Very hard to decide. I picked "vertex weighting" because it is essential imo and we cant get to the next artistic step without them.
main reason is because all updates so far are focused on scene managament, lightning and code and i am afraid if this gets put asaid again we will have to wait another year for it.
Also a benefit would be: once it is done its done. It wont need lots of updates and fixes and shouldnt cause major errors. lightining, shadow aso sound more "fragile" to me and would also take longer to work on all systems.

as said, its very hard to pick from (kids in a candy store) and i wouldnt mind any of them.

a personal note:
After observing the last few weeks and months i have the feeling that conitec is investing a lot more power and energy into gs and especially into the updates.
it looks like they take the user complaints about long update-less times very serious and try to solve it. (reminds me of the first years i spent with gs)

I am sure i am not the only one who likes this new-old way.

cheers