This example came to me as part of our responses to a discussion of "terrorists as a virus."  I quote the writer (Don Patterson) with permission ...
 

Don Patterson wrote (September 27, 2001):

I, too, think metaphors are indispensable.  Scientific Method is based on recognition that all things we think we know are actually only metaphors for a reality we can never see.  And a "black box" type of metaphor is often amazingly useful for doping out how a complex system works.  What I object to is the strained use of metaphor by people who seem to know less about the appositive of the metaphor than they do about the subject.

If I were answering [the proposition that terrorists are like a virus], I'd point out that the only proof a metaphor is accurate lies in its predictive power.  Does equating terrorism with a virus allow us to predict anything?  Possibly.

First, though, we have to fine-tune the metaphor, [regarding] a logical error: viruses are not the same thing as viral vectors. If terrorism can be well described as a virus, terrorists cannot. They instead carry the disease.

There is a well-documented and widely observed protocol for addressing known vectors of a dangerous disease.  Those infected with a deadly virus should be identified, quarantined if the virus is highly infectious or causes violent behavior, and treated in such a way as to halt the spread of the virus.  Secondarily, the infected individual should be protected from his own irrational acts.

Rabies is a good case in point. The virus that causes rabies has a predilection for nervous tissue. In the infected individual it
causes "cerebral dysfunction, anxiety, confusion, agitation, progressing to delirium, abnormal behavior, hallucinations, and
insomnia." This is hell on the poor mammal infected. It's no picnic for those whom the mammal encounters, either.

Who would want to have a large agitated, sleepless, hallucinating dog wandering the streets? When faced with such a dog, few people would place its welfare above their family's. Even fewer would advocate letting the dog roam until society properly addressed the causes of the disease. Throwing meat to it would be a useful option only if one wished to divert its attention.

Even essentially benign carriers of rabies, such as bats and squirrels, could not be tolerated if their behavior posed a threat to surrounding populations.

Imagine then that terrorism is a virus like rabies: it wreaks havoc on most of the infected and on those with whom they come in contact.  It also has its carriers, like Bin Laden.  In many respects, the metaphor works, except that, unlike the carriers of rabies, the carriers of terrorism suffer themselves from the disease to the extent that their behavior becomes dangerous to others.

Since there is no antiserum to terrorism, the only thing left us is to capture and quarantine the infected or, when that is not feasible, kill them. This is an operation on just as high a priority as finding and minimizing the causes of the "disease."   

Now, if this metaphor does not seem to account for all the outcomes [one] would like to see, then … change the metaphor.