CSI as defined by Dembski is only identified by ruling out all non-design explanations, showing that they are too improbably to be accepted.
The problem is you are removing things bexcause you think it is improbable, and on the other hand you are adding a inteligent desighner that has no evidence to support him but the theory you are using. Every theory you are trying to rule out as being improbable has evidence to support it yours does not so what is more improbable.