There is a class of problems where "when things get really bad" is also the point when it's too late to implement a solution. Driving off a cliff, for example.

What would you do if you studied this matter for several years, and found overwhelming evidence that this is such a category of problem?