I think what actually works best is local-level individual targeting of key leadership nodes.
“If you want to study one of these strange organisms, you had better have a good justification. It's not good to study gene organisation in some obscure insect that no one's ever heard about.”
“It's definitely time to stop. We're getting too old. We both realised that the show wasn't as engaging as it used to be. We were starting to look a bit ridiculous.”