The heart of the proof was showing that the lumps in rounding get out of balance, thereby preventing the pure logarithmic number of steps. But if you accept the smallest lump, it does enough to hold the count.