Thanks for playing! I could group packets (and will if the rendering becomes a bottleneck), but it’s currently doing some other absurdly inefficient things that should be easier to optimize. There’s a lot of low-hanging fruit for fixing the performance that I didn’t get to during the jam itself.
(The number one thing is improving the search for routes; whenever a node is captured, it has to traverse the whole graph and recompute every route, which creates lag spikes.)