In my personal experience, anything done to reduce back-pressure or increase flow after the collector is free horsepower assuming the fuel curve is there. All engine/exhaust combinations have dips & peaks in the power curve resulting from that particular combination of primary tube diameter & length as well as collector size & length. As most of you already know, the first consideration header builders design for is for max flow in the primary tubes - they have to be big enough to eliminate restriction while maintaining the velocity needed for scavaging as much burnt gas as possible. If only that was all you had to worry about. Turns out the exiting gas also creates a pressure wave that turns around when it reaches the "end" of the pipe and heads back toward the engine. When it gets back to the engine it disrupts/slows the velocity of the still exiting gas and cause a dip in the torque/power curve. A "tuned" exhaust is one where pipe diameter & length is adjusted so this happens (hopefully) at a point in the power band away from the needed power curve, as well as achieving maximum scavaging where most needed. Anti-reversion lips, stepped primaries, coatings to keep heat in and velocity up, are all part of the arsenal used to get that last horsepower. Builders' these days use sophisticated computer programs that consider all this in relation to intake-cam-head flow before spitting out design parameters.
I mention all this because 98% of this happens before the end of the header collector. On a stock bike, almost every consideration after that point is meant to meet noise laws (or pollution laws in the case of the catalytic converter). Increasing flow at the can by reducing back pressure can add a few horsepower without significantly impacting any of the header design characteristics, but may creat a lean situation at some RPM's that require additional fuel. Virtually every a/m can has significantly less back pressure and more flow than most stock systems. Interestly, one of the reasons the 2008 exhaust is so huge is to try to quell the noise without adding unneeded back pressure. But if you don't care about noise,achieving this gets a lot easier.
A racing exhaust can be a significantly different animal since it's designed from the start for max power at high RPMs. In some racing exhausts, a portion of the midpipe & canister is actually an extension of the header collector. The circumferance and length of the collector dramatically impact peaks/dips in the power curve, but this doesn't really apply to the stock exhaust the same way it does to a 4 into 1 or a 4:2 racing exhaust. Look at Brock's exhausts to illustrate an example, You can get his 4 into 1 header with A/H, meg, or std canister with similiar power but requiring somewhat different maps. The point being that even on those hi-po systems the canister has a lot less to do with power output than the header design. Also consider the 1/2 dozen a/m systems where you can insert plugs or additional baffling to reduce the noise. None of them claim to increase power by adding a flow restrictor which increases back pressure. Although some do claim not to lose much.
There is a difference between the design parameters of a stock exhaust and a racing exhaust. One may react differently than the other to the same modification or change. Pure science aside, sometime's it cool to just try something different.