4 ms·

Now and know sound nothing alike, so hashing by character composition would get you much closer than the hypothetical native speaker.