The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty

CHAPTER 45

Putting It into Practice

The technology surrounding Probability Management is improving fast, and recent breakthroughs promise to make it more accessible than ever. For perspective, I will review the evolution to date. The first-generation models, like the Bessemer and Shell demos at ProbabilityManagement.org, displayed the two revolutionary hall-marks of this approach. First, they were based on scenario libraries; so they were modular, meaning their results could be consolidated, or added together. Second, they were interactive, connecting the seat of users’ intellects to the seat of their pants.

What was frustrating, on the other hand, was that the models, as implemented in practice, required literally 20,000 to 40,000 spreadsheet formulas. This was because they were actually comprised of a thousand identical models, one for each of 1,000 potential scenarios stored in the SLURP. Then in 2007, with the introduction of Frontline Systems’ Risk Solver, we were able to greatly reduce the number of formulas in our clients’ models. Instead of requiring one submodel per row of the SLURP, we needed just a single model, through which the Risk Solver fired all 1,000 scenarios like bullets through an Uzi. When people say that size doesn’t matter, they are not talking about factors of 1,000. A spreadsheet with 40 formulas is easy to build and maintain, whereas one with 40,000 formulas, even if highly repetitive, is a nightmare. In addition, the new technology made the models even ...