post-2015, with mean intra-environment semantic similarity falling from 0.467 to 0.291. Fourth, the cross-domain academic baseline (n=2,000 papers across finance, health, and philosophy) shows stable low precision (P approximately 0.26), indicating that equity was semantically heterogeneous in academic usage long before the DEI debate. Fifth, Google Ngrams shows an equity/equality ratio inversion beginning in 2013, one year before the Froehle image began circulating, suggesting the book corpus was already shifting before social media dynamics accelerated. Together these findings support a lexical exaptation thesis: the DEI meme did not 1 displace philosophical precision from a shared term. It colonized a term that philosophy had already abandoned because it considered the term too imprecise. The Semantic Memetic Divergence (SMD) Index, operationalized through fitness (F) and precision (P) vectors, is introduced as a falsable measurement instrument for tracking these dynamics. The social media environment (I-3) was not computationally analyzed due to the absence of a verified corpus; it is treated qualitatively. All data and replication scripts are publicly available (https://github.com/adrianlerer/semantic-memetic-divergence). This record is part of the Law as Extended Phenotype research program. Replication materials are available at github.com/adrianlerer/semantic-memetic-divergence.