ID Metrics and an Active Information Tutorial
|January 20, 2012||Posted by johnnyb under Intelligent Design|
One of my favorite parts of ID is the fact that it is creating good tools for biologists to use. ID is often misconceived as a conclusion about whether or not X was designed. Instead, ID presupposes only the *possibility* that something was designed, and that intelligent agents are not mechanistic. In accordance with this, several metrics have been developed.
The first metric that I am aware of is CSI. The method for measuring CSI was originally developed by Dembski in The Design Inference. The main problem for CSI is in the difficulty of actually taking the measurements it requires.
The second metric (well, metric probably isn’t quite the right word, it’s a qualitative measure) is Irreducible Complexity as described in Darwin’s Black Box. As originally proposed by Behe, Irreducible Complexity is fully testable, and has been successfully tested by Minnich and Meyer in the lab. While Irreducible Complexity as proposed by Behe only conceptually argues against Darwinism, further theoretical work shows more specifically, based on computational principles, why Irreducible Complexity argues for intelligence, as well as practical uses of ID in biological research.
The third metric, however, is my favorite. It’s a simpler conception, yet very powerful, and is based directly on the No Free Lunch theorems. It is “Active Information”. Active Information is basically the measurement of how much information a search algorithm knows about the pattern of the search space that it is searching. It is measured by looking at the performance of the search algorithm vs a blind search. The paper describing it is here. This concept has been further applied to measure the amount of active information that is used by the immune system during somatic hypermutation (about 22 bits), and additional research is ongoing to apply it more generally to cells in hypermutable states.
Anyway, Active Information has a huge potential in biology to help detect which processes have frontloaded information, and how much information the cell is actually supplying for mutational processes. Anyway, below, Robert Marks gives a *great* lecture on information generally, and ends the lecture specifically talking about Active Information in evolutionary systems.