Simulation
a = 0.00705
p = a · k = 0.0282
| akn = 0.423
η = k · exp(−akn) = 2.63 / 4 bits (65.5%)
η = k · exp(−akn) = 2.63 / 4 bits (65.5%)
akn = 0.42 — within capacity
Watch how per-step-per-bit errors accumulate. Each cell is one bit at one step. At every step, each surviving bit independently corrupts with probability p = a · k.