Inflation's smoking gun has refused to smoke for forty years. Any inflationary epoch stretches quantum fluctuations of spacetime itself into a background of primordial gravitational waves, imprinting a distinctive curl pattern, B-modes, into the CMB's polarization, with amplitude parameterized by the tensor-to-scalar ratio r. The classic inflationary models that defined the paradigm predicted r between 0.01 and 0.1, comfortably detectable; the measurements have been a forty-year retreat: BICEP2's famous 2014 detection dissolved into Galactic dust, and the combined BICEP/Keck-Planck constraint now sits at r < 0.036 (and tightening toward 0.01-class sensitivity with current data), with no primordial signal at any sensitivity yet reached. Each improvement has executed another tranche of the model space: single-field monomials are gone, natural inflation is gone, and the survivors cluster at small r by design, plateau potentials engineered to predict what has not been seen.
The retreat is methodologically corrosive because r was the discriminator: the one signature separating "inflation happened" from "inflation is a parameterization." A paradigm whose flagship prediction migrates downward after each null, with model space always available below the current bound, risks unfalsifiability in practice, the criticism pressed by Steinhardt and others from within the field. Meanwhile the null carries information: whatever generated the primordial perturbations did so without an accompanying tensor background at the classic scale, a fact that fits awkwardly with the high-energy inflationary epochs that solve the other problems best.
The standing is a forty-year null hardening into evidence: CMB-S4 and LiteBIRD will reach r near 0.001, deep enough to exclude the remaining natural plateau models, and a continued null at that depth converts the absence into one of cosmology's most consequential measurements.