What you have here would be known as observed ASC, since conditional Kolmogorov complexity cannot be calculated. But, it seems a valid implementation of observed ASC, so I agree to it.

Quick background on ASC for those new to the subject:

Algorithmic Specified Information (ASC) is defined as ASC(X, P, C), where X is the event being analyzed, P is the probability distribution that we think generated X, and C is a context for describing X that is independent from P. It is calculated as:

ASC(X, P, C) = -\log_2(P(X)) - K(X|C)

where P(X) is the probability of P generating X, and K(X|C) is the conditional Kolmogorov complexity based on a prefix Turing machine, so K(X|C) obeys the Kraft inequality:

\sum_X 2^{K(X|C)} \leq 1.

Because K(X|C) follows the Kraft inequality, the probability of generating B bits of ASC from P is less than 2^{-B}.

Kolmogorov complexity is uncomputable, so in an empirical setting we have to estimate an upper bound using compression. We denote the compression of X given C as L(X|C). When we use a compression algorithm to estimate K(X|C), the resulting metric is known as Observed ASC (OASC):

OASC(X, P, C) = -\log_2 (P(X)) - L(X|C).

And since L(X|C) \geq K(X|C), and thus also obeys the Kraft inequality, then the probability of generating B bits of OASC from P is also less than 2^{-B}.