Many scientific questions concern events or causes that cannot be observed directly. Investigators compare possible explanations with the evidence that remains.
An inference to the best explanation asks which proposed cause has the greatest explanatory power and scope, fits what is already known, and possesses the causal ability needed to produce the effect being explained.
This kind of reasoning does not make an explanation certain. A better explanation can later be replaced if new evidence appears.
The evidence assembled in these investigations permits several explanations to be compared rather than considered in isolation. Natural processes have demonstrated capabilities. Intelligent agents also have demonstrated capabilities. The question is which kinds of causes are adequate for the particular effects under investigation.
Judgments about the "best" explanation can be disputed. Different investigators may weigh simplicity, probability, explanatory scope, prior assumptions, and causal adequacy differently.
An inference is therefore not the same thing as a proof.
What criteria should be used to compare design with competing explanations, and what observations could cause us to prefer one explanation over another?
Intelligent Design is strongest when presented as a comparative inference rather than as a conclusion forced by a single unexplained phenomenon. The case must show not only difficulties for alternatives but positive reasons why intelligence has relevant causal powers.