Slightly more seriously, you could perhaps make an argument that, just like weight decay, an apparent "anti-contribution" moves the learning trajectory along, and helps the network settle into a more optimal basin eventually.
That way, my contribution is still valuable on the net, and I'm owed $0.00000003 positive dollars instead.
Was that not the joke?