I would tend to believe what the CMP-5 was refering to was a single steel beam with support posts of wood...which is often installed in a single family dwelling.
If George Little's steel I beam supported by a steel post set on concrete in contact with the earth were to be exposed to a ground fault and it wasn't bonded according to 250.104(C)...there would not be a low-impedance path to carry the ground fault current, which would not allow the over current device to open.This steel beam would be live.
In a industrial or commercial enviroment, this steel may not be likely to become energized, for hopefully only qualified personal would be installing wiring apparatus nearby.
But in the residential world it may be likely to become energized by some of the wiring I have seen by DIY homeowners.I just had to clean up a basement with extension cords running everywhere and the guy had adaptors in his keyless with 2-wire lamp cords strung all over the place, wrapped around gas piping and strung over metal duct work.
I still believe after all that has been debated here that George Little's (exposed) steel would have to be evaluated by the AHJ to see if it were to present a hazard.
As for the wording change to 250.104(C) from "steel" to "metal"...I think that this wording change adds a whole new list of bonding requirements.
Metal: any of class of elementry substances that are good conductors of electricity.
Steel: a hard,tough alloy of iron with carbon.
You see, the word METAL does't just limit bonding to steel.
Structure...the first sentence of 250.104(C) mentions exposed STRUCTURAL steel. (2002 NEC)
As far as some of the other equipment mentioned here such as the instant hot water or the heated towel bar...those pieces of equipment must be grounded(not bonded) according to 250.134 or 250.138
shortcircuit
