A special case of Godwin's law?

In discussions of software engineering methodology, unless you're actually working on life support or military targeting systems, the first person to bring up the space shuttle     loses the argument.

I'm not entirely sure why comparisons to physical engineering just won't go away.  To compare the task of engineering say a piston   for an engine block and a data access component of a web site only serves to invite comparisons of the number of things we know about the piston and the number of things we don't know about the data access component.

For the piston, we understand all the relevant properties; it's tensile strength, melting point, elasticity, coefficients of friction, etc.  The only thing we don't know until it's tested is whether this particular instance has a manufacturing defect.

On the other hand, when we call a method we don't know its performance characteristics, it's memory characteristics, it's reentrancy, or thread safety characteristics. Possibly some of these might be documented, but that only goes to highlight another critical difference - the number of parts and interactions and sane engine block vs. a modest piece of software are of different orders of magnitude.

I'd like to find another industry that makes for good comparisons to software engineering.  Law perhaps.