A little more to the discussion. When the exhaust valve opens there is a supersonic or near-supersonic pulse that shoots down the pipe ahead of the exhaust flow, mostly along the wall. When it reaches something, like a baffle or even the end of the pipe, part of it is reflected back up the pipe where it creates "back pressure". But it's not really back pressure, just what most people think of when they hear the term. Back pressure isn't the enemy, it's this reflected pulse or reversion. The best example is on a 2-stroke engine where the pipe length and shape is tuned to have the reflected pulse arrive just before the exhaust port closes. Two cycles being what they are, a little air-fuel mixture goes out the port before it closes (why they run dirty), and the pulse pushes it back in. It's a very low pressure forced induction. The problem with 4-cycles is there isn't any air-fuel mixture coming out the exhaust in most cases (you people with massively overlapping cams aside), just stuff you want to get rid of, so the pulse just pushes exhaust back into the cylinder. Slash-cut pipes aren't just for show. The hope (and it actually works) is that without a sudden end to the pipe that a straight cut one has there is less of a reflection. Welding lolli-pops in the outlet does the same thing, you vary the reflection until you find the place where things cancel each other out the best. Torque cones, along with increasing port velocity, keep the exiting pulse off the side wall and keep the reflected pulse, which is along the wall, away from the port.
The step in the header is doing the same thing, it breaks up the pulse clinging to the wall on the way out, and gives a surface on which a return pulse can reflect back off of. No covalent bonds or Brownian motion involved!