This talk will revisit some classical "decomposition space shrinking" arguments like Mike Freedman has been discussing in his current lecture series. I'll focus on Freedman's revolutionary 1981 argument, which has come to be known as stated in the title above. The proof is best appreciated by first looking at the most elementary cases, where the number of non-trivial point-inverses is assumed to be one, then two, then three, by which time the process of proof should be clear. Nice pictures too.