The fundamental problem is that iceberg collapse usually follows an exponential model, but as with all exponential formulas, a small change in the constants yields humongous differences in the long run.