Putting back pressure on the cylinder is never a goo thing, it' can cause damage to valves and other important parts of the head. What people think is back pressure is something called scavenging. Basically there is a perfect point at which your exhaust flows just right to have an optimal exhaust velocity. This does not however mean that it is pushing gas back in to the cylinder or putting pressure backwards on the exhaust valves (the term 'back pressure'). What it does mean is that your exhaust flows free enough that the speed at which it exits is at optimum efficiency for evacuating cylinder and exhaust system space. If you go to big the velocity is lower (reread my example) because of physics. The same basic principal applies to headers/exhaust. You want the highest exhaust gas velocity and that doesn't mean the largest exhaust. As was stated, you reach a point of diminishing return and when the exhaust velocity lowers, you start to lose some power because you're not moving exhaust gases out fast enough. With the right diameter exhaust you create the right type and amount of pressure which moves forward through the exhaust system and not backwards.
I am running out of different ways to say it. Essentially, you have x amount of gaseous volume and it can go through a pipe that is too small, just right, or too large.
With the pipe that is too small, velocity may be the highest here but the actual flow rate is not high enough to satisfy the total evacuation of exhaust gas before the next combustion takes place, this forces gas towards the path of least resistance (usually slightly open or opening valves)
With a pipe that is just right, velocity in excellent and at the same time the pipe can handle the volume of exhaust produced. There is little to no measurable back pressure, exhaust gases continue to flow in the direction of least resistance and out the back.
With a pipe that is too large, exhaust velocity is decreased but volume capacity is larger. Even though the capacity is greater, there is less vacuum to empty the cylinders and once in-stream, it exits the tube at a slower rate.
I think I made things worse but the best way to demonstrate the proper effects of exhaust velocity is the straw demo.