The Forecasting Debate and the Brittleness of PAYGO

I've become frustrated with what I'll call the “forecasting” debate over social security. And my frustration has turned around into an additional argument against the PAYGO system.
It is now clear to me that the forecasting debate works by choosing your favored assumptions about growth, aging, immigration, etc., extrapolating into the future, and then arguing either that we are “headed for an iceberg” or that there is no iceberg, or at least there is no iceberg that can't be evaded with marginal tinkering.
However, the fact remains that no one knows what the growth rate is going to be in ten years. No one can tell you whether there will be a huge jump in life expetancy due to technological innovation in 20 years. No one can tell you whether President Jeb Bush will usher in a new era of mass immigration. Maybe those fascistic millenials will have six kids per pair!
Whatever the case may be, whether or not social security-as-we-know-it is sustainable depends on a lot of what we don't know and can't know. We can and should try to see what the future will look like if certain trends continue, given various different assumptions. Yet, we don't know how to assign probabilities to the assumptions or to the possible futures. It is likely that some unpredictable exogenous factor will render any such assignment moot.
But frustrations about the futility of the forecasting debate point to a deep flaw in the design of social security: the system is fragile. Brittle, even. The fact that the projected sustainability of social security is so sensitive to fairly small changes in growth rates, demographic change, unemployment rates and is a sure sign that it is extremely poorly designed policy.
Advocates of status quo-ish approaches are stuck arguing that the future's going to thread the needle of conditions under which the system is viable. Now, I don't know, and neither do they, whether their favored forecast will become reality. But it remains that a well-designed institution should be robust under a broad range of future conditions. Our PAYGO system just isn't. Small differences in the rate of growth, rate of increase of life expectancy, and so on, shouldn't make or break the system. Of course, there is no system that can reliably withstand dramatic changes in any variable. But we should at least aim for a system that is fairly adaptive and robust against moderate changes in growth, population, employement, and aging.