How does Elster avoid the problem that if you have a toolbox of explanations that you can select from to explain anything, you actually aren't explaining anything?

On pages 16-20 of Explaining Social Behaviour: More Nuts and Bolts for the Social Sciences, Elster argues that one should refute rival explanations and show that additional (preferably novel) corollaries of the invoked explanation are observed. "These two criteria – refuting the most plausible alternatives and generating novel facts – are decisive for the credibility of an explanation."