so this is my new visualisation taking the polar representation and unrolling it with a new focus on the variables that form interconnections between modules a.k.a. Atlas

Its quite hard flattening a cyclical network into a 1D layout but I think this works ok. Colors are keyed by the start and end module so you get natural same-color bundling. The X axis is proportional to the size, but its not a fixed arc because that had a tendency to explode the X axis. They are not straight lines either, because that has a tendency to alias

the lopecode version is a much bigger diagram because the network is so much more complex (its the Atlas implementation AND the development IDE whereas on Observable the IDE is not part of the userspace runtime). Atlas doesn't really scale all the way to Lopecode size but it is an improvement over cellMap which had multiple thousands of individual elements. Atlas is basically cellMap but without the internal nodes. I realized the most interesting variables are the ones that cross module boundaries; they are extra important as they sort of represent something that is useful in general settings beyond the borders. Internal module variables are often internal details that are not very interesting, so we can get a signal-to-noise boost by removing all the internal nodes. Atlas does that and also makes them more readable.