Examining arguments from design › Even a designer would leave questions open

Part V presses the consequences of Cleanthes's method. If we infer a cause from its observed effect, the effect limits what we are entitled to attribute to the cause. An effect available to our examination does not establish unlimited power merely because a familiar religious description includes it.

Unity is another problem. Human productions often involve several people. If the world is compared to an artifact, why infer a single maker rather than several cooperating makers? The analogy cannot both borrow the authority of human craft and ignore features of that craft whenever they produce an inconvenient possibility.

Perfection is similarly difficult to infer. Philo proposes unsettling possibilities: a world might be an inexperienced maker's attempt, the work of a subordinate intelligence, or an imperfect product compared with worlds beyond our knowledge. He is not presenting these possibilities as discoveries. They expose how many different accounts remain compatible with a loosely specified designer.

A modern illustration makes the limitation easier to see. Suppose an unfamiliar object clearly shows evidence of deliberate assembly. That evidence might tell us that someone worked on it. It would not, by itself, tell us whether one person or a team assembled it, whether it is their best work, or whether its maker is morally admirable. Those further conclusions require further evidence. The illustration reconstructs the inferential gap rather than supplying a new case from Hume's text.

Cleanthes replies that Philo's possibilities continue to involve design. He can treat that concession as preserving a significant conclusion. Philo's challenge nevertheless remains: what makes that conclusion sufficient for religion? The debate concerns both whether intelligence is inferred and how much religious content the inference can carry.

The strength of the inference and the richness of its conclusion can therefore come apart. An argument might provide a defensible reason for a modest claim while offering little support for the further claims its proponent values most. Testing the modest claim and testing those additions are separate jobs. Treating them separately allows a reader to grant something to Cleanthes without granting everything he wants from natural religion.