This is a guest appearance of Erik Tellgren.
I (Mark Perakh) have not contributed anything to this essay and am posting it as a courtesy to Erik.
Here starts Erik’s text:
William Dembski has been one of the most influential contributors to the Intelligent Design (ID) movement. Among other things, his work has added the terms specified information, specified complexity, and complex specified information to the basic vocabulary of the ID movement. These terms are all directly related to the logarithms of special types of probabilities, e.g. the probability of a pattern of interest given that it was produced in some way that excludes the foresight and guidance of an intelligent agent.
In a recent draft manuscript, Dembski and his coauthor Marks extend the vocabulary with three new terms : endogenous information, exogenous information, and active information. They consider as given a search space and a fixed subset, called a target, that makes up some fraction ps of the search space. An issue of interest to them is how to measure how well a search algorithm  exploits the structure of the search problem. Two possible candidates are the probability p that a search algorithm is successful and the ratio p/ps. Readers of Dembski’s previous writings will not be surprised to discover that Marks and Dembski prefer to log-transform their probabilities and rename them ‘information’. In equations, their definitions are
endogenous information = -log2(ps),
exogenous information = -log2(p),
active information = -log2(ps/p).
Continue reading Comments on Active Information at Talk Reason