Barry Arrington has started a new thread over at UD, where he argues that it is perfectly okay if CSI (complex specified information) cannot be quantified.

I responded, in a comment, that he seems to be making the case that CSI is not a scientific concept.  It occurred to me that folk here might want to have a discussion on what is required of a concept, for it to be part of a scientific theory.

It seems to me that, at the very least, the concept has to be defined precisely enough that one can form testable hypotheses.  Moreover, these hypotheses need to be testable by independent researchers, and the tests need to provide some kind of reliability (i.e. reasonable agreement between results obtained by independent researchers).  Perhaps that’s weaker that quantifiable.  However, I think even that weaker requirement is a problem for CSI, as it is currently “defined”.

The topic is open for discussion.