Assume there is a function `rand_bit()` that returns `0` or `1` with equal probability, independently on each call. A standard way to approximate a random number uniformly distributed on `[0, 1)` is to use the generated bits as the binary digits after the decimal point. If the first `n` calls return bits `b1, b2, ..., bn`, define `X = b1/2 + b2/4 + b3/8 + ... + bn/2^n` In this coding version, the bit sequence is given to you directly as a list `bits`, and you must return the corresponding value `X`. This matches the simple algorithm for the original randomized problem: call `rand_bit()` exactly `n` times and interpret the results as a binary fraction. Using `n` bits produces one of `2^n` equally likely values in `[0,1)`. The approximation has resolution `1/2^n`, and the truncation error compared with the full infinite binary expansion is always less than `1/2^n`.