Fiesta is on record saying that this rate puzzle is largely conventional, in ways that the cost metric is absolutely not.  But what he didn't mention is that, under these restrictions, RG feels quite similar to RA.  Because when you are required to use track to move anything, every hex that your product moves is going to cost you at least 5G. Because of that, we had two main goals: first, arrange all of our inputs and glyphs to minimize the distance they all would have to travel; and second, attach single atoms to the product as soon as possible, to minimize the number of arms necessary to manipulate things. 

In our opinion, the most crucial aspect of this solve is the unusual monomer. We chose this one because it mostly eliminates the need to have one of the three components to "switch sides." That both reduced The amount of glyphs needed and reduced the distance that atoms needed to travel, thereby saving track.

We imagine that the best solves will utilize more multiarms on period 4 that are useful in multiple directions and during successive loops, as you can see on our water input and mors handling.  But as our steak in the ground clocked in at 760g, this level of optimization feels plenty respectable.