Joe,

Your reasoning is exactly correct -- friction is greater in the smaller pipe, restricting the overall flow. But flow isn't continuous; it's pulsed -- and we'd like low pressure behind one departing pulse to assist the following cylinder's flow. As long as pressure down the pipe can be evacuated between pulses, the fast momentary flow near the headpipe(s) leaves lower pressure behind it than slow flow would -- and the flow velocity at that point is determined by the diameter of the pipe.

In other words, the small pipe is better at low overall flow volume, but builds pressure at higher flow volumes that kills the benefits (and then some). The large pipe doesn't provide much benefit at low flow volume, but hits its balance between velocity and pressure at higher flow volume.

If we have a header that's well tuned to do all that work, we don't want the rest of the system to mess with it.

Does that clarify at all?