In his book "Unweaving the rainbow", Richard Dawkins explains how the use of analogies and metaphors can be a useful tool to understand scientific ideas and concepts. He also describes why doing this can sometimes be bad and actually make something more difficult to understand and even make you understand things wrong. Dawkins himself refers to this as using either good poetry (when better understanding is reached) or bad poetry (when understanding is hindered) respectively. Reading through intelligent design (ID) material on the web, it strikes me that ID proponents are quite fond of bad peotry. Most importantly, they seem to think that analogies are actually a good descriptor of the way things really are. The analogies become, then, not mere analogies but literal truth. The best example I have seen so far is from none other that William Dembski:
William Dembski quotes an Ivan Amato who talks about mutations that are usually thought to be silent (i.e. have no phenotypic effect): "The more scientists study the genetic code, the more it reads like poetry. In a poem, every word, every line break, even every syllable can carry more than a literal meaning. So too can the molecular letters, syllables, and words of the genetic code carry more biologically relevant meanings than they appear to at first." Amato's poetry is "good" (heck, he even used poetry as an analogy) IMO, but of course Dembski has to take it literally. The reason is obvious: poets write poetry. Poets intelligently designed the poems. But it's just an ANALOGY. Bill took a good piece of poetry and presented bad poetry to his readers.