That seems a rather flawed way of viewing the density, since that key represents 2^256 possible combinations. So if you take a way a bit, the density by that measure goes up 0.3%. Practically speaking, it should double.
Why don't we instead calculate the density as the expected value of guessing one combination. In other words, let's say I have stolen an ATM card and can wipe a bank account containing $100,000 by guessing the 4 digit pin:
$100,000/(10^4) = $10/guess
Or in our case:
$161,590,330/(2^256) = $1.39e-69/guess
Not so impressive now.