Head porting is not a black art. It's basic physics.
It's all about matching piston CFM demand at a targeted RPM while making sure the port isn't too slow or too fast.
Often when people think of porting causing the an increase in RPM, it's because material was removed, making the port slower - THUS the engine has to spin higher to equal the same port velocity.
You can port to increase efficiency. This will increase CFM and increase velocity (and decrease it in other areas) without removing so much material to cause the engine to want to peak higher.
In other words the engine will continue to rev higher and higher until it reaches a point where flow goes sonic (air reaches the speed of sound) somewhere in the port, and the amount of mass the engine sees is level despite an increase in pressure drop in the cylinder (piston CFM) demand. This is the point where engines nose over.
If the port is bigger, the it will require more piston cfm demand (rpm) to reach this point.
If the port is too smaller, it will occur at a lower point.
The 4 cycle otto engine is happiest with velocities in the 280-320fps range @ 28" H20 depression.
The straighter the port, the higher the average velocity can be.
The flatter the port, the more you have to slow down the air around the short side radius to get it to turn.
Valve shape and valve seat angles are critical.
There is a lot of good reading on the above elsewhere (there is also a lot of bad reading about it too). I'll post some links when I get a chance, but you should look into buying this book:
http://www.amazon.com/Practical-Engine-Airflow-Performance-Applications/dp/1613251572
Here is my head (SR500):
The head was sized to peak at 7200rpm with an average velocity of 300fps (1.77in^2 MCSA)
I graphed the flow numbers:
You can use a program like PipeMax to help with the math:
http://www.maxracesoftwares.com/forum/viewtopic.php?f=13&t=242
We tried several valve shapes, valve seat angles, etc. The biggest problem we had was port stability, but Nick (my head porter) spent a lot of time working on the short side radius. The port never backs up now and it is quiet on the bench. The depression numbers on the bench are stable as well. You look for this because a running engine will see higher depressions than most flow benches can flow at. If it's not stable at 28", it won't be at higher depressions. Our limiting factor was actually the casting. He ran out of room on the sides of the ports over the short side - so what you see there is about as wide and flat as the short side can be without adding material to the outside of the head.
As for the exhaust, you do NOT want to put a restriction right off the port. Match the port area to the tubing and carry that for about 8" before stepping up. The trick with the exhaust is to make it as small as you can without restricting - this requires using large 6" bend radii.. anti reversion should not happen at the head. which is partly what the D-shape port is trying to accomplish. The other is the ports on the vintage Yamahas are much too big for any power the castings could withstand.
I'll post a lot more later when I'm not pressed for time, but in the mean time, I have a lot posted here:
http://www.xs650.com/forum/showthread.php?t=25773&page=3