mark, i once thought exactly the way you did. in fact, when i started futzing around this past weekend, i fully expected the boost to be uncontrollable. imagine my surprise (and joy) at finding out it was not.
when the solenoid is not energized, the pneumatic circuit is "shorted" -- the line from the nipple goes through the solenoid and right to the WGA. the common port of the solenoid goes to the nipple, the NC port is capped, and the NO port leads to the WGA.
when the solenoid is energized, the pneumatic circuit is "opened" -- ie, the line from the comp nipple to the solneoid goes NOWHERE, and the line from the solenoid to the WGA goes NOWHERE. the only "control" of boost comes from the ball+spring MBC.
i would say that there was reverse flow back through the ball+spring mbc, but when i first got it i tried to blow back through it, since i have found that the best ball+spring MBCs do NOT allow backflow. i was happy to find that it made a very admirable seal to reverse flow, so i cannot say that appreciable pressure is being lost via that route.
the last possibility is that the WGA itself has a small bleed built into it. i haven't tried the very scientific "put mouth on hose and blow" method of testing for that yet, though, so i can't tell you anything more. all i can say is that boost is more solid than ever with NO discreet/discernable bleed hole on my setup, and it doesn't "latch up" in the "limit boost" condition.