When developers face enormous pressure to get a first implementation working quickly, they could commit to an iterative implementation strategy that starts with a throwaway prototype, adding support for exact GC in a later design. Although this approach is sound in principle, in practice the temptation to hold on to the prototype is too great for most development teams. This conundrum is vivid in the example of Perl, which has been intending to move to an entirely new engine (Perl 6) for over ten years now. We argue that the software engineering discipline of exact GC is not burdensome when followed from the beginning of VM development and actually improves compiler correctness as well [2, 18]. The development history of scripting languages such as Ruby, AppleScript, and Perl make a compelling argument for early investment and deferred gratification.