There once was a class V safety level. It was decided eliminate it to reduce the complexity of the levels and make IV cover everything. If you ever get to work with anything well into IV, ie in the tens of watts region, you will quickly see why four is enough.
Its bad enough having to remember the new IIIM and IIIR classes for certain light sources as well.
If you read deep enough into Dr Sliney's books, you will find that the levels are based on long ago animal tests and cadaver tests, statistics, and built in safety factors, done in the 1970s.
IIIAs levels might be chosen for ensuring less then a 1 in 9,999 incident probability of detectable damage. One day I will go back and look up what that number really is.
A warning for the simple minded. Just because its only a 1 in 9,999 chance of damage, don't assume you can try something five times and quit. The math and the reality does not work that way.
Its still Russian Roulette to do so.
One should not suggest changes to existing laws that work. Back then engineers and scientists wrote the laws, lawyers cleaned up the language, and it was good. Now, a lawyer consults a engineer, laughs, and then writes what the highest bidding lobbyist wants. Often Not good!
The OP does have a valid point, the 4.95 mW of IIIA could possibly be taken up to say 15 mW if the beam size is not too small. However the safety factor starts to erode. Real damage thresholds are in mW per square centimeter, but for simplification and ease of understanding, it was decided to use total beam power.
Steve