Three falsifiable predictions from a zero-free-parameter model of HIV escape from the eCD4-Ig decoy

A network engineer, not a virologist, modelled the eCD4-Ig decoy as ablack-hole route — a destination that accepts the packet and discards it —and asked what it would cost HIV to route around it. The model refuses curve-fitting. Every rate is taken from publishedmeasurement (Perelson 1996; BioNumbers 105728; the 1-day eclipse phase).Unknown quantities are not guessed: when a number is bounded and small, all99 possible values are swept, and the answer becomes a bound rather than anassumption. THREE FALSIFIABLE PREDICTIONS 1. Escape cost above 73 percent. Sweeping all 99 fitness costs, hardenedacross mutation rates from 1e-3 to 1e-8 (693 runs), the observed in vivoprotection is reproduced only above 73 percent; the bound moves by 2 points.Pixel-level reading of Fig. 5 of Fellinger et al., J Virol 2019 (PMC6600210)gives 92 to 97 percent for Q428K alone. This prediction holds. 2. Serum neutralisation of the compensated variant at least 67 percent. Atthe roughly 18 percent cost read for K171E/N301D/Q428K on Cf2Th cells,protection breaks unless serum neutralises at least 67 percent of it. Fig. 6gives about 75 percent for Q428K alone and about 47 percent for N301D alone,against about 0 percent for wild type; the triple variant cannot be separatedat the published resolution, and the missing value is reported as missing. 3. Decoy clearance at least 76 percent. Modelling analytical treatmentinterruption in a virologically suppressed person — reservoir reactivationcalibrated on the published 16-day median to above 50 copies/mL, immune brakecalibrated on the 32-day median to above 10 000, the 21-day median to above400 left as a prediction and landing at 18 days, inside the publishedinterquartile range — rebound is abolished only if the decoy clears at least76 percent of virion output. The transition is sharp: 70 percent stillrebounds at day 127; 77 percent never does. WHERE THE MODEL FAILS, STATED BY THE AUTHOR In primary infection the model requires target cells to fall to about1 percent to turn the acute peak over, against a measured nadir near 304cells per microlitre: it is wrong there. The immune brake improves the10 000-copy rebound mark by 11 days but changes the 400-copy mark by zero, anunexplained gap. A pass threshold that had been chosen after seeing the errorit needed to accommodate was replaced by published interquartile ranges,which the author did not choose. This work demonstrates nothing about HIV. It states conditions under which amodel reproduces published measurements, and offers three numbers that canrefute it. All code, the source-by-source provenance of every figure used, and the rawunedited output are included.