Your analysis makes sense, and if I were to use his method I might be tempted to fix the two-step discrepancy.