large political actions
The paper does not look at size of action, but method, such as use of violence, innocent victims, …
You earn and spend support
The paper suggests something else as more effective: small unknown groups benefit from violent action, as they’ve little to no support to lose. Later, when emotional response has died down, but having gained notoriety, they gain broader support with a more moderate messaging.
For well known causes, it just reduces long term support, but still makes it more likely for short term policy change.
EDIT: This is basically the origin story of most major religions, come to think of it