I don't see the point of drawers unless the order of attached cables is functional. In any case they're less like drawers and more like open shelves.
I'd suggest they be retractable from the base of the nod. Could expand and retract by double clicking on the associated node input label e.g. "Objects" label on the Render Scene to Window node.
Hopefully the lists of lists feature will get rid of the three(!) drawers on the **Copy 3D Object **node.
To my eye drawers as they currently stand do add considerable clutter to the graph. It occurred to me that Drawers could actually just be a unique node type exclusively for list construction with many input ports and one output port (think Grasshopper lists typed out on the graph as a node). So they could be positioned conveniently by the user, and with type introspection quite handy too if pulling values from far off nodes. Then i realised that one difference is that drawers are unique in that they can have inputs created and removed easily where-as Vuo nodes aren't able to do that easily ATM, calculate expression is one of the few i can think of that rebuilds the ports to match the expression input. That's why there is one node that has a 2 input and a 8 input version i guess (don't recall which on though, sorry :-) ).
replacement drawers closed:
replacement drawers Objects drawer (only) open (speculating that introspection of types is possible) I would think that dragging a cable onto the Objects port it would create a new entry in the drawer and add it: