IMHO the '83% rule' is a response to having conservative assumption piled on top of conservative assumption, but it doesn't itself have a good direct physics basis.
Article 220 dwelling calculations are known to be very conservative. Just look at the conductor and transformer sizes used by power companies.
Conductor ampacity tables are known to be conservative.
Most houses with 200A service (nominal 48kW capacity) operate at an average power level of 1-2kW.
On the other hand, breakers have a certain tolerance. A 100A breaker might never trip at 125A and be perfectly in spec.
In any case there are several analogous situations in the code, where the breaker is _not_ the weak link, and the conductors are considered protected by the combination of the breaker and the load characteristics. Look at the protection for motor feeders, for example, where the breaker has a trip rating much greater than the conductor ampacity and provides short circuit protection, and the motor overload device also provides overload protection for the wires feeding the motor. Or look at the 'next size up' rule, where you can use a breaker that is larger than the conductor ampacity as long as the calculated load is less than the conductor ampacity.
-Jon