1/4 + 1/20 on both outputs. Hopefully decent? My first ever RG solve actually optimised for G, and I'm pretty happy with it. Many hexarm.

My wild guess of a prediction is that winner will be ~500G and that this solution will place around 20th, but who knows tbh. I wouldn't be particularly surprised if it places anywhere from 11th-30th. 

I started with the left hand side, then attempted to build the right hand side around it. No planning done, but many redesigns required.

Arm 15 needing to exist at all is a Big Geometry Sad. I wanted to have Arm 16 two hexes down-right, but that swing doesn't work. However, there's probably much more potential savings with better layouts.

I went with the first algorithm that came to mind - There's almost certainly better out there. Dual Pipeline with Odd Ones Out might not be the way to go with how much of a nightmare routing is - I suppose I'll find out what the alternatives are.