Diagnosing Mycoplasma pneumoniae (MP) has followed three technological paths — seeing the organism, reading the host's antibodies, reading the pathogen's genome — and the winners were rarely the best assays. Culture was analytically supreme and never adopted; serology ruled the clinic for fifty years despite documented misclassification; PCR outperformed both by 1990 yet took a decade and a half to reach routine use. We reconstructed 120 years of MP diagnostics as a diffusion dataset, scoring each technology on Rogers' five adoption attributes against its contemporaneous competitor, and tested the scores against regulatory archives (FDA, CE-IVD, China NMPA; search concluded 30 September 2026). Every documented displacement is consistent with four empirical laws: relative advantage in time-to-result outweighed analytic accuracy; workflow compatibility decided more contests than sensitivity; displacement occurred when a successor's turnaround crossed the clinical decision window; and adoption lagged roughly a decade behind technological maturity (mean 8.5 years across the eight measurable transitions) — a severity-gated lag that relaxed measurably after the 2023 resurgence. Applying the validated rubric prospectively, we predict that CRISPR-based point-of-care testing will obtain first regulatory approvals (NMPA before FDA) by end-2027 and achieve decisive adoption in pediatric emergency settings by end-2028, conditional on extraction-free, sample-to-answer workflow, while metagenomic sequencing consolidates in severe and atypical niches without crossing into first-line use before 2030. The predictions are falsifiable and dated — the archive is offered as an evaluation instrument, not an opinion.