It took me a lot of work to optimize the cost down to this point, and almost as much work to get a 2.5-width solution working in the first place. Still, it works! I can't see any way to save more money on this (maybe somehow getting the product assembly done with only one bonder? Or finding a way to remove a track from the top arm; I don't think you can save anything in the input disassembly arm), but there's definitely cycles optimization that can be done beyond that which I already did. Still, for a first time doing a tournament puzzle I'm satisfied with my work. It'd be cool to see if anyone did seriously different layouts to me, b/c the width primary seemed to sharply limit what (min-primary) layouts can work quite dramatically. I'm sad to be submitting a non-looping design, but my looping version was 110G more expensive. I made sure that when it crashes it crashes politely though so it's alright :).