Channel-Only Contraction Coefficients for BSC Tensor Products: Exact Formulas and Monotonicity [Erratum: Not JSD-Specific]

**ERRATUM (March 2026):** The main results of this preprint — the exact formula for η(BSC(δ)⊗2), monotonicity in blocklength, and convergence to 1 — are mathematically correct. However, the original framing is misleading. The channel-only contraction coefficient η_f(W) := sup_{P≠Q} D_f(PW,QW)/D_f(P,Q) equals η_χ²(W) for ALL f-divergences and all channels (via Ordentlich–Polyanskiy binary reduction). Therefore super-tensorization is universal across all f-divergences, not JSD-specific. The contrast with χ² multiplicative tensorization (Equation 2) conflates the channel-only and source-dependent definitions. The formula η(BSC(δ)⊗2) = (1−2δ)²/((1−δ)²+δ²) is correct but follows equally from a standard Rayleigh quotient computation for χ². Original abstract preserved below for reference. --- This work studies the tensorization behavior of the strong data-processing inequality (SDPI) contraction coefficient for the Jensen–Shannon divergence. For the binary symmetric channel BSC(δ), an exact closed-form expression is derived for the two-fold product channel, showing strict super-tensorization. It is further proved that the contraction coefficients for tensor powers are nondecreasing in blocklength and converge to one. The underlying mechanism is identified as the non-factorization of the midpoint distribution under product measures.