# Info-gap decision theory — the origins

December 28, 2011 Leave a comment

The evolution of info-gap decision theory into a prime example of a voodoo decision theory makes interesting reading:

- The theory had a previous incarnation, but in that version it showed no signs of the voodoo theory that it eventually became.
- There are no clues in the literature on info-gap decision theory to explain the reasons for this transformation.

The central element of the theory, namely its robustness model, originated in the book *Robust Reliability in the Mechanical Sciences* (Ben-Haim 1996), where it was treated as a model designed to handle **small perturbations** in the nominal value of a parameter.

So apart from the fact that this book was completely oblivious that this robustness model was in fact a radius of stability model, the proposition to use such a model for the analysis of **small perturbations** was utterly sound.

What is so puzzling then is that in the book * Information-gap Decision Theory* (Ben-Haim, 2001), the very same model began to be treated as a model for pursuing robustness against severe uncertainty where the severity of the uncertainty is characterized by these three working assumptions:

- The uncertainty space can be
**vast**(e.g. unbounded). - The point estimate is
**poor**(e.g. a guess). - The quantification of uncertainty is
**likelihood-free**.

There seems to be no explanation in the info-gap literature as to what triggered this transformation. Was this the result of an accidental thought, or that of a long and careful analysis of the challenges posed by a severe uncertainty of this type?